Amm Vol121n07

New
in the
MAA eBooks Store
MATHEMATICAL ASSOCIATION OF AMERICA
Ordinary Diferential Equations is, frst and foremost, a text
for the introductory course in ordinary diferential equa-
tions, usually taken by sophomore engineering and science
majors afer a two or three term calculus sequence. Te
driving idea behind this particular text is that all science
majors need to be convinced to take the diferential equa-
tions course along with the engineers. An understanding
of dynamical systems is gradually becoming a necessity
for everyone in the sciences.
One way to encourage science majors of all kinds to take a course in
diferential equations is to show students that problems involving diferential equations are
ofen a lot more interesting than the problems seen in calculus. Tis is especially true in the
area of nonlinear equations, and is one reason why this text contains much more than the
usual amount of material on the geometry of nonlinear systems.
Recognizing that not all students are fortunate enough to take a diferential equations
course in their undergraduate career, another aim has been to make this book as readable
as possible so it can be used for self-study. Each section of the book also has many examples
followed by their worked out solution. Tose two things (readability and full solutions to
the examples) also make this text a likely candidate for a professor who wants to teach a
fipped course in diferential equations.
Each section of the book has its own set of exercises. Answers to odd-numbered exercises
are included in the back of the book.
2014, 330 pages
Electronic edition ISBN: 9781614446149
Hardcover ISBN: 9781939512048
ebook: $30.00
To order go to www.maa.org/ebooks/FCDS
Ordinary Diferential Equations
from Calculus to Dynamical Systems
Virginia W. Noonburg
a course in
ms
MAA textbooks:
great books at
unbeatable prices!
THE AMERICAN MATHEMATICAL
MONTHLY
Volume 121, No. 7 AugustSeptember 2014
EDITOR
Scott T. Chapman
Sam Houston State University
NOTES EDITOR BOOK REVIEW EDITOR
Sergei Tabachnikov Jeffrey Nunemacher
Pennsylvania State University Ohio Wesleyan University
PROBLEM SECTION EDITORS
Douglas B. West Gerald Edgar Doug Hensley
University of Illinois Ohio State University Texas A&M University
ASSOCIATE EDITORS
William Adkins
Louisiana State University
David Aldous
University of California, Berkeley
Elizabeth Allman
University of Alaska, Fairbanks
Jonathan M. Borwein
University of Newcastle
Jason Boynton
North Dakota State University
Edward B. Burger
Southwestern University
Minerva Cordero-Epperson
University of Texas, Arlington
Allan Donsig
University of Nebraska, Lincoln
Michael Dorff
Brigham Young University
Daniela Ferrero
Texas State University
Luis David Garcia-Puente
Sam Houston State University
Sidney Graham
Central Michigan University
Tara Holm
Cornell University
Roger A. Horn
University of Utah
Lea Jenkins
Clemson University
Daniel Krashen
University of Georgia
Ulrich Krause
Universit at Bremen
Jeffrey Lawson
Western Carolina University
C. Dwight Lahr
Dartmouth College
Susan Loepp
Williams College
Irina Mitrea
Temple University
Bruce P. Palka
National Science Foundation
Vadim Ponomarenko
San Diego State University
Catherine A. Roberts
College of the Holy Cross
Rachel Roberts
Washington University, St. Louis
Ivelisse M. Rubio
Universidad de Puerto Rico, Rio Piedras
Adriana Salerno
Bates College
Edward Scheinerman
Johns Hopkins University
Anne Shepler
University of North Texas
Frank Sottile
Texas A&M University
Susan G. Staples
Texas Christian University
Daniel Ullman
George Washington University
Daniel Velleman
Amherst College
EDITORIAL ASSISTANT
Bonnie K. Ponce
NOTICE TO AUTHORS
The MONTHLY publishes articles, as well as notes and
other features, about mathematics and the profes-
sion. Its readers span a broad spectrum of math-
ematical interests, and include professional mathe-
maticians as well as students of mathematics at all
collegiate levels. Authors are invited to submit arti-
cles and notes that bring interesting mathematical
ideas to a wide audience of MONTHLY readers.
The MONTHLYs readers expect a high standard of ex-
position; they expect articles to inform, stimulate,
challenge, enlighten, and even entertain. MONTHLY
articles are meant to be read, enjoyed, and dis-
cussed, rather than just archived. Articles may be
expositions of old or new results, historical or bio-
graphical essays, speculations or denitive treat-
ments, broad developments, or explorations of a
single application. Novelty and generality are far
less important than clarity of exposition and broad
appeal. Appropriate gures, diagrams, and photo-
graphs are encouraged.
Notes are short, sharply focused, and possibly infor-
mal. They are often gems that provide a new proof
of an old theorem, a novel presentation of a familiar
theme, or a lively discussion of a single issue.
Submission of articles, notes, and ller pieces is re-
quired via the MONTHLYs Editorial Manager System.
Initial submissions in pdf or L
A
T
E
X form can be sent
to the Editor Scott Chapman at
http://www.editorialmanager.com/monthly
The Editorial Manager System will cue the author
for all required information concerning the paper.
Questions concerning submission of papers can
be addressed to the Editor at monthly@shsu.edu.
Authors who use L
A
T
E
X can nd our article/note tem-
plate at http://www.shsu.edu/
~
bks006/Monthly.
html. This template requires the style le maa-
monthly.sty, which can also be downloaded from the
same webpage. Aformatting document for MONTHLY
references can be found at http://www.shsu.edu/
~
bks006/FormattingReferences.pdf. Follow the
link to Electronic Publications Information for
authors at http://www.maa.org/pubs/monthly.
html for information about gures and les, as well
as general editorial guidelines.
Letters to the Editor on any topic are invited.
Comments, criticisms, and suggestions for mak-
ing the MONTHLY more lively, entertaining, and
informative can be forwarded to the Editor at
monthly@shsu.edu.
The online MONTHLY archive at www.jstor.org is a
valuable resource for both authors and readers; it
may be searched online in a variety of ways for any
specied keyword(s). MAA members whose institu-
tions do not provide JSTOR access may obtain indi-
vidual access for a modest annual fee; call 800-331-
1622.
See the MONTHLY section of MAA Online for current
information such as contents of issues and descrip-
tive summaries of forthcoming articles:
http://www.maa.org/
Proposed problems or solutions should be sent to:
DOUG HENSLEY, MONTHLY Problems
Department of Mathematics
Texas A&M University
3368 TAMU
College Station, TX 77843-3368.
In lieu of duplicate hardcopy, authors may submit
pdfs to monthlyproblems@math.tamu.edu.
Advertising correspondence should be sent to:
MAA Advertising
1529 Eighteenth St. NW
Washington DC 20036.
Phone: (877) 622-2373,
E-mail: tmarmor@maa.org.
Further advertising information can be found online
at www.maa.org.
Change of address, missing issue inquiries, and
other subscription correspondence can be sent to:
MAA Service Center, maahq@maa.org.
All of these are at the address:
The Mathematical Association of America
1529 Eighteenth Street, N.W.
Washington, DC 20036.
Recent copies of the MONTHLY are available for pur-
chase through the MAA Service Center:
maahq@maa.org, 1-800-331-1622.
Microlm Editions are available at: University Micro-
lms International, Serial Bid coordinator, 300 North
Zeeb Road, Ann Arbor, MI 48106.
The AMERICAN MATHEMATICAL MONTHLY (ISSN
0002-9890) is published monthly except bimonthly
June-July and August-September by the Mathe-
matical Association of America at 1529 Eighteenth
Street, N.W., Washington, DC 20036 and Lancaster,
PA, and copyrighted by the Mathematical Asso-
ciation of America (Incorporated), 2014, including
rights to this journal issue as a whole and, except
where otherwise noted, rights to each individual
contribution. Permission to make copies of individ-
ual articles, in paper or electronic form, including
posting on personal and class web pages, for ed-
ucational and scientic use is granted without fee
provided that copies are not made or distributed for
prot or commercial advantage and that copies bear
the following copyright notice: [Copyright the Math-
ematical Association of America 2014. All rights re-
served.] Abstracting, with credit, is permitted. To
copy otherwise, or to republish, requires specic
permission of the MAAs Director of Publications and
possibly a fee. Periodicals postage paid at Washing-
ton, DC, and additional mailing ofces. Postmaster:
Send address changes to the American Mathemati-
cal Monthly, Membership/Subscription Department,
MAA, 1529 Eighteenth Street, N.W., Washington, DC,
20036-1385.
A Bit of Tropical Geometry
Erwan Brugall e and Kristin Shaw
Abstract. This friendly introduction to tropical geometry is meant to be accessible to rst year
students in mathematics. The topics discussed here are basic tropical algebra, tropical plane
curves, some tropical intersections, and Viros patchworking. Each denition is explained with
concrete examples and illustrations. The text is a modication of a translation from a French
text by the rst author. There is also a newly-added section highlighting new developments
and perspectives on tropical geometry. In addition, the nal section provides an extensive list
of references on the subject.
What kinds of strange spaces with mysterious properties hide behind the enigmatic
name of tropical geometry? In the tropics, just as in other geometries, it is difcult to
nd a simpler example than that of a line. So let us start from there.
A tropical line consists of three usual half lines in the directions (1, 0), (0, 1),
and (1, 1), emanating from any point in the plane (see Figure 1(a)). Why call this
strange object a line, in the tropical sense or any other? If we look more closely, we
nd that tropical lines share some of the familiar geometric properties of usual or
classical lines in the plane. For instance, most pairs of tropical lines intersect in a
single point
1
(see Figure 1(b)). Also, for most choices of pairs of points in the plane,
there is a unique tropical line passing through the two points
2
(see Figure 1(c)).
(a) (b) (c)
Figure 1. The tropical line
What is even more important, although not at all visible from the picture, is that
classical and tropical lines are both given by an equation of the form ax +by +c =
0. In the realm of standard algebra, where addition is addition and multiplication is
multiplication, we can determine without too much difculty the classical line given
http://dx.doi.org/10.4169/amer.math.monthly.121.07.563
MSC: Primary 14T05, Secondary 14P25
1
It might happen that two distinct tropical lines intersect in innitely many points, as a ray of a line might
be contained in the parallel ray of the other. We will see in Section 3 that there is a more sophisticated notion
of stable intersection of two tropical curves, and that any two tropical lines have a unique stable intersection
point.
2
The situation here is similar to the case of intersection of tropical lines; there exists the notion of a stable
tropical line passing through two points in the plane, and any two such points dene a unique stable tropical
line.
AugustSeptember 2014] A BIT OF TROPICAL GEOMETRY 563
by such an equation. In the tropical world, addition is replaced by the maximum and
multiplication is replaced by addition. Just by doing this, all of our objects drastically
change form! In fact, even being equal to 0 takes on a very different meaning.
Classical and tropical geometries are developed following the same principles, but
from two different methods of calculation. They are simply the geometric faces of two
different algebras.
Tropical geometry is not a game for bored mathematicians looking for something
to do. In fact, the classical world can be degenerated to the tropical world in such a
way so that the tropical objects conserve many properties of the original classical ones.
Because of this, a tropical statement has a strong chance of having a similar classical
analogue. The advantage is that tropical objects are piecewise linear, and thus much
simpler to study than their classical counterparts!
We could therefore summarize the approach of tropical geometry as follows:
Study simple objects to provide theorems concerning complicated objects.
The rst part of this text focuses on tropical algebra and tropical curves, and some
of their properties. We then explain why classical and tropical geometries are related
by showing how the classical world can be degenerated to the tropical one. We illus-
trate this principle by considering a method known as patchworking, which is used
to construct real algebraic curves via objects called amoebas. Finally, in the last sec-
tion we go beyond these topics to showcase some further developments in the eld.
These examples are a little more challenging than the rest of the text, as they illustrate
more recent directions of research. We conclude by delivering some bibliographical
references.
Before diving into the subject, we should explain the use of the word tropical. It is
not due to the exotic forms of the objects under consideration, nor to the appearance of
the previously-mentioned amoebas. Before the termtropical algebra, the more cut-
and-dried name of max-plus algebra was used. Then, in honor of the work of their
Brazilian colleague, Imre Simon, the computer science researchers at the University of
Paris decided to trade the name of max-plus for tropical. Leaving the last word to
Wikipedia
3
, the origin of the word tropical simply reects the French view on Brazil.
1. TROPICAL ALGEBRA.
Tropical operations. Tropical algebra is the set of real numbers where addition is
replaced by taking the maximum, and multiplication is replaced by the usual sum. In
other words, we dene two new operations on R, called tropical addition and multi-
plication, and denoted + and , respectively, in the following way:
x + y = max(x, y), x y = x + y.
In this entire text, quotation marks will be placed around an expression to indicate
that the operations should be regarded as tropical. Just as in classical algebra, we of-
ten abbreviate x y to xy. To familiarize ourselves with these two new strange
operations, lets do some simple calculations:
1 +1 = 1, 1 +2 = 2, 1 +2 +3 = 3, 1 2 = 3,
1 (2 +(1)) = 3, 1 (2) = 1, and (5 +3)
2
= 10.
3
March 15, 2009.
564 c THE MATHEMATICAL ASSOCIATION OF AMERICA [Monthly 121
These two tropical operations have many properties in common with the usual ad-
dition and multiplication. For example, both are commutative, and tropical multipli-
cation is distributive with respect to tropical addition + (i.e., (x + y)z =
xz + yz). There are, however, two major differences. First of all, tropical addi-
tion does not have an identity element in R (i.e., there is no element x such that
max{y, x} = y for all y R). Nevertheless, we can naturally extend our two tropi-
cal operations to by
x T, x +() = max(x, ) = x, and x () = x +() = ,
where T = R {} are the tropical numbers. Therefore, after adding to R,
tropical addition now has an identity element. On the other hand, a major difference
remains between tropical and classical addition: An element of R does not have an
additive inverse. Said in another way, tropical subtraction does not exist. Neither
can we solve this problem by adding more elements to T to try to cook up additive
inverses. In fact, + is said to be idempotent, meaning that x + x = x for all x in
T. Our only choice is to get used to the lack of tropical additive inverses!
Despite this last point, the tropical numbers T, equipped with the operations +
and , satisfy all of the other properties of a eld. For example, 0 is the identity
element for tropical multiplication, and every element x of T different from has
a multiplicative inverse
1
x
= x. Then T satises almost all of the axioms of a eld,
so by convention we say that it is a semi-eld.
Take care when writing tropical formulas, as 2x = x + x but 2x = x + 2.
Similarly, 1x = x but 1x = x + 1, and once again 0x = x and (1)x =
x 1.
Tropical polynomials. After having dened tropical addition and multiplication, we
naturally come to consider functions of the form P(x) =
d
i =0
a
i
x
i
with the a
i
s in
T, in other words, tropical polynomials
4
. By rewriting P(x) in classical notation, we
obtain P(x) = max
d
i =1
(a
i
+i x). Lets look at some examples of tropical polynomials:
x = x, 1 + x = max(1, x), 1 + x +3x
2
= max(1, x, 2x +3),
1 + x +3x
2
+(2)x
3
= max(1, x, 2x +3, and 3x 2).
Now, lets nd the roots of a tropical polynomial. Of course, we must rst ask, what
is a tropical root? In doing this, we encounter a recurring problem in tropical mathe-
matics. A classical notion may have many equivalent denitions, yet when we pass to
the tropical world these could turn out to be different, as we will see. Each equivalent
denition of the same classical object potentially produces as many different tropical
objects.
The most basic denition of classical roots of a polynomial P(x) is an element x
0
such that P(x
0
) = 0. If we attempt to replicate this denition in tropical algebra, we
must look for elements x
0
in T such that P(x
0
) = . Yet, if a
0
is the constant term
of the polynomial P(x), then P(x) a
0
for all x in T. Therefore, if a
0
= , the
polynomial P(x) would not have any roots. This denition is surely not adequate.
We may take an alternative, yet equivalent, classical denition. An element x
0
T
is a classical root of a polynomial P(x) if there exists a polynomial Q(x) such that
P(x) = (x x
0
)Q(x). We will soon see that this denition is the correct one for
4
In fact, we consider tropical polynomial functions instead of tropical polynomials. Note that two different
tropical polynomials may still dene the same function.
tropical algebra. To understand it, lets take a geometric point of view of the problem.
A tropical polynomial is a piecewise linear function and each piece has an integer
slope (see Figure 2). What is also apparent from Figure 2 is that a tropical polynomial
is convex, or concave up. This is because it is the maximum of a collection of linear
functions.
We call tropical roots of the polynomial P(x) all points x
0
of T for which the graph
of P(x) has a corner at x
0
. Moreover, the difference in the slopes of the two pieces
adjacent to a corner gives the order of the corresponding root. Thus, the polynomial
0 + x has a simple root at x
0
= 0, the polynomial 0 + x + (1)x
2
has simple
roots 0 and 1, and the polynomial 0 + x
2
has a double root at 0.
0
0
(, ) 0
0
1 (, ) 0
0
(, )
(a) P(x) = 0 + x (b) P(x) = 0 + x +(1)x
2
(c) P(x) = 0 + x
2
Figure 2. The graphs of some tropical polynomials

The roots of a tropical polynomial P(x) =
d
i =0
a
i
x
i
= max
d
i =1
(a
i
+ i x) are
therefore exactly the tropical numbers x
0
for which there exists a pair i = j such
that P(x
0
) = a
i
+i x
0
= a
j
+ j x
0
. We say that the maximum of P(x) is obtained (at
least) twice at x
0
. In this case, the order of the root at x
0
is the maximum of |i j | for
all possible pairs i , j that realize this maximum at x
0
. For example, the maximum of
P(x) = 0 + x + x
2
is obtained three times at x
0
= 0 and the order of this root is 2.
Equivalently, x
0
is a tropical root of order at least k of P(x) if there exists a tropical
polynomial Q(x) such that P(x) = (x + x
0
)
k
Q(x). Note that the factor x x
0
in
classical algebra gets transformed to the factor x + x
0
, since the root of the polyno-
mial x + x
0
is x
0
and not x
0
.
This denition of a tropical root seems to be much more satisfactory than the rst
one. In fact, using this denition, we have the following proposition.
Proposition 1.1. The tropical semi-eld is algebraically closed. In other words, every
tropical polynomial of degree d has exactly d roots when counted with multiplicities.
For example, we may check that we have the following factorizations:
5
0 + x +(1)x
2
= (1)(x +0)(x +1) and 0 + x
2
= (x +0)
2
.
Exercises.
1. Why does the idempotent property of tropical addition prevent the existence of
inverses for this operation?
2. Draw the graphs of the tropical polynomials P(x) = x
3
+2x
2
+3x +(1)
and Q(x) = x
3
+(2)x
2
+2x +(1), and determine their tropical roots.
5
Once again, the equalities hold in terms of polynomial functions not on the level of the polynomials. For
example, 0 + x
2
and (0 + x)
2
are equal as polynomial functions but not as polynomials.
3. Let a R and b, c T. Determine the roots of the polynomials ax +b and
ax
2
+bx +c.
4. Prove that x
0
is a tropical root of order at least k of P(x) if and only if there
exists a tropical polynomial Q(x) such that P(x) = (x + x
0
)
k
Q(x).
5. Prove Proposition 1.1.
2. TROPICAL CURVES.
Denition. Carrying on boldly, we can increase the number of variables in our poly-
nomials. A tropical polynomial in two variables is written P(x, y) =
i, j
a
i, j
x
i
y
j
,
or better yet P(x, y) = max
i, j
(a
i, j
+i x + j y) in classical notation. In this way, our
tropical polynomial is again a convex piecewise linear function, and the tropical curve
C dened by P(x, y) is the corner locus of this function. Said in another way, a tropi-
cal curve C consists of all points (x
0
, y
0
) in T
2
for which the maximum of P(x, y) is
obtained at least twice at (x
0
, y
0
).
We should point out that, until Section 6, we will focus on tropical curves contained
in R
2
and not in T
2
. This does not affect at all the generality of what will be discussed
here; however, it renders the denitions, the statements, and our drawings simpler and
easier to understand.
Let us look at the tropical line dened by the polynomial P(x, y) =
1
2
+ 2x +
(5)y. We must nd the points (x
0
, y
0
) in R
2
that satisfy one of the following three
systems of equations:
2 + x
0
=
1
2
5 + y
0
, 5 + y
0
=
1
2
2 + x
0
, 2 + x
0
= 5 + y
0

1
2
.
We see that our tropical line is made up of three standard half-lines:
__
3
2
, y
_
y
11
2
_
,
__
x,
11
2
_
x
3
2
_
, and
_
(x, x +7)
x
3
2
_
(see Figure 3(a)).
2
2
(a)
1
2
+2x +(5)y (b) 3 +2x +2y +3xy + y
2
+ x
2
(c) 0 + x + y
2
+(1)x
2
Figure 3. Some tropical curves

We are still missing one bit of information to properly dene a tropical curve. The
corner locus of a tropical polynomial in two variables consists of line segments and
half-lines, which we call edges. These intersect at points, which we will call vertices.
Just as in the case of polynomials in one variable, for each edge of a tropical curve, we
must take into account the difference in the slope of P(x, y) on the two sides of the
edge. Doing this, we arrive at the following formal denition of a tropical curve.
Denition 2.1. Let P(x, y) =
i, j
a
i, j
x
i
y
j
be a tropical polynomial. The tropical
curve C dened by P(x, y) is the set of points (x
0
, y
0
) of R
2
, such that there exists
pairs (i, j ) = (k, l) satisfying P(x
0
, y
0
) = a
i, j
+i x
0
+ j y
0
= a
k,l
+kx
0
+l y
0
.
We dene the weight w
e
of an edge e of C to be the maximum of the greatest
common divisor (gcd) of the numbers |i k| and | j l| for all pairs (i, j ) and (k, l)
that correspond to this edge. That is to say,
w
e
= max
M
e
(gcd(|i k|, | j l|)) ,
where
M
e
=
_
(i, j ), (k, l) | x
0
e, P(x
0
, y
0
) = a
i, j
+i x
0
+ j y
0
= a
k,l
+kx
0
+l y
0
_
.
In Figure 3, the weight of an edge is only indicated if the weight is at least two.
For example, in the case of the tropical line, all edges are of weight 1. Thus, Figure
3(a) represents the tropical line fully. Two examples of tropical curves of degree 2
are shown in Figures 3(b) and (c). The tropical conic in Figure 3(c) has two edges of
weight 2. Note that, given an edge e of a tropical curve dened by P(x, y), the set M
e
can have any cardinality between 2 and w
e
+1: In the example of Figure 3(c), we have
M
e
= {(0, 0), (0, 1), (0, 2)} if e is the horizontal edge, and M
e
= {(2, 0), (0, 2)} if e is
the other edge of weight 2.
Dual subdivisions. To recap, a tropical polynomial is given by the maximum of a -
nite number of linear functions corresponding to monomials of P(x, y). Moreover, the
points of the plane R
2
for which at least two of these monomials realize the maximum
are exactly the points of the tropical curve C dened by P(x, y). Let us rene this a bit
and consider at each point (x
0
, y
0
) of C, all of the monomials of P(x, y) that realize
the maximum at (x
0
, y
0
).
Let us rst go back to the tropical line C dened by the equation P(x, y) =
1
2
+
2x +(5)y (see Figure 3(a)). The point (
3
2
,
11
2
) is the vertex of the line C. This is
where the three monomials
1
2
=
1
2
x
0
y
0
, 2x = 2x
1
y
0
, and (5)y = (5)x
0
y
1
take the
same value. The exponents of those monomials, that is to say, the points (0, 0), (1, 0),
and (0, 1), dene a triangle
1
(see Figure 4(a)). Along the horizontal edge of C, the
value of the polynomial P(x, y) is given by the monomials 0 and y, in other words,
the monomials with exponents (0, 0) and (0, 1). Therefore, these two exponents dene
the vertical edge of the triangle
1
. In the same way, the monomials giving the value
of P(x, y) along the vertical edge of C have exponents (0, 0) and (1, 0), which dene
the horizontal edge of
1
. Finally, along the edge of C that has slope 1, P(x, y) is
given by the monomials with exponents (1, 0) and (0, 1), which dene the edge of
1
that has slope 1.
What can we learn from this digression? In looking at the monomials that give the
value of the tropical polynomial P(x, y) at a point of the tropical line C, we notice that
(a) (b) (c)
Figure 4. Some dual subdivisions
the vertex of C corresponds to the triangle
1
and that each edge e of C corresponds
to an edge
e
of
1
, whose direction is perpendicular to that of e.
Let us illustrate this with the tropical conic dened by the polynomial P(x, y) =
3 + 2x + 2y + 3xy + x
2
+ y
2
, and drawn in Figure 3(b). This curve has as its
vertices four points (1, 1), (1, 2), (1, 1), and (2, 1). At each of these vertices
(x
0
, y
0
), the value of the polynomial P(x, y) is given by three monomials:
P(1, 1) = 3 = y
0
+2 = x
0
+ y
0
+3,
P(1, 1) = 3 = x
0
+2 = x
0
+ y
0
+3,
P(1, 2) = y
0
+2 = x
0
+ y
0
+3 = 2y
0
,
and
P(2, 1) = x
0
+2 = x
0
+ y
0
+3 = 2x
0
.
Thus, for each vertex of C, the exponents of the three corresponding monomials dene
a triangle, and these four triangles are arranged as shown in Figure 4(b). Moreover, just
as in the case of the line, for each edge e of C, the exponents of the monomials giving
the value of P(x, y) along the edge e dene an edge of one (or two) of these triangles.
Once again, the direction of the edge e is perpendicular to the corresponding edge of
the triangle.
To explain this phenomenon in full generality, let P(x, y) =
i, j
a
i, j
x
i
y
j
be any
tropical polynomial. The degree of P(x, y) is the maximum of the sums i + j for all
coefcients a
i, j
different from . For simplicity, we will assume in this text that all
polynomials of degree d satisfy a
0,0
= , a
d,0
= , and a
0,d
= . Thus, all
the points (i, j ) such that a
i, j
= are contained in the triangle with vertices (0, 0),
(0, d), and (d, 0), which we call
d
. Given a nite set of points A in R
2
, the convex
hull of A is the unique convex polygon with vertices in A and containing
6
A. From
what we just said, the triangle
d
is precisely the convex hull of the points (i, j ) such
that a
i, j
= .
If v = (x
0
, y
0
) is a vertex of the curve C dened by P(x, y), then the convex hull of
the points (i, j ) in
d
Z
2
such that P(x
0
, y
0
) = a
i, j
+i x
0
+ j y
0
is another polygon
v
, which is contained in
d
. Similarly, if (x
0
, y
0
) is a point in the interior of an edge
e of C, then the convex hull of the points (i, j ) in
d
Z
2
such that P(x
0
, y
0
) =
a
i, j
+i x
0
+ j y
0
is a segment
e
contained in
d
. The fact that the tropical polynomial
P(x, y) is a convex piecewise linear function implies that the collection of all
v
form
a subdivision of
d
. In other words, the union of all of the polygons
v
is equal to the
triangle
d
, and two polygons
v
and
v
have either an edge in common, a vertex in
common, or do not intersect at all. Moreover, if e is an edge of C adjacent to the vertex
v, then
e
is an edge of the polygon
v
, and
e
is perpendicular to e. In particular, an
edge e of C is innite, i.e., is adjacent to only one vertex of C, if and only if
e
is
contained in an edge of
d
. This subdivision of
d
is called the dual subdivision of C.
For example, the dual subdivisions of the tropical curves in Figure 3 are drawn in
Figure 4 (the black points represent the points of R
2
with integer coordinates; notice
that they are not necessarily the vertices of the dual subdivision).
The weight of an edge may be read off directly from the dual subdivision.
Proposition 2.2. An edge e of a tropical curve has weight w if and only if Card(
e

Z
2
) = w +1.
6
Equivalently, it is the smallest convex polygon containing A.
It follows from Proposition 2.2 that the degree of a tropical curve may be deter-
mined easily only from the curve itself. It is the sum of weights of all innite edges
in the direction (1, 0) (we could equally consider the directions (0, 1) or (1, 1)).
Moreover, up to a translation and choice of lengths of its edges, a tropical curve is
determined by its dual subdivision.
Balanced graphs and tropical curves. The rst consequence of the duality from the
last section is that a certain relation, known as the balancing condition, is satised at
each vertex of a tropical curve. Suppose that v is a vertex of C adjacent to the edges
e
1
, . . . , e
k
with respective weights w
1
, . . . , w
k
. Recall that every edge e
i
is contained
in a line (in the usual sense) dened by an equation with integer coefcients. Because
of this, there exists a unique integer vector v
i
= (, ) in the direction of e
i
such that
gcd(, ) = 1 (see Figure 5(a)). We orient the boundary of
v
in the counterclockwise
direction, so that each edge
e
i
of
v
dual to e
i
is obtained from a vector w
i
v
i
by
rotating by an angle of exactly /2 (see Figure 5(b)). Then, following the previous
section, the polygon
v
dual to v yields immediately the vectors w
1
v
1
, . . . , w
k
v
k
.
3
v

v
(a) (b)
Figure 5. Balancing condition
The fact that the polygon
v
is closed immediately implies the following balancing
condition:
k
i =1
w
i
v
i
= 0.
A graph in R
2
, whose edges have rational slopes and are equipped with positive
integer weights, is a balanced graph if it satises the balancing condition at each one
of its vertices. We have just seen that every tropical curve is a balanced graph. In fact,
the converse is also true.
Theorem 2.3 (G. Mikhalkin). Tropical curves in R
2
are exactly the balanced graphs.
Thus, this theorem afrms that there exist tropical polynomials of degree 3 whose
tropical curves are the weighted graphs in Figure 6. We have also drawn the associated
dual subdivision of
3
for each curve.
Exercises.
1. Draw the tropical curves dened by the tropical polynomials P(x, y) = 5 +
5x + 5y + 4xy + 1y
2
+ x
2
and Q(x, y) = 7 + 4x + y + 4xy + 3y
2
+
(3)x
2
, as well as their dual subdivisions.
2
(a) (b) (c)
Figure 6.
2. A tropical triangle is a domain of R
2
bounded by three tropical lines. What are
the possible forms of a tropical triangle?
3. Prove Proposition 2.2.
4. Show that a tropical curve of degree d has at most d
2
vertices.
5. Find an equation for each of the tropical curves in Figure 6. The following re-
minder might be helpful: If v is a vertex of a tropical curve dened by a tropical
polynomial P(x, y), then the value of P(x, y) in a neighborhood of v is given
uniquely by the monomials corresponding to the polygon dual to v.
3. TROPICAL INTERSECTION THEORY.
B ezouts theorem. One of the main interests in tropical geometry is to provide a
simple model of algebraic geometry. For example, the basic theorems from intersec-
tion theory of tropical curves require much less mathematical background than their
classical counterparts. The theorem we have in mind is B ezouts theorem, which states
that two algebraic curves in the plane of degrees d
1
and d
2
, respectively, intersect in
d
1
d
2
points.
7
Before we tackle the general case, let us rst consider tropical lines and
conics.
As mentioned in the introduction, two tropical lines generally intersect in exactly
one point (see Figure 7(a)), just as in classical geometry. Now, do a tropical line and
a tropical conic intersect each other in two points? If we naively count the number
of intersection points, the answer is sometimes yes (Figure 7(b)) and sometimes no
(Figure 7(c)).
(a) (b) (c)
Figure 7. Intersections of tropical lines and conics
7
Warning, this is a theorem in projective geometry! For example, we may have two lines in the classical
plane that are parallel: If they do not intersect in the plane, they intersect nevertheless at innity. Also, we have
to count intersection points with multiplicity. For example, a tangency point between two curves will count as
two intersection points.
In fact, the unique intersection point of the conic and the tropical line in Figure 7(c)
should be counted twice. But why twice in this case and not in the previous case? To
nd the answer, we look to the dual subdivisions.
We start by restricting to the case when the curves C
1
, C
2
intersect in a nite col-
lection of points away from the vertices of both curves. Note that the union of the two
tropical curves C
1
and C
2
is again a tropical curve. In fact, we can easily verify that the
union of two balanced graphs is again a balanced graph. Or, we could also easily check
that if C
1
and C
2
are dened by tropical polynomials P
1
(x, y), P
2
(x, y), respectively,
then Q(x, y) = P
1
(x, y)P
2
(x, y) denes precisely the curve C
1
C
2
. Moreover, the
degree of C
1
C
2
is the sum of the degrees of C
1
and C
2
.
The dual subdivisions of the unions of the two curves C
1
and C
2
in the three cases
of Figure 7 are represented in Figure 8. In every case, the set of vertices of C
1
C
2
is
the union of the vertices of C
1
, the vertices of C
2
, and the intersection points of C
1
and
C
2
. Moreover, since each point of intersection of C
1
and C
2
is contained in an edge of
both C
1
and C
2
, the polygon dual to such a vertex of C
1
C
2
is a parallelogram. To
make Figure 8 more transparent, we have drawn each edge of the dual subdivision in
the same color as its corresponding dual edge. We can conclude that in Figures 8(a)
and (b), the corresponding parallelograms are of area one, whereas the corresponding
parallelogram of the dual subdivision in Figure 8(c) has area two! Hence, it seems that
we are counting each intersection point with the multiplicity we will describe below.
(a) (b) (c)
Figure 8. The subdivisions dual to the union of the curves in Figure 7
Denition 3.1. Let C
1
and C
2
be two tropical curves that intersect in a nite number
of points and away from the vertices of the two curves. If p is a point of intersection
of C
1
and C
2
, the tropical multiplicity of p as an intersection point of C
1
and C
2
is the
area of the parallelogram dual to p in the dual subdivision of C
1
C
2
.
With this denition, proving the tropical B ezouts theorem is a walk in the park!
Theorem 3.2 (B. Sturmfels). Let C
1
and C
2
be two tropical curves of degrees d
1
and
d
2
respectively, intersecting in a nite number of points away from the vertices of the
two curves. Then the sum of the tropical multiplicities of all points in the intersection
of C
1
and C
2
is equal to d
1
d
2
.
Proof. Let us call this sum of multiplicities s. Notice that there are three types of
polygons in the subdivision dual to the tropical curve C
1
C
2
:
those dual to a vertex of C

1
. The sum of their areas is equal to the area of
d
1
, in
other words,
d
2
1
2
;
those dual to a vertex of C

2
. The sum of their areas is equal to
d
2
2
2
;
those dual to an intersection point of C

1
and C
2
. The sum of their areas we have
called s.
Since the curve C
1
C
2
is of degree d
1
+d
2
, the sum of the area of all of these poly-
gons is equal to the area of
d
1
+d
2
, which is
(d
1
+d
2
)
2
2
. Therefore, we obtain
s =
(d
1
+d
2
)
2
d
2
1
d
2
2
2
= d
1
d
2
,
which completes the proof.
Stable intersection. In the last section, we considered only tropical curves that inter-
sect nicely, meaning they intersect only in a nite number of points and away from
the vertices of the two curves. But what can we say in the two cases shown in Fig-
ures 9(a) (two tropical lines that intersect in an edge) and 9(b) (a tropical line passing
through a vertex of a conic)? Thankfully, we have more than one tropical trick up our
sleeve.
v
v
(a) (b) (c) (d)
Figure 9. Non-transverse intersection and a translation
Let be a very small positive real number and v a vector such that the quotient of
its two coordinates is an irrational number. If we translate in each of the two cases
one of the two curves in Figures 9(a) and (b) by the vector v, we nd ourselves back
in the case of nice intersection (see Figures 9(c) and (d)). Of course, the resulting
intersection depends on the vector v. On the other hand, the limit of these points, if
we let shrink to 0, does not depend on v. The points in the limit are called the stable
intersection points of two curves. The multiplicity of a point p in the stable intersection
is equal to the sum of the intersection multiplicities of all points that converge to p
when tends to zero.
For example, there is only one stable intersection point of the two lines in Figure
9(a). The point is the vertex of the line on the left. Moreover, this point has multiplic-
ity 1, so again, our two tropical lines intersect in a single point. The point of stable
intersection of the two curves in Figure 9(b) is the vertex of the conic, and it has mul-
tiplicity 2.
Notice that if a point is in the stable intersection of two tropical curves, then it is
either an isolated intersection point, or a vertex of one of the two curves. Thanks to
stable intersection, we can remove from our previous statement of the tropical B ezout
theorem the hypothesis that the curves must intersect nicely.
Theorem 3.3 (B. Sturmfels). Let C
1
and C
2
be two tropical curves of degree d
1
and
d
2
. Then the sum of the multiplicities of the stable intersection points of C
1
and C
2
is
equal to d
1
d
2
.
In passing, we may notice a surprising tropical phenomenon: A tropical curve has
a well dened self-intersection!
8
Indeed, all we have to do is to consider the stable
intersection of a tropical curve with itself. Following the discussion above, the self-
intersection points of a curve are the vertices of the curve (see Figure 10).
Figure 10. The four points in the self-intersection of a conic
Exercises.
1. Determine the stable intersection points of the two tropical curves in Exercise 1
of Section 2, as well as their multiplicity.
2. A double point of a tropical curve is a point where two edges intersect (i.e., the
dual polygon to the vertex of the curve is a parallelogram). Show that a tropical
conic with a double point is the union of two tropical lines. Hint: Consider a line
passing through the double point of the conic and any other vertex of the conic.
3. Show that a tropical curve of degree 3 with two double points is the union of a
line and a tropical conic. Show that a tropical curve of degree 3 with three double
points is the union of three tropical lines.
4. A FEW EXPLANATIONS. Let us pause for a while from our introduction of
tropical geometry to explain briey some connections between classical and tropical
geometry. In particular, our goal is to illustrate the fact that tropical geometry is a limit
of classical geometry. If we were to realize roughly the content of this section in one
sentence, it would be: Tropical geometry is the image of classical geometry under the
logarithm with base +.
Maslov dequantization. First of all, let us explain how the tropical semi-eld arises
naturally as the limit of some classical semi-elds. This procedure, studied by Victor
Maslov and his collaborators beginning in the 1990s, is known as dequantization of
the real numbers.
A well-known semi-eld is the set of positive or zero real numbers together with
the usual addition and multiplication, denoted (R
+
, +, ). If t is a strictly positive
real number, then the logarithm of base t provides a bijection between the sets R and
T. This bijection induces a semi-eld structure on T with the operations denoted by
+
t
and
t
, and given by:
x +
t
y = log
t
(t
x
+t
y
) and x
t
y = log
t
(t
x
t
y
) = x + y.
8
In classical algebraic geometry, only the total number of points in the self-intersection of a planar curve
is well dened, not the positions of the points on the curve. We can still say that a line intersects itself in one
point, but it is not at all . . . clear which point.
The equation on the right-hand side already shows classical addition appearing as
an exotic kind of multiplication on T. Notice that by construction, all of the semi-elds
(T, +
t
,
t
) are isomorphic to (R
+
, +, ). The trivial inequality max(x, y)
x + y 2 max(x, y) on R
+
, together with the fact that the logarithm is an increasing
function, gives us the following bounds for +
t
:
t > 1, max(x, y) x +
t
y max(x, y) +log
t
2.
If we let t tend to innity, then log
t
2 tends to 0, and the operation +
t
therefore
tends to the tropical addition +! Hence, the tropical semi-eld comes naturally from
degenerating the classical semi-eld (R
+
, +, ). From an alternative perspective, we
can view the classical semi-eld (R
+
, +, ) as a deformation of the tropical semi-
eld. This explains the use of the term dequantization, coming from physics and
referring to the procedure of passing from quantum to classical mechanics.
Dequantization of a line in the plane. Now we will apply a similar reasoning to the
line in the plane R
2
dened by the equation x y +1 (see Figure 11(a)). To apply the
logarithm map to the coordinates of R
2
, we must rst take their absolute values. Doing
this results in folding the four quadrants of R
2
onto the positive quadrant (see Figure
11(b)). The image of the folded-up line under the coordinate-wise logarithm with base
t applied to (R
+
)
2
is drawn in Figure 11(c). By denition, taking the logarithm with
base t is the same thing as taking the natural logarithm and then rescaling the result by
a factor of
1
ln t
. Thus, as t increases, the image under the logarithm with base t of the
absolute value of our line becomes concentrated around a neighborhood of the origin
and three asymptotic directions, as shown in Figures 11(c), (d), and (e). If we allow t
to go all the way to innity, then we see the appearance in Figure 11(f) of a tropical
line!
(a) (b) (c)
(d) (e) (f)
Figure 11. Dequantization of a line
5. PATCHWORKING. Reading Figure 11 from left to right, we see how starting
from a classical line in the plane we can arrive at a tropical line. Reading this gure
from right to left is, in fact, much more interesting. Indeed, we see as well how to
construct a classical line given a tropical one. The technique known as patchworking
is a generalization of this observation. In particular, it provides a purely combinatorial
procedure to construct real algebraic curves from a tropical curve. To explain this
procedure in more detail, we rst go a bit back in time.
Hilberts 16th problem. A planar real algebraic curve is a curve in the plane R
2
dened by an equation of the form P(x, y) = 0, where P(x, y) is a polynomial whose
coefcients are real numbers. The real algebraic curves of degree 1 and 2 are simple
and well known; they are lines and conics, respectively. When the degree of P(x, y)
increases, the form of the real algebraic curve can become more and more complicated.
If you are not convinced, take a look at Figure 12, which shows some of the possible
drawings realized by real algebraic curves of degree 4.
(a) (b) (c) (d)
Figure 12. Some real algebraic curves of degree 4
A theorem due to Axel Harnack at the end of the 19th century states that a planar
real algebraic curve of degree d has a maximum of
d(d1)+2
2
connected components.
But how can these components be arranged with respect to each other? We call the
relative position of the connected components of a planar real algebraic curve in the
plane its arrangement. In other words, we are not interested in the exact position of
the curve in the plane, but simply in the conguration that it realizes. For example, if
one curve has two bounded connected components, we are only interested in whether
one of these components is contained in the other (Figure 12(c)) or not (Figure 12(a)).
At the second International Congress in Mathematics in Paris in 1900, David Hilbert
announced his famous list of 23 problems for the 20th century. The rst part of his
16th problem can be very widely understood as the following:
Given a positive integer d, establish a list of possible arrangements of real alge-
braic curves of degree d.
At Hilberts time, the answer
9
was known for curves of degree at most 4. There
have been spectacular advances in this problem in the 20th century, due mostly in part
to mathematicians from the Russian school. Despite this, there remain numerous open
questions.
Real and tropical curves. In general, it is a difcult problem to construct a real alge-
braic curve of a xed degree and realizing a given arrangement. For a century, math-
ematicians have proposed many ingenious methods for doing this. The patchworking
method invented by Oleg Viro in the quantization is actually one of the most powerful.
9
A more reasonable and natural problem is to consider arrangements of connected components of non-
singular real algebraic curves in the projective plane, rather than in R
2
. In this more restrictive case, the answer
was known up to degree 5 in Hilberts time, and is now known up to degree 7. The list in degree 7 is given in
a theorem by Oleg Viro and patchworking is an essential tool in its proof.
At this time, tropical geometry was not yet in existence, and Viro announced his the-
orem in a language different from what we use here. However, he realized by the end
of the 1990s that his patchworking could be interpreted as a quantization of tropical
curves. Patchworking is, in fact, the process of reading Figure 11 from right to left
instead of left to right. Thanks to the interpretation of patchworking in terms of tropi-
cal curves, shortly afterwards, Grigory Mikhalkin generalized Viros original method.
Here we will present a simplied version of patchworking. The interested reader may
nd a more complete version in the references indicated in Section 7.
In what follows, for a, b two integer numbers, we denote by s
a,b
: R
2
R
2
the
composition of a reections in the x-axis with b reections in the y-axis. Therefore,
the map s
a,b
depends only on the parity of a and b. To be precise, s
0,0
is the identity,
s
1,0
the reection in the x-axis, s
0,1
the reection in the y-axis, and s
1,1
the reection
in the origin (equivalently, rotation by 180 degrees).
We will now explain in detail the procedure of patchworking. We start with a tropi-
cal curve C of degree d, which has only edges of odd weight, and such that the poly-
gons dual to all vertices are triangles. For example, take the tropical line from Figure
13(a). For each edge e of C, choose a vector v
e
= (
e
,
e
) in the direction of e such
that
e
and
e
are relatively prime integers (note that we may choose either v
e
or v
e
,
but this does not matter in what follows). For the tropical line, we may take the vectors
(1, 0), (0, 1), and (1, 1). Now, let us think of the plane R
2
where our tropical curve
sits as being the positive quadrant (R
+
)
2
of R
2
, and take the union of the curve with
its three symmetric copies obtained by reections in the axes. For the tropical line, we
obtain Figure 13(b). For each edge e of our curve, we will erase two out of the four
symmetric copies of e, denoted e
and e
. Our choice must satisfy the following rules:
= s
e
,
e
(e
), and
for each vertex v of C adjacent to the edges e

1
, e
2
, and e
3
, and for each pair (
1
,
2
)
in {0, 1}
2
, exactly one or three of the copies of s
1
,
2
(e
1
), s
1
,
2
(e
2
), and s
1
,
2
(e
3
) are
erased.
We call the result after erasing the edges a real tropical curve. For example, if C is
a tropical line, using the rules above, it is possible to erase six of the edges from the
symmetric copies of C to obtain the real tropical line represented in Figure 13(c). It is
true that this real tropical curve is not a regular line in R
2
, but they are arranged in the
plane in same way (see Figure 13(d)).
This is not just a coincidence; it is in fact a theorem.
(a) (b) (c) (d)
Figure 13. Patchworking of a line
Theorem 5.1 (O. Viro). Given any real tropical curve of degree d, there exists a real
algebraic curve of degree d with the same arrangement.
We should take a second to realize the depth and elegance of the above statement.
A real tropical curve is constructed by following the rules of a purely combinatorial
game. It seems like magic to assert that there is a relationship between these combi-
natorial objects and actual real algebraic curves! We will not get into details here, but
Viros method even allows us to determine the equation of a real algebraic curve with
the same arrangement than a given real tropical curve.
Of course, much skill and intuition are still required to construct real algebraic
curves with complicated arrangements, and both of these are acquired with practice.
Therefore, we invite the reader to try the exercises at the end of this section. Neverthe-
less, patchworking provides a much more exible technique to construct arrangements
than dealing directly with polynomials. In particular, it requires no a priori knowledge
of algebraic geometry. To illustrate the use of Viros patchworking, let us now con-
struct the arrangements of two real algebraic curves, one of degree 3 and the other of
degree 6.
First of all, consider the tropical curve of degree 3 shown in Figure 14(a). For suit-
able choices of edges to erase, Figures 14(b) and (c) represent the two stages of the
patchworking procedure. With this, we have proved the existence of a real algebraic
curve of degree 3, resembling the conguration of Figure 14(d).
(a) (b) (c) (d)
Figure 14. Patchworking of a cubic
To conclude, consider the tropical curve of degree 6 drawn in Figure 15(a). Once
again, for a suitable choice of edges to erase, the patchworking procedure gives the
curve in Figure 15(c). A real algebraic curve of degree 6 that realizes the same arrange-
ment as the real tropical curve was rst constructed by using much more complicated
techniques in the 1960s by Gudkov. An interesting piece of trivia: Hilbert announced
in 1900 that such a curve could not exist!
(a) (b) (c)
Figure 15. Gudkovs curve
Amoebas. Even though dequantization of a line is the main idea underlying patch-
working in its full generality, the proof of Viros theorem is quite a bit more technical
to explain rigorously. Here, we will give a sketch of what is involved in the proof.
First of all, since the eld R is not algebraically closed, we will not work with real
algebraic curves but rather complex algebraic curves, in other words, subsets of the
space (C
)
2
dened by equations of the form P(x, y) = 0, where C
= C \ {0}, and
P(x, y) is a polynomial with complex coefcients (which can therefore be real). For t
a positive real number, we dene the map Log
t
on (C
)
2
by
Log
t
(C
)
2
R
2
(x, y) (log
t
|x|, log
t
|y|).
We denote the image of a curve given by the equation P(x, y) = 0 under the map
Log
t
by A
t
(P), and call it the amoeba of base t of the curve. Exactly as in Section 4,
the limit of A
t
(P) when t goes to +is a tropical curve having a single vertex, with
innite rays corresponding to asymptotic directions of the amoebas A
t
(P). However,
we can obtain more interesting limiting objects by considering families of algebraic
curves, that is to say, for each t , not only the base of the logarithm changes, but also the
curve whose amoeba we are considering. Such a family of complex algebraic curves
takes the form of a polynomial P
t
(x, y), whose coefcients are given as functions of
the real number t . Therefore, for each choice of t > 1 we obtain an honest polynomial
in x and y that denes a curve in the plane.
The following theorem says that all tropical curves arise as the limit of amoebas
of families of complex algebraic curves, and provides a fundamental link between
classical algebraic geometry and tropical geometry.
Theorem 5.2 (G. Mikhalkin, H. Rullg ard). Let P
(x, y) =
i, j
a
i, j
x
i
y
j
be a
tropical polynomial. For each coefcient a
i, j
different from , choose a nite set
I
i, j
R such that a
i, j
= max I
i, j
, and choose
i, j
(t ) =

rI
i, j

i, j,r
t
r
with
i, j,r
= 0.
For every t > 0, we dene the complex polynomial P
t
(x, y) =

i, j

i, j
(t )x
i
y
j
. Then
the amoeba A
t
(P
t
) converges to the tropical curve dened by P
(x, y) when t tends

to +.
The dequantization of the curve seen in Section 4 is a particular case of the above
theorem. The amoeba of base t of the line with equation t
0
x t
0
y + t
0
1 = 0 con-
verges to the tropical line dened by 0x +0y +0. We can deduce Viros theorem
from the preceding theorem by remarking, among other details, that if the
i, j,r
are
real numbers, then the curves dened by the polynomials P
t
(x, y) are real algebraic
curves.
Exercises.
1. Construct a real tropical curve of degree 2 that realizes the same arrangement as
a hyperbola in R
2
. Do the same for a parabola. Can we construct a real tropical
curve realizing the same arrangement as an ellipse?
2. By using patchworking, show that there exists a real algebraic curve of degree 4
realizing the arrangement shown in Figure 12(b). Use the construction illustrated
in Figure 14 as a guide.
3. Show that for every degree d, there is a real algebraic curve with
d(d1)+2
2
con-
nected components.
6. FURTHER DIRECTIONS. So far, we only considered tropical polynomials in
one variable, their tropical roots, and tropical curves in the plane. Tropical geometry,
however, extends far beyond these basic objects, and we would like to showcase briey
some horizons that t within the topics of the material discussed in this text.
Hyperelds. Recall that we started our tour of tropical geometry by introducing the
tropical semi-eld T, which we obtained from the real numbers by Maslovs dequanti-
zation procedure. We then proceeded to consider the solutions to polynomial equations
over this semi-eld by choosing our denitions wisely. Since the appearance of tropi-
cal algebra, there have been several proposals to enrich this simple tropical semi-eld
to other algebraic structures, in order to better understand relations between tropical
algebra and geometry. Here, we present an example of such an enrichment, suggested
by Viro, known as hyperelds.
Similarly to a eld, a hypereld is a set equipped with two operations called mul-
tiplication and addition, which satisfy some axioms. The main difference between hy-
perelds and (semi-)elds is that addition is allowed to be multivalued; this means
that the sum of two numbers is not necessarily just one number as we are used to,
but can actually be a set of numbers. The simplest hypereld is the sign hypereld,
which consists of only three elements 0, +1, and 1. The multiplication is as usual,
i.e., (1) (1) = 1, but the multivalued addition is dened by the following table.
In fact, this hypereld is quite natural. Suppose you were to add or multiply two real
numbers but you only knewtheir signs; then the corresponding operations of the signed
hypereld give all the possible signs of the output.
+ 0 +1 1
0 {0} {+1} {1}
+1 {+1} {+1} {0, 1}
1 {1} {0, 1} {1}
There exist several hyperelds related to tropical geometry. Here we will show just
one, which resembles our tropical semi-eld quite a bit. It consists of the same under-
lying set T = R {}, and multiplication is again the usual addition. However, the
multivalued addition is now the following:
x y =
_
{max(x, y)}, if x = y,
{z T | z x}, if x = y.
Notice that can still be considered as the tropical zero, since x = {x} for
any x T.
Let us take a look at how this foreign concept of multivalued addition can simplify
some denitions and resolve some peculiarities in tropical geometry. In the tropical
hypereld T, we declare x
0
to be a root of a tropical polynomial P(x) if the set P(x
0
)
contains the tropical zero . Notice that both denitions of the roots of a polynomial
coincide, whether it is considered over the tropical hypereld or the tropical semi-eld.
In the case of curves, every classical line in the plane is given by the equation
y = ax +b. Points in T
2
satisfying the same equation y = ax +b = max(a +x, b)
over the tropical semi-eld form a piecewise linear graph, as depicted in Figure 2(a).
However, this is not at all a tropical curve, as the balancing condition is not satised
at (b a, b), where the linearity breaks. If, instead, we replace our semi-eld addi-
tion with the new mutlivalued hypereld addition, then there is one point where the
function P(x) = (a + x) b is truly multivalued:
P(b a) = {x R {} | x b} .
Hence, the previous graph grows a vertical tail at x = b a when the function is
considered over the tropical hypereld; this is the dotted line in Figure 2(a). Therefore,
we obtain the tropical line that we rst encountered in Figure 1.
In conclusion, we see that using the tropical hypereld already simplies some
basic aspects of tropical geometry. The denition of the zeroes of polynomials seems
more natural, and sets in R
2
dened by a simple equation of the form y = P(x) are
tropical curves, which was not the case over the tropical semi-eld.
Tropical modications. If we compare the above example again to the classical situ-
ation, then we may notice another phenomenon particular to tropical geometry. Clas-
sically, the shape of a line does not change depending on where it lives. For example, a
non-vertical line in R
2
projects bijectively to the x-axis; more generally, any line in R
n
is in bijection with a coordinate axis via a projection. This dramatically fails in tropical
geometry. The projection to the x-axis of a generic tropical line in T
2
no longer gives
a bijection to T, since the vertical edge of the line is contracted to a point. In general,
the possible shapes of a tropical line depend on the space T
n
where it sits. We can,
nevertheless, describe them all. A generic line in T
n
has the shape of a line in T
n1
with an additional tail grown at an interior point. We depicted in Figure 16 some pos-
sible shapes of tropical lines in T
3
and T
5
. Notice that, according to this description, a
generic tropical line in T
n
has n +1 ends, and is always a tree, meaning a graph with
no cycles.
10
(a) A tropical line in T
3
(b) A snowake line in T
5
(c) A caterpillar line in T
5
Figure 16. Different tropical lines
This phenomenon is not specic to tropical lines; in general, there are innitely
many possible ways of modelling an object from classical geometry by a tropical ob-
ject. This presents tropical geometers with some interesting problems, such as the fol-
lowing. What do the different tropical models of the same classical object have in
common? How are two different models of the same object related? Does there exist a
model better than the others?
It turns out that different tropical models are related by an operation known as trop-
ical modication. In fact, the example we just saw with a tropical line in T
2
along with
the projection to T in the vertical direction is precisely a tropical modication of T. Let
us take a look at another example in two dimensions. Consider the tropical polynomial
10
Alternatively, a tree is a graph in which there is exactly one path joining any two vertices.
in two variables P(x, y) = x y 0 over the tropical hypereld. The subset of T
3
dened
11
by z P(x, y) is drawn in Figure 17. It is a piecewise linear surface with
2-dimensional faces; three of them are in the downward vertical direction and arise
due to P(x, y) being multivalued. These six faces are glued along any pair of the four
rays starting at (0, 0, 0) and in the directions
u
1
= (1, 0, 0), u
2
= (0, 1, 0), u
3
= (0, 0, 1), and u
0
= (1, 1, 1).
0
y
x
T
2
(0, t, 0)
(t, t, t )
(t, 0, 0)
(0, 0, t )
Figure 17. A modication of the tropical afne plane T
2
Just as in Section 5, we may consider the limit of amoebas Log
t
(P) of the plane
P R
3
dened by the equation x + y + z +1 = 0. This limit is again the piecewise
linear surface , and therefore this latter provides another tropical model of a classical
plane!
12
Notice that once again the projection that forgets the third coordinate, estab-
lishes a bijection between P and R
2
, but is not a bijection between and T
2
. This
projection map T
2
is a tropical modication of the tropical plane T
2
.
We might ask why we would want to use this more complicated model of a plane
rather than just T
2
. Let us show an example. Consider the real line
L = {(x, y) R
2
| x + y +1 = 0} R
2
,
11
Here, we use the tropical hypereld for consistency. Alternatively, denitions of Sections 1 and 2 admit
natural generalizations to polynomials with any number of variables, and is the tropical surface dened by
the tropical polynomial 0 + x + y + z.
12
The same phenomenon as in the case of curves happens here. We may consider for simplicity a plane in
R
3
, because it is dened by a linear equation. However, if we want to study tropical surfaces of higher degree
and their approximations by amoebas of classical surfaces as in Theorem 5.2, then we have to leave R for its
algebraic closure C.
and let L T
2
be the corresponding tropical line, i.e., L = lim
t +
Log
t
(L). Let L
t
be a family of lines in R
2
whose amoebas converge to some tropical line L
T
2
as
in Theorem 5.2, and dene p
t
= L L
t
. Now, let us ask the following question: Can
we determine lim
t +
Log
t
( p
t
) by only looking at the tropical picture?
If the set-theoretic intersection of L and L
consists of a single point p, then

Log
t
( p
t
) has no choice but to converge to p. But what happens when this set-theoretic
intersection is innite?
Suppose that L
is given by the tropical polynomial 1 x + y + 0, so that the

position of L and L
is as in Figure 9(a) with L in blue and L
in red. Then the stable

intersection of L and L
is the vertex of L
, that is to say the point (1, 0). Note that

this is independent of the family L
t
, as long as Log
t
(L
t
) converges to L
. However,
depending on the family L
t
, the point lim
t +
Log
t
( p
t
) may be located anywhere on
the half-line {(x, 0), x 1}. Indeed, given the family
L
t
=
_
(x, y) R
2
| (t +1)x + y +(1 t
b+1
) = 0
_
with b 1,
then by Theorem 5.2, lim
t +
Log
t
(L
t
) = L
. Yet we have p
t
= (t
b
, 1 t
b
), and so
lim
t +
Log
t
( p
t
) = (b, 0). Similarly, we get lim
t +
Log
t
( p
t
) = (, 0) by taking
L
t
=
_
(x, y) R
2
| t x + y +1 = 0
_
.
This shows that using the tropical model L, L
T
2
, it is impossible to extract in-
formation about the location of the intersection point of L and L
t
. It turns out that
changing the model T
2
via a tropical modication unveils the location of the limit of
the intersection point p
t
. To obtain our new model, let us consider again the plane
P R
3
with equation x + y + z +1 = 0, and the projection
: P R
2
(x, y, z) (x, y).
The map is clearly a bijection, and
1
(x, y) = (x, y, x y 1). The inter-
section of P with the plane z = 0 in R
3
is precisely the line L. This implies that
Log
t
(
1
(L)) {z = } and
lim
t +
Log
t
(
1
(L)) = {z = } T
3
.
If L
t
is a family of lines in R
2
as above, then
1
(L
t
) is a family of lines in P, and,
moreover, the intersection point p
t
must satisfy
1
( p
t
) {z = 0}. In particular, we
have
lim
t +
Log
t
(
1
( p
t
)) {z = }.
Now it turns out that L
= lim
t +
Log
t
(
1
(L
t
)) is a tropical line in , which
intersects the plane {z = } in a single point! Therefore, upon projection back to
R
2
, this point is nothing else but lim
t +
Log
t
( p
t
); in other words, the location of
lim
t +
Log
t
( p
t
) is now revealed by considering the tropical picture in T
3
. In Figure
18 we depict the case when L
t
is given by (t +1)x + y +(1 t
b+1
) = 0.
Hence, there is no unique or canonical choice of a tropical model of a classical
space, and tropical modication is the tool used to pass between the different models.
There is a single space that captures in a sense all the possible tropical models, known
L
L
(b, 0, )
Figure 18. The tropical lines L and L
in the modied plane

as the Berkovich space of the classical object. The structure of a Berkovich space is
extremely complicated due to the innitely many different tropical models. For ex-
ample, the Berkovich line is an innite tree. In a sense, a tropical model gives us just
a snapshot of the complicated Berkovich space and modications are what allow us to
change the resolution.
Higher (co)dimension. In the rst part of this text, our main object of study was
tropical curves in the plane. Even in this limited case, there are many applications
to classical geometry. In higher dimensions, things become a bit trickier as the links
between the tropical and classical worlds become more intricate.
As we mentioned earlier, the denitions of Sections 1 and 2, as well as Theorem 5.2,
admit natural generalizations to polynomials with any number of variables. A tropical
polynomial in n variables P(x
1
, . . . , x
n
) denes a tropical hypersurface in T
n
, which
is the corner locus of the graph of P equipped with some weights. A tropical hypersur-
face in T
n
is glued fromat pieces, has dimension n 1, satises a balancing condition
similar to curves, and arises as the limit of amoebas of a family of complex hypersur-
faces. Because of this, we say that all tropical hypersurfaces are approximable.
It is also natural to consider tropical objects of dimension less than n 1 in T
n
. A
k-dimensional tropical variety is a weighted polyhedral complex of dimension k, sat-
isfying a generalization of the balancing condition that we saw for curves. It would be
too technical and probably useless to give here the balancing condition in full general-
ity. Nevertheless, in the case of curves, i.e., when k = 1, the denition is the same as in
Section 2: A tropical curve in R
n
is a graph whose edges have a rational direction and
are equipped with positive integer weights. In addition, the graph must satisfy the bal-
ancing condition given in Section 2 at each vertex. Unfortunately, Theorem 5.2 does
not generalize to the case of higher codimensions. There exist tropical varieties, said to
be not approximable, which do not arise as limits of amoebas of complex varieties of
the same degree. For our purposes, we may think of a complex variety in C
n
as being
the solution set of a system of polynomial equations in n variables.
Already in T
3
there are many examples of tropical curves that are not approximable.
More intricately, there are examples of pairs of tropical varieties Y X T
n
for
which both Y and X are approximable independently; however, there do not exist
approximations Y
t
and X
t
of Y and X with Y
t
X
t
. Let us give an explicit example
of such a pair.
Consider the modied tropical plane T
3
we just encountered, which is ap-
proximated by the amoebas of the complex plane P C
3
dened by the equation
x + y + z +1 = 0. Now, the union of three rays in the directions
(2, 3, 0), (0, 1, 1), (2, 2, 1)
with each ray equipped with weight 1 is a tropical curve C T
3
. We may check that
the balancing condition is satised:
(2, 3, 0) +(0, 1, 1) +(2, 2, 1) = 0.
Moreover, C is contained in the tropical plane , since we may express the three rays
as
(2, 3, 0) = 2u
1
+3u
2
, (0, 1, 1) = u
0
+u
1
, (2, 2, 1) = 2u
0
+3u
3
,
where the u
i
are the directions of the rays of given in Section 6. The tropical curve
C is approximable,
13
but there are no approximations C
t
and P
t
of C and with
C
t
P
t
. The full proof of this statement involves several steps, and we only focus
here on the essential one. Hence, let us assume that we may take both families P
t
and C
t
to be constant, with P
t
equal to the above plane P for all t . That is to say,
we are reduced to showing that there does not exist a complex curve C P such that
C = lim
t +
Log
t
(C).
Suppose, on the contrary, that such a complex curve C exists. The projection of C
to T
2
that forgets the z-coordinate is a tropical curve C
as dened in Section 2. The

curve C
and its Newton polygon are shown in Figure 19. A polynomial with such a
Newton polygon is of degree 3 and has the form P(x, y) = ax
3
+by
2
+ , where
the monomials after the dots have higher degree. A complex curve in C
2
with such
an equation has a type of singularity at (0, 0) known as a cusp. Hence the ray of C in
the direction (2, 3, 0) = 2u
1
+ 3u
2
tells us that the projection of C has a cusp at
(0, 0). Since C is contained in P and P projects bijectively to C
2
, we deduce that C
itself has a cusp at (0, 0, 1). In a similar way, the ray in direction 2u
0
+3u
3
implies
that C must have a second cusp at innity.
14
However, it is known that a plane curve
of degree 3 can have at most one cusp, which contradicts the existence of C. This tells
us that our tropical curve is not approximable by a complex algebraic curve of degree
3 contained in the plane P.
2
Figure 19. The projection of the curve C P to T
2
along with its dual polygon
13
For example, check that C = lim
t +
Log
t
(C), where C = {(u
2
(u 1), u
3
, (u 1)), u C}.
14
This statement makes sense in the framework of projective geometry.
Many other examples of non-approximable tropical objects exist, including lines in
surfaces, curves in space, and even linear spaces. What must be stressed is that, from
a purely combinatorial viewpoint, it is very difcult to distinguish these pathological
tropical objects from approximable ones. This is just one of the problems that makes
tropical geometry continue to be a challenging and exciting eld of research, especially
in terms of its relations to classical geometry.
7. REFERENCES. The references given belowcontain the proofs of statements con-
tained throughout the text. They also present the many directions in which tropical ge-
ometry is developing. The references to the literature go far beyond what we mention
here; it would be impossible to provide a complete list. However, we hope to give an
account of the multitude of perspectives on tropical geometry.
First, we should note that tropical curves rst appeared under the name of ( p, q)-
webs, in the work of the physicists O. Aharony, A. Hanany, and B. Kol [2]. In mathe-
matics, roots of tropical geometry can be traced back at least to the work of Bergman
[8] on logarithmic limit sets of algebraic varieties, to Bieri and Groves [10] on non-
archimedean valuations, to Viro on patchworking (see references below), as well to the
study of toric varieties, which provide a rst relation between convex polytopes and
algebraic geometry, see for example [17], [19].
There are other general introductions to tropical geometry. The texts [9] and [49]
are aimed at the interested reader with a minimal background in mathematics, similar
to the one assumed in this text. There the authors emphasize different applications
of tropical geometry, enumerative geometry in [9], and phylogenetics in [49]. More
condent readers can also read the works of [20], [24], [25], [30], [31], [36], [37], and
[46]. For established geometers, we recommend the state of the art [33] and [35].
In particular, stable intersections of tropical curves in R
2
are discussed in detail
in [46]. The proof of the B ezout theorem we gave is contained in [20]. Foundations
of a more sophisticated tropical intersection theory are exposed in [35], and further
developed in [4], [28], [47].
Amoebas of algebraic varieties were introduced in [22]. The interested reader can
also see the texts [43], [56], and the more sophisticated [42]. For a deeper look at
patchworking, amoebas, and Maslovs dequantization, as well as their applications to
Hilberts 16th problem, we point you to the texts [32], [33], [55], and [57], as well as
the website [59]. We especially recommend the text [26], dedicated to a large audience,
which explains how patchworking has been used to disprove the Ragsdale conjecture,
which had been open for decades.
For more on tropical hyperelds, we point the reader toward [58]. As we men-
tioned in the text, there are other enrichments of the tropical semi-eld, some examples
of which can be found in [14], [18], and [27]. Tropical modications were intro-
duced by Mikhalkin in the above-mentioned text [35], and their relations to Berkovich
spaces can be found in [44] and [45]. This perspective from Berkovich spaces al-
lows one to relate tropical geometry to algebraic geometry, not only over C, but
also over any eld equipped with a so-called non-archimedean valuation (e.g., the
p-adic numbers equipped with the p-adic valuation). For this point of view, we re-
fer, for example, to [1], [7], [40], and references therein. The approximation prob-
lem has been studied mostly in the case of curves. For tropical curves, see [5], [29],
[34], [35], [38], [39], [48], and [52]. For a look at the approximation problem of
curves in surfaces, as in the example provided in the last section, see [11], [13], and
[21]. Among all sources of motivation to study the latter problem, we recommend
the very nice study of tropical lines in tropical surfaces in R
3
by Vigeland in [53]
and [54].
To end this introduction to tropical geometry, we point out that this subject has
very successful applications in many elds other than Hilberts 16th problem, such
as enumerative geometry [34], algebraic geometry [51], mirror symmetry [23], mathe-
matical biology [6], [41], computational complexity [3], computational geometry [16],
[50], and algebraic statistics [15], [41], just to name a few.
ACKNOWLEDGMENT. We are very grateful to Oleg Viro for his encouragement and extremely helpful
comments. We also thank Ralph Morisson and the anonymous referees for their useful suggestions on a pre-
liminary version of this text.
REFERENCES
1. D. Abramovich, L. Caporaso, S. Payne, The tropicalization of the moduli space of curves (2012), avail-
able at http://arxiv.org/abs/1212.0373.
2. O. Aharony, A. Hanany, B. Kol, Webs of (p,q) 5-branes, ve dimensional eld theories and grid diagrams,
J. High Energy Phys. 1 (1998) 150.
3. M. Akian, S. Gaubert, A. Guterman, Tropical polyhedra are equivalent to mean payoff games, Internat.
J. Algebra Comput. 22 (2012) 1250001, available at http://arxiv.org/abs/0912.2462.
4. L. Allermann, J. Rau, First steps in tropical intersection theory, Math. Z. 264 (2010) 633670.
5. O. Amini, M. Baker, E. Brugall e, J. Rabinoff, Lifting harmonic morphisms of tropical curves, metrized
complexes, and Berkovich skeleta (2013), available at http://arxiv.org/abs/1303.4812.
6. F. Ardila, C.J. Klivans, The Bergman complex of a matroid and phylogenetic trees, J. Comb. Theory Ser.
B 96 (2006) 3849.
7. M. Baker, S. Payne, J. Rabinoff, Nonarchimedean geometry, tropicalization, and metrics on curves
(2011), available at http://arxiv.org/abs/1104.0320.
8. G. M. Bergman, The logarithmic limit-set of an algebraic variety, Trans. Amer. Math. Soc. 157 (1971)
459469.
9. G eom etrie Tropicale. Edited by N. Berline, A. Plagne, and C. Sabbah.

Editions de l
Ecole Polytechnique,
Palaiseau, 2008.
10. R. Bier, J. Groves, The geometry of the set of characters induced by valuations, J. Reine Angew. Math.
347 (1984) 168195.
11. T. Bogart, E. Katz, Obstructions to lifting tropical curves in surfaces in 3-space, SIAM J. Discrete Math.
26 (2012) 10501067.
12. E. Brugall e, Un peu de g eom etrie tropicale, Quadrature 74 (2009) 1022.
13. E. Brugall e, K. Shaw, Obstructions to approximating tropical curves in surfaces via intersection theory
(2011), available at http://arxiv.org/abs/arXiv:1110.0533.
14. A. Connes, C. Consani, The universal thickening of the eld of real numbers (2012), http://arxiv.
org/abs/arXiv:1202.4377.
15. M. A. Cueto, J. Morton, B. Sturmfels, Geometry of the restricted Boltzmann machine, in Algebraic
Methods in Statistics and Probability II. Vol. 516, Contemp., Math. American Mathematical Society,
Providence, RI, 2010. 135153.
16. M. A. Cueto, E. A. Tobis, J. Yu, An implicitization challenge for binary factor analysis, J. Symbolic
Comput. 45 no. 12 (2012) 12961315.
17. V. I. Danilov, A. G. Khovanski, Newton polyhedra and an algorithm for calculating Hodge-Deligne
numbers, Izv. Akad. Nauk SSSR Ser. Mat. 50 (1986) 925945.
18. A. Dress, W. Wenzel, Algebraic, tropical, and fuzzy geometry, Beitr. Algebra Geom. 52 (2011) 431461.
19. W. Fulton, Introduction to Toric Varieties. Ann. Math. Studies. Vol. 131, Princeton University Press,
Princeton, NJ, 1993.
20. A. Gathmann, Tropical algebraic geometry, Jahresber. Deutsch. Math.-Verein. 108 (2006) 332.
21. A. Gathmann, K. Schmitz, A. Winstel, The realizability of curves in a tropical plane (2013), available at
http://arxiv.org/abs/arXiv:1307.5686.
22. I. M. Gelfand, M. M. Kapranov, A. V. Zelevinsky. Discriminants, Resultants, and Multidimensional De-
terminants. Mathematics: Theory & Applications, Birkh auser Boston, Boston, MA, 1994.
23. M. Gross, Tropical geometry and mirror symmetry. CBMS Regional Conference Series in Mathematics.
Vol. 114, published for the Conference Board of the Mathematical Sciences Washington, DC, 2011.
24. I. Itenberg, G. Mikhalkin, Geometry in the tropical limit, Math. Semesterber. 59 (2012) 5773.
25. I. Itenberg, G. Mikhalkin, E. Shustin, Tropical Algebraic Geometry. Vol. 35, Oberwolfach Seminars
Series, Birkh auser, 2007.
26. I. Itenberg, O. Viro, Patchworking algebraic curves disproves the Ragsdale conjecture, Math. Intelli-
gencer 18 no. 4 (1996) 1928.
27. Z. Izhakian, L. Rowen, A guide to supertropical algebra, in Advances in Ring Theory. Trends Math.
Birkh auser/Springer Basel AG, 2010. 283302.
28. E. Katz, A tropical toolkit, Expo. Math. 27 (2009) 136.
29. , Lifting tropical curves in space and linear systems on graphs, Adv. Math. 230 (2012) 853875.
30. D. Maclagan, Introduction to tropical algebraic geometry, in Tropical Geometry and Integrable Systems.
Contemp. Math. Vol. 580, American Mathematical Society, Providence, RI, 2012. 119.
31. D. Maclagan, B. Sturmfels, Introduction to Tropical Geometry. Book in progress, available at http:
//homepages.warwick.ac.uk/staff/D.Maclagan/papers/papers.html.
32. G. Mikhalkin, Real algebraic curves, the moment map and amoebas, Ann. of Math. 151 (2000) 309326.
33. , Amoebas of algebraic varieties and tropical geometry, in Different Faces of Geometry. Int. Math.
Ser. (N.Y.), Vol. 3, Kluwer/Plenum, New York, 2004. 257300.
34. , Enumerative tropical algebraic geometry in R
2
, J. Amer. Math. Soc. 18 no. 2 (2005) 313377.
35. , Tropical geometry and its applications, in International Congress of Mathematicians. Vol. 2,
Eur. Math. Soc., Z urich, 2006. 827852.
36. , What is . . . a tropical curve? Notices Amer. Math. Soc. 54 no. 4 (2007) 511513.
37. G. Mikhalkin, J. Rau, Tropical geometry. Book in progress.
38. T. Nishinou, Correspondence theorems for tropical curves (2009), available at http://arxiv.org/
abs/arXiv:0912.5090.
39. T. Nishinou, B. Siebert, Toric degenerations of toric varieties and tropical curves, Duke Math. J. 135 no. 1
(2006) 151.
40. J. Rabinoff, Tropical analytic geometry, Newton polygons, and tropical intersections, Adv. Math. 229
(2012) 31923255.
41. Algebraic Statistics for Computational Biology. Edited by L. Pachter and B. Sturmfels. Cambridge Uni-
versity Press, New York, 2005.
42. M. Passare, H. Rullg ard, Amoebas, Monge-Amp` ere measures, and triangulations of the Newton polytope,
Duke Math. J. 121 no. 3 (2004) 481507.
43. , How to compute

1/n
2
by solving triangles, Amer. Math. Monthly 115 (2008) 745752.
44. S. Payne, Topology of nonarchimedean analytic spaces and relations to complex algebraic geometry,
available at http://arxiv.org/abs/arXiv:1309.4403.
45. , Analytication is the limit of all tropicalizations, Math. Res. Lett. 16 (2009) 543556.
46. J. Richter-Gebert, B. Sturmfels, T. Theobald, First steps in tropical geometry, in Idempotent Mathematics
and Mathematical Physics. Contemp. Math. Vol. 377, American Mathematical Society, Providence, RI,
2005. 289317.
47. K. Shaw, A tropical intersection product in matroidal fans, SIAM J. Discrete Math. 27 (2013) 459491.
48. D. Speyer, Uniformizing tropical curves I: Genus zero and one (2007), available at http://arxiv.
org/abs/0711.2677.
49. D. Speyer, B. Sturmfels, Tropical mathematics, Math. Mag. 82 (2009) 163173.
50. B. Sturmfels, J. Tevelev, J. Yu, The Newton polytope of the implicit equation, Mosc. Math. J. 7 (2007)
327346.
51. J. Tevelev, Compactications of subvarieties of tori, Amer. J. Math. 129 (2007) 10871104.
52. I. Tyomkin, Tropical geometry and correspondence theorems via toric stacks, Math. Ann. 353 (2012)
945995.
53. M. D. Vigeland, Smooth tropical surfaces with innitely many tropical lines, Arkiv f or Matematik 48
(2009) 177206.
54. , Tropical lines on smooth tropical surfaces (2007), available at http://arxiv.org/abs/
0708.3847.
55. O. Viro, Dequantization of real algebraic geometry on logarithmic paper, in European Congress of Math-
ematics, Vol. I (Barcelona, 2000), Vol. 201 of Progr. Math., Birkh auser, Basel, 2001. 135146.
56. , What is an amoeba? Notices of the AMS 49 (2002) 916917.
57. , From the sixteenth Hilbert problem to tropical geometry, Japanese Journal of Mathematics 3
(2008) 185214.
58. , Hyperelds for tropical geometry I. Hyperelds and dequantization (2010), available at http:
//arxiv.org/abs/1006.3034.
59. , Patchworking, texts for students available at https://www.math.sunysb.edu/
~
oleg/
patchworking.html.
ERWAN BRUGALL
E (brugalle@math.cnrs.fr) got his Ph.D. in Rennes in 2004. After holding a position at

Jussieu University, he is now working at

Ecole Polytechnique in Paris. His main interests are real, complex,
and tropical algebraic varieties, and enumerative geometry.
Ecole polytechnique, Centre Math ematiques Laurent Schwartz, 91 128 Palaiseau Cedex, France
KRISTIN SHAW (shawkm@math.utoronto.ca) received her Ph.D. from the University of Geneva in 2011
and is now a postdoctoral fellow at the University of Toronto. Her research is mainly in complex and tropical
geometry, with particular interests in surfaces, as well as intersection theory.
Department of Mathematics, University of Toronto, 40 St. George St., Toronto, Ontario, M5S 2E4, Canada
Another Algebraic Pythagorean Proof
We refer the reader to Figure 1 below.
Slide the right triangle DBD
1
along the cathetus(leg) (BD = a) by a distance
equal to the hypothenuse (DD
1
= c).The new right triangle is EGE
1
. We create
the rhombus DEE
1
D
1
(each side is equal to c).
Since the diagonals in the rhombus are perpendicular (DE
1
and ED
1
) the
angles BD
1
E = GDE
1
are the same. Thus, the right triangles D
1
EB and
GDE
1
are similar (A, A, A):
BD
1
BE
=
DG
GE
1
b
c a
=
c +a
b
a
2
+b
2
= c
2
. (1)
Thus we obtain the Pythagorean Theorem.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . .
. . . . . . .
. . .
B a E c a
c
c
G D
D
1
E
1
b b
90
90
c
Submitted by Marcelo Brafman, Tel Aviv University
MSC: Primary 00A05
Four Quotient Set Gems
Bryan Brown, Michael Dairyko, Stephan Ramon Garcia,
Bob Lutz, and Michael Someck
Abstract. Our aim in this note is to present four remarkable facts about quotient sets. These
observations seem to have been overlooked by the MONTHLY, despite its intense coverage of
quotient sets over the years.
INTRODUCTION. If A is a subset of the natural numbers N = {1, 2, . . .}, then we
let R(A) = {a/a
: a, a
A} denote the corresponding quotient set (sometimes called

a ratio set). Our aim in this short note is to present four remarkable results, which seem
to have been overlooked in the MONTHLY, despite its intense coverage of quotient sets
over the years [3, 4, 5, 9, 10, 11, 14]. Some of these results are novel, while others have
appeared in print elsewhere but somehow remain largely unknown.
In what follows, we let A(x) = A [1, x] so that | A(x)| denotes the number of
elements in A which are x. The lower asymptotic density of A is the quantity
d(A) = liminf
n
| A(n)|
n
,
which satises the obvious bounds 0 d(A) 1. We say that A is fractionally dense
if the closure of R(A) in R equals [0, ) (i.e., if R(A) is dense in [0, )).
Our four gems are as follows.
1. The set of all natural numbers whose base-b representation begins with the digit
1 is fractionally dense for b = 2, 3, 4, but not for b 5.
2. For each [0,
1
2
), there exists a set A N with d(A) = that is not fraction-
ally dense. On the other hand, if d(A)
1
2
, then A must be fractionally dense
[15].
3. We can partition N into three sets, each of which is not fractionally dense. How-
ever, such a partition is impossible using only two sets [2].
4. There are subsets of N that contain arbitrarily long arithmetic progressions, yet
that are not fractionally dense. On the other hand, there exist fractionally dense
sets that have no arithmetic progressions of length 3.
BASE-b REPRESENTATIONS. In [5, Example 19], it was shown that the set
A = {1} {10, 11, 12, 13, 14, 15, 16, 17, 18, 19} {100, 101, . . .}
of all natural numbers whose base-10 representation begins with the digit 1 is not
fractionally dense. This occurs despite the fact that d(A) =
1
9
, so that a positive pro-
portion of the natural numbers belongs to A. The consideration of other bases reveals
the following gem.
MSC: Primary 11A99, Secondary 11B99; 11A16
Gem 1. The set of all natural numbers whose base-b representation begins with the
digit 1 is fractionally dense for b = 2, 3, 4, but not for b 5.
To show this, we require the following more general result.
Proposition 1. Let 1 < a b. The set
A =
_
k=0
[b
k
, ab
k
) N (1)
is fractionally dense if and only if b a
2
. Moreover, we also have
d(A) =
a 1
b 1
.
Proof. We rst compute d(A). Since the counting function | A(x)| is nondecreas-
ing on each interval of the form [b
k
, ab
k
) and constant on each interval of the form
[ab
k
, b
k+1
), it follows that
liminf
n
| A(n)|
n
= liminf
n
1
b
n
n1
k=1
[b
k
, ab
k
) N
= lim
n
1
b
n
n1
k=1
(a 1)b
k
= lim
n
(a 1)(b
n
1)
b
n
(b 1)
=
a 1
b 1
. (2)
Note that in order to obtain (2), we used the fact that the difference between ab
k
b
k
and |[b
k
, ab
k
) N| is 1. Having computed d(A), we now turn our attention to
fractional density.
Case 1. Suppose that a
2
< b. By construction, each quotient of elements of A be-
longs to an interval of the form I
= (a
1
b
, ab
) for some integer . If j < k, then

a
2
< b
kj
, whence ab
j
< a
1
b
k
so that every element of I
j
is strictly less than ev-
ery element of I
k
. Therefore, R(A) contains no elements in any interval of the form
[ab
, a
1
b
+1
], each of which is nonempty since a
2
< b. Thus A is not fractionally
dense.
Case 2. Now let b a
2
, noting that
(0, ) =
_
j Z
_
b
j
a
, b
j
_
[b
j
, ab
j
).
Suppose that belongs to an interval of the form [b
j
, ab
j
) for some integer j . Given
> 0, let k be so large that 1 < b
k
, and observe that
b
j +k
b
k
< ab
j +k
.
AugustSeptember 2014] FOUR QUOTIENT SET GEMS 591
Let be the unique natural number satisfying
b
j +k
+ b
k
< b
j +k
+ +1 (3)
and
0 (a 1)b
j +k
1. (4)
From (3), we nd that
0 b
k
(b
j +k
+) < 1,
from which it follows that
0
b
j +k
+
b
k
< .
Since (4) ensures that b
j +k
+ belongs to the interval [b
j +k
, ab
j +k
), we conclude
that (b
j +k
+ )/b
k
belongs to R(A). A similar argument applies if belongs to an
interval of the form [
b
j
a
, b
j
). Putting this all together, we conclude that A is fractionally
dense.
To see that Gem 1 follows from the preceding proposition, observe that if a = 2
and b 2 is an integer, then the set A dened by (1) is precisely the set of all natural
numbers whose base-b representation begins with the digit 1. The inequality b a
2
holds precisely for the bases b = 2, 3, 4 and fails for all b 5.
CRITICAL DENSITY. Having seen that there exist sets that are not fractionally
dense, yet whose lower asymptotic density is positive, it is natural to ask whether
there exists a critical value 0 < 1 such that d(A) ensures that A is fractionally
dense. The following gem establishes the existence of such a critical density, namely
=
1
2
.
Gem 2. If d(A)
1
2
, then A is fractionally dense. On the other hand, if 0 <
1
2
,
then there exists a subset A N with d(A) = that is not fractionally dense.
Establishing this result will take some work, although we are now in a position
to prove the second statement (originally obtained by Strauch and T oth [15, Thm. 1]
and by Harman [8, p. 167] using slightly different methods). If 0 <
1
2
, then we
may write =
1
2+
where > 0. Now let a = 1 +

2
and b = 1 + +
1
2
2
so that
1 < a
2
< b. For these parameters, Proposition 1 tells us that the set A dened by (1)
satises d(A) = and fails to be fractionally dense. If = 0, then we note that the set
A = {2
n
: n N} is not fractionally dense and has d(A) = 0.
Thus, we can construct sets having lower asymptotic density arbitrarily close to
1
2
,
yet which fail to be fractionally dense. On the other hand, it is clear that fractionally
dense sets with d(A) =
1
2
exist. Indeed, one such example is A = {2, 4, 6, . . .}. How-
ever, the question of whether a non-fractionally dense set can have lower asymptotic
density equal to the critical density
1
2
is much more difcult.
In the late 1960s,

Sal at proposed an example of a set A N that is not fractionally
dense and such that d(A) =
1
2
[12, p. 278]. However,

Sal ats example was awed, as
pointed out in the associated corrigendum [13]. In 1998, Strauch and T oth established
a more general result [15, Thm. 2], which implies that a set satisfying d(A)
1
2
must
be fractionally dense (fortunately, the lengthy corrigendum [16] to this paper does not
affect the result in question). For d(A) >
1
2
, the fractional density of A also follows
from a sophisticated theorem in metric number theory [8, Thm. 6.6] (although Harman
was kind enough to show us an elementary proof in this case). The proof below, which
covers the critical case d(A) =
1
2
, is essentially due to Strauch and T oth.
Theorem 2. If d(A)
1
2
, then A is fractionally dense.
Proof. Let d(A)
1
2
and suppose toward a contradiction that 0 < < 1 and
R(A) (, ) = . Noting that A must be innite, we enumerate the elements of A
in increasing order a
1
< a
2
< . Let k be a natural number, which is so large that
k > 1, and let 0 < < 1. For all m and n in N, let
J
n
m
=
_
a
k(n+m)
, (a
k(n+m)
+k)
_
.
For each n in N, we claim that the intervals
J
n
0
, J
n
1
, . . . , J
n
nn1
, (a
kn
, a
kn
), (5)
used below in a delicate counting argument, are pairwise disjoint. Indeed, since the a
i
are integers, it follows that
(a
k(n+m)
+k) (a
k(n+m)+k
) = (a
k(n+m+1)
),
so that the right endpoint of J
n
m
is at most the left endpoint of J
n
m+1
. A similar argument
shows that the right endpoint of J
n
nn1
is at most a
kn
.
Next, let n be so large that n

. We claim that
J
n
m

_
a
k(n+m)
, a
k(n+m)
_
(6)
for each m = 0, 1, . . . , n n 1. To see this, note that
k

kn a
kn
< a
kn+m
,
which implies that (a
k(n+m)
+k) a
k(n+m)
so that (6) holds.
Since (a
n
, a
n
) A = by assumption, it follows now that the intervals (5) are
contained in the complement [0, )\A. Letting B = N \ A, a nave counting argu-
ment gives
|B(a
kn
)|
_
( )a
kn
1
_
+(n n)(k 1).
Dividing through by a
kn
and taking limits superior yields
1 d(A) = limsup
n
|B(n)|
n
limsup
n
|B(a
kn
)|
a
kn
limsup
n
_
( )a
kn
1
a
kn
+
(n n)(k 1)
a
kn
_
+liminf
n
(n n)(k 1)
a
kn
1

+(1 ) liminf
n
kn n
a
kn
1

+(1 )
_
liminf
n
kn
a
kn
1
k
limsup
n
kn
a
kn
_
1

+(1 )
_
d(A)
1
k
_
.
Taking the limit as 0 and k and rearranging gives
_
1 +

_
d(A)

. (7)
If d(A)
1
2
, then the preceding inequality implies that , a contradiction.
The astute reader will note that we have actually shown that if (, ) R(A) = ,
then (7) must hold. In fact, with only a little more work, we can show that
d(A)

min
_
d(A), 1 d(A)
_
and d(A) 1 ( ), where
d(A) = limsup
n
| A(n)|
n
denotes the upper asymptotic density of A [15, Thm. 2]. Moreover, it is also known
that d(A) +d(A) 1 implies that A is fractionally dense [15, p. 71].
PARTITIONS OF N. Clearly, N itself is fractionally dense since R(N) = Q
(0, ), the set of positive rational numbers. An interesting question now presents
itself. If N is partitioned into a nite number of disjoint subsets, does one of these
subsets have to be fractionally dense? The following result gives a complete answer to
this problem.
Gem 3. We can partition N into three sets, each of which is not fractionally dense.
However, such a partition is impossible using only two sets.
This gem is due to Bukor,

Sal at, and T oth [2]. These three, along with P. Erd os,
later generalized the second statement by showing that if the set A is presented as
an increasing sequence a
1
< a
2
< satisfying lim
n
a
n+1
/a
n
= 1, then for each
B A, either B or A\B is fractionally dense [1].
The proof of our third gem is contained in the following two results.
Proposition 3. There exist disjoint sets A, B, C N, none of which are fractionally
dense, such that N = A B C.
Proof. Let
A =
_
k=0
_
5
k
, 2 5
k
_
N,
B =
_
k=0
_
2 5
k
, 3 5
k
_
N,
and
C =
_
k=0
_
3 5
k
, 5 5
k
_
N,
so that A, B, and C consist of those natural numbers whose base-5 expansions be-
gin, respectively, with 1, with 2, and with 3 or 4. We consider only C here, for the
remaining two cases are similar (in fact, the set A is already covered by Proposition
1). Observe that each quotient of elements of C is contained in an interval of the form
I
=
_
3
5
5
,
5
3
5
_
for some integer . If j < k, then every element of I
j
is strictly less
than every element of I
k
, since
5
3
5
j
<
3
5
5
k
holds if and only if
25
9
< 5
kj
. Thus,
R(C) [
5
3
5
,
3
5
5
+1
] = for any integer .
Theorem 4. If A and B are disjoint sets such that N = A B, then at least one of
A, B is fractionally dense.
Proof. Without loss of generality, we may assume that both A and B are innite.
Suppose, toward a contradiction, that neither A nor B is fractionally dense. Thus, there
exists , > 1, and > 0 such that ( , +) R(A) and ( , +) R(B)
are both empty. Now, let n
0
N be such that
1 + + +2
n
0
< .
Since A and B are both innite, there exists n > (n
0
+1) such that n A and
n +1 B. If s =
n
1 belongs to A, then setting t = s yields
t
s

s s
s
<
1
s

1
n
0
< . (8)
Since ( , +) R(A) = , we conclude that t belongs to B. Now, observe that
n +1
t

=
n +1 s
t
<
n +1 (s 1)
t
=
1 + + +n
n
t
<
1 + +2
t
1 + +2
n
0
< .
Hence,
n+1
t
belongs to ( , +) R(B), which is a contradiction. Therefore, it
must be the case that s belongs to B.
Assuming now that s =
n
1 belongs to B, we now let t = s. Proceeding

as in (8), we can show that |
t
s
| < and hence that t belongs to A. Moreover,
n
t

n s
t
<
n (s 1)
t
=
+n s
t
=
+n (
n
1)
t
=
+ +(n
n
)
t
<
+2
n
0
< .
Thus,
n
t
belongs to ( , +) R(A), a contradiction which implies that s belongs
to neither A nor B, an absurdity, since N = A B.
Although we might be tempted to postulate a relationship between Theorems 2 and
4, lower asymptotic density has no bearing on the preceding result, since it is possible
to partition N into two subsets both having lower asymptotic density zero.
Proposition 5. There exist disjoint sets A, B N such that N = A B, d(A) =
d(B) = 0, and d(A) = d(B) = 1.
Proof. We require the StolzCes` aro Theorem [6], which tells us that if x
n
and y
n
are
two increasing and unbounded sequences of real numbers, then
lim
n
x
n+1
x
n
y
n+1
y
n
= L = lim
n
x
n
y
n
= L.
Now put the rst 1! natural numbers into A, then the next 2! into B, the next 3! into A,
and so forth, yielding the sets
A = {1, 4, 5, 6, 7, 8, 9, 34, . . .}, B = {2, 3, 10, 11, 12, . . . , 32, 33, 154, 155, . . .}.
By construction, A B = and N = A B. Let x
n
=

2n1
k=1
k! and y
n
=

2n
k=1
k!,
observing that | A(x
n
)| = | A(y
n
)| =

n
k=1
(2k 1)!. Since
| A(y
n+1
)| | A(y
n
)|
y
n+1
y
n
=
(2n +1)!
(2n +2)! +(2n +1)!
=
1
2n +3
0
and
| A(x
n+1
)| | A(x
n
)|
x
n+1
x
n
=
(2n +1)!
(2n +1)! +(2n)!
=
1
1 +
1
2n+1
1,
it follows from the StolzCes` aro Theorem that d(A) = 0 and d(A) = 1. An analo-
gous argument using x
n
=

2n
k=1
k! and y
n
=

2n+1
k=1
k! now reveals that d(B) = 0 and
d(B) = 1 as well.
ARITHMETIC PROGRESSIONS. Our nal gem concerns arithmetic progressions
of natural numbers. Recall that an arithmetic progression with common difference b
and length n is a sequence of the form a, a + b, a + 2b, . . . , a + (n 1)b. Subsets
of the natural numbers that contain arbitrarily long arithmetic progressions are often
thought of as being thick or dense in some qualitative sense.
Gem 4. There exists a subset of N containing arbitrarily long arithmetic progressions
that is not fractionally dense. On the other hand, there exists a fractionally dense set
that has no arithmetic progressions of length 3.
The rst statement of Gem 4 has already been proven. Indeed, Proposition 1 pro-
duces sets that are not fractionally dense and that contain arbitrarily long blocks of
consecutive natural numbers. We therefore need only produce a fractionally dense set
having no arithmetic progressions of length three (see also [9, Thm. 2]).
Proposition 6. The set A = {2
j
: j 2} {3
k
: k 2} is fractionally dense and con-
tains no arithmetic progressions of length three.
Proof. First, recall that Kroneckers approximation theorem [7, Thm. 440] asserts that
if > 0 is irrational, R, and > 0, then there exist n, m N such that |n
m| < . Let , > 0 and note that = log
2
3 > 0 is irrational. By the continuity
of f (x) = 2
x
at log
2
, there exists > 0 such that
| log
2
x log
2
| < = |x | < . (9)
Kroneckers theorem with = log
2
3 and = log
2
now yields n, m N so that
log
2
_
3
n
2
m
_
log
2

= |n log
2
3 log
2
m| < .
In light of (9), it follows that |3
n
/2
m
| < , whence A is fractionally dense.
We now show that A contains no arithmetic progressions of length three. Assume,
toward a contradiction, that such an arithmetic progression exists. By the denition of
A, the rst term in this progression is either a power of 2 or a power of 3. We consider
both cases separately, letting b denote the common difference in each progression.
Case 1. Suppose that 2
j
, 2
j
+b, and 2
j
+2b belong to A. Since 2
j
+2b is even and
belongs to A, it must be of the form 2
k
for some k > j , whence b = 2
k1
2
j 1
is
even since j 2. Thus, 2
j
+b must also be of the form 2
for some > j so that

2
= 2
j
+b = 2
j
+(2
k1
2
j 1
) = 2
j 1
(2
kj
+1).
Upon dividing the preceding by 2
j 1
we obtain a contradiction.
Case 2. Now suppose that 3
j
, 3
j
+ b, 3
j
+ 2b is an arithmetic progression in A of
length three that begins with 3
j
. In this case, 3
j
+2b is odd and hence must be of the
form 3
k
for some k > j . Therefore,
b =
3
k
3
j
2
= 3
j
(3
kj
1)
2
,
so that the second term 3
j
+b in our progression is divisible by 3
j
. Thus, 3
j
+b = 3
for some > j , from which it follows that

3
= 3
j
+b = 3
j
+3
j
(3
kj
1)
2
.
Since this implies that 2 3
j
= 2 +(3
kj
1), which is inconsistent modulo 3, we
conclude that A has no arithmetic progressions of length three.
ACKNOWLEDGMENT. Partially supported by National Science Foundation Grant DMS-1001614.
REFERENCES
1. J. Bukor, P. Erd os, T.

Sal at, J. T. T oth, Remarks on the (R)-density of sets of numbers, II, Math. Slovaca
47 (1997) 517256.
2. J. Bukor, T.

Sal at, J. T. T oth, Remarks on R-density of sets of numbers, Tatra Mt. Math. Publ. 11 (1997)
159165.
3. J. Bukor, J. T. T oth, On accumulation points of ratio sets of positive integers, Amer. Math. Monthly 103
(1996) 502504.
4. S. R. Garcia, Quotients of Gaussian primes, Amer. Math. Monthly 120 (2013) 851853.
5. S. R. Garcia, V. Selhorst-Jones, D. E. Poore, N. Simon, Quotient sets and Diophantine equations, Amer.
Math. Monthly 118 (2011) 704711.
6. R. Gelca, T. Andreescu, Putnam and Beyond. Springer, New York, 2007.
7. G. H. Hardy, E. M. Wright, An Introduction to the Theory of Numbers. Sixth edition. Oxford University
Press, Oxford, 2008.
8. G. Harman, Metric Number Theory. London Mathematical Society Monographs. New Series. Vol. 18.
Clarendon Press Oxford University Press, New York, 1998.
9. S. Hedman, D. Rose, Light subsets of N with dense quotient sets, Amer. Math. Monthly 116 (2009)
635641.
10. D. Hobby, D. M. Silberger, Quotients of primes, Amer. Math. Monthly 100 (1993) 5052.
11. A. Nowicki, Editors endnotes, Amer. Math. Monthly 117 (2010) 755756.
12. T.

Sal at, On ratio sets of sets of natural numbers, Acta Arith. 15 (1969) 273278.
13. , Corrigendum to the paper On ratio sets of sets of natural numbers, Acta Arith. 16 (1970) 103.
14. P. Starni, Answer to two questions concerning quotients of primes, Amer. Math. Monthly 102 (1995)
347349.
15. O. Strauch, J. T. T oth, Asymptotic density of A N and density of the ratio set R(A), Acta Arith. 87
(1998) 6778.
16. , Corrigendum to Theorem 5 of the paper: Asymptotic density of A N and density of the ratio
set R(A), Acta Arith. 103 (2002) 191200.
BRYAN BROWN is a junior at Pomona College majoring in mathematics and minoring in computer science.
bryanbrown740@gmail.com
MICHAEL DAIRYKO graduated from Pomona College in 2013 with a degree in mathematics and is cur-
rently a Ph.D. student at Iowa State University. He hopes to become a professor at a small liberal arts college
where the climate is warm. When not mathing, he enjoys the outdoors, socializing with friends, and playing
water polo.
Department of Mathematics, Iowa State University, 396 Carver Hall, Ames, IA 50011
mdairyko@iastate.edu
STEPHAN RAMON GARCIA grew up in San Jos e, California before attending U.C. Berkeley for his B.A.
and Ph.D. After graduating, he worked at U.C. Santa Barbara for several years before moving to Pomona
College in 2006. He has earned three NSF research grants and ve teaching awards from three different in-
stitutions. He is the author of fty research articles in operator theory, complex analysis, matrix analysis, and
number theory. http://pages.pomona.edu/
~
sg064747
Department of Mathematics, Pomona College, Claremont, CA 91711
Stephan.Garcia@pomona.edu
BOB LUTZ is a recent graduate from Pomona College and a Ph.D. student in mathematics at the University
of Michigan. An ardent gourmand and sometimes-author, he aspires to teach at a liberal arts college one day,
but is tempted by the more stable prospect of a career in food blogging.
Department of Mathematics, University of Michigan, 2074 East Hall, 530 Church Street,
Ann Arbor, MI 48109-1043
boblutz@umich.edu
MICHAEL SOMECK is a junior mathematics and philosophy double major at Pomona College, currently
spending a year abroad at the University of Oxford. He is interested in the ways in which mathematicians
and philosophers often use the same tool, logic, to solve completely different problems. In his free time, he
enjoys traveling, spending time with friends, reading a good book, or relaxing at a beach in his hometown of
San Diego.
msomeck@gmail.com
100 Years ago this Month in The American Mathematical Monthly
Edited by Vadim Ponomarenko
That there has been and still is an unrest among thoughtful high school teachers
of mathematics is manifest. This problem certainly exists: How are we to adapt
the courses in algebra and geometry to the experiences of high school students,
in order that these may contribute to their broader development in both a cultural
and a practical way? On the one hand we have the student who is preparing for
the university and on the other hand the student whose school life ends with the
high school or who, going on to the university, will not continue the subject of
mathematics. The one who would solve this problem must have these two classes
in view, and the solution for the one must be the solution for the other since it
is only in the largest schools that distinction can be made by having separate
classes, even if it were advisable.
Excerpted from Miscellaneous Questions 21 (1914), 239.
Blind-friendly von Neumanns Heads or Tails
Vincius G. Pereira de S a and Celina M. H. de Figueiredo
Abstract. The toss of a coin is usually regarded as the epitome of randomness, and has
been used for ages as a means to resolve disputes in a simple, fair way. Perhaps as ancient
as consulting objects such as coins and dice is the art of maliciously biasing them in order
to unbalance their outcomes. However, it is possible to employ a biased device to produce
equiprobable results in a number of ways, the most famous of which is the method suggested
by von Neumann back in 1951. This paper addresses how to extract uniformly distributed bits
of information from a nonuniform source. We study some probabilities related to biased dice
and coins, culminating in an interesting variation of von Neumanns mechanism that can be
employed in a more restricted setting where the actual results of the coin tosses are not known
to the contestants.
1. INTRODUCTION. Estimating probabilities is one of those tasks at which the
human brain seems to be not very good. Conditional probabilities, in particular, fre-
quently defyand defeatones intuition, from the uninitiated to the specialist. This
explains to a certain extent why people lose money gambling and on similar activ-
ities. Led astray by instinct, the inadvertent player overlooks probabilistic subtleties
and misestimates the odds.
In this paper, we study some probabilities that may be rather counterintuitive. Al-
though our results were formulated in the context of games between two opponents,
they could as well have been framed in terms of the general engineering problem of
converting biased randomness into unbiased randomness, a fascinating subject with
plenty of literature available (see, for instance, [3, 4, 10, 11, 13, 14]).
In Section 2, we describe a game where the winning chances depend on the fairness
of a die with n 2 sides in a nontrivial way. Indeed, even knowing beforehand the
exact probability associated to each side of the die, it may not be so easy to decide,
between two seemingly equivalent strategies, which one is the most advantageous.
Still, it is possible to show that one of the strategies is never worse than the other,
no matter the bias or the number of sides of the die. We then look at some special
cases. Particularly for n = 2, we derive the winning chances as a function of the bias,
showing how to maximize the probability that a given player wins.
In Section 3, we consider the following questions regarding three independent and
identically distributed (iid) random variables A, B, and C. Are the events C = B and
B = A always independent? When, and to which extent, may (the knowledge of) the
former affect (the probability of) the latter? The answers to these questions are closely
related to the results in Section 2.
Finally, in Section 4, we leverage the results from the previous sections into a vari-
ation of von Neumanns method of playing a fair heads or tails with a biased coin.
Numerous improvements on von Neumanns original idea have been studied over the
decades ([1, 7, 9, 12], to mention but a few). Our method handles, however, an extra
restriction: The players do not know the actual result of each coin toss; instead, they
are only aware of whether each toss produced the same result as the previous one along
a sequence of independent tosses.
MSC: Primary 60C05, Secondary 60G40
2. A TRIPLE OR TWO STRAIGHT DOUBLES? In this section, we discuss a
very simple dice game, not only for its own sake, but also for the useful inequality we
obtain from it.
Consider a die with n 2 sides. We propose a game where each player must choose
between two strategies: playing for a triple, or two consecutive doubles. In the for-
mer strategy, the player throws the die three times, scoring a point if the die produces
three identical results. In the latter, the player throws the die four times, scoring a point
if the results after the rst and the second throws match one another and the third and
fourth results also match one another. After repeating their sequences of throws a pre-
viously arranged number of times, the player with more points wins the game.
We might think that both strategies are equally good (or equally terrible, depending
on how big n is). Indeed, when playing for a triple, there are two critical moments
at which the player needs to be lucky: at the second throw, whose result is required
to match the rst one; and at the third throw, which is also required to match the
other two. When playing for two doubles in a row, there are also two such critical
moments: at the second throw, when the player wants the die to match the result that
just preceded it, and again at the fourth throw, for the same reason. In spite of their
apparent equivalence, the odds for both strategies are not always as good, and how
more advantageous one strategy is depends on the fairness of the die.
In order to get some intuition, suppose our die is strongly biased towards one partic-
ular outcome (say it results in a 1 in 90% of the throws). What difference does it make,
now, to play for a triple or two straight doubles? When playing for a triple, we have
only three chances of going wrong, that is, of throwing the die and obtaining some-
thing other than 1. (Of course, it is possible that some other-than-1 outcome appears
three times in a row, but that is denitely unlikely.) On the other hand, when playing
for two straight doubles, we have in practice four chances of going wrong, since any
other-than-1 outcome that ends up occurring will most likely remain unmatched. Now
we proceed to a more formal discussion.
Claim 1. If A, B, C, and D are iid discrete random variables, then
Pr{A = B C = D} Pr{A = B = C}.
Proof. Let be the range of possible values of A, B, C, and D. For i , let p
i
be the
probability Pr{A = i } = Pr{B = i } = Pr{C = i } = Pr{D = i } that a given variable
takes value i . Let E
2,2
denote the event that A = B and C = D, and let E
3
denote the
event that A = B = C. Since the variables are independent, we can write
Pr{E
2,2
} =
i
p
2
i
2
(1)
and
Pr{E
3
} =
i
p
3
i
. (2)
We now show that Pr{E
2,2
} Pr{E
3
}. Inspired by its use in [5], we recall the Cauchy
inequality [2, p. 373]
i
x
i
y
i
i
x
2
i
i
y
2
i
.
AugustSeptember 2014] 601
Setting x
i
= p
3/2
i
and y
i
= p
1/2
i
, we obtain
i
p
2
i
i
p
3
i
i
p
i
i
p
3
i
as desired.
As a consequence, the probability of obtaining a triple in three throws of a die is
never less than the probability of obtaining two straight doubles in four throws of that
die. It should now be clear that equality must not be taken for granted. We look at some
special cases.
From now on, let A, B, C, and D be specically the results of four consecutive
throws of an n-sided die; that is, iid random variables dened by selections on the
sample space = {1, . . . , n}. We denote by p
i
the probability that i is the result
obtained by throwing that die, for i .
Perfectly fair dice. Consider the case where p
i
= 1/n for all i ; that is, our die is
perfectly fair. In this situation, equations (1) and (2) yield
Pr{E
3
} = Pr{E
2,2
} =
1
n
2
,
and therefore both players have the same probability of winning the game. As a matter
of fact, it is a simple exercise to show that Claim 1 holds with equality if, and only
if, either the random variables are uniformly distributed over a subset of their sample
space, or one of the possible outcomes occur with probability 1.
Perfectly loaded dice. Suppose the die was manufactured in such a way that the same
side always ends upwards after a throw, e.g. p
1
= 1, and p
i
= 0 for i {2, . . . , n}. Of
course, both strategies will score a point at each and every sequence of throws, and
the game will almost certainly end in a draw. As in the perfectly fair case, there is no
preferable strategy.
Coins. We nowfocus on the case n = 2 and call our die a coin. We are interested in the
function d : [0, 1] [0, 1] that quanties the advantage d( p) = Pr{E
3
} Pr{E
2,2
} of
playing for a triple when the probability of our coin landing heads is p. By making
p
1
= p and p
2
= 1 p in (1) and (2), we obtain
d( p) =
p
3
+(1 p)
3
p
2
+(1 p)
2
2
= 4p
4
+8p
3
5p
2
+ p.
It is now easy to see (Figure 1) that d has minimum value 0 at p {0, 0.5, 1} as ex-
pected, and maximum value 0.0625 at p = 1/2 1/(2
2) 0.5 0.3536. In other

words, 6.25% is the largest possible probabilistic advantage that can be obtained in
this game.
1
This is achieved by the player who plays for a triple when the probability
of the coin landing heads (or tails, by symmetry) is approximately 85.36%.
1
Should a winning margin of 6.25% appear to be rather small, compare it with the casinos winning margin
of 5.26% in American roulette, and of only 2.70% in European roulette. Yet we do not see casinos going
bankrupt too often!
Figure 1. Advantage of playing for a triple as a function of p
3. SEEMINGLY IRRELEVANT KNOWLEDGE. Let us look at a different exper-
iment. Someone throws a die with n 2 sides three times in a row, and writes down
the sequence of results, calling them A, B, and C. You want to guess whether C = B.
Does it make any difference if you are told that B = A?
Because the throws of the die are mutually independent, it is tempting to say that
the fact that the rst two results (A and B) do not match has no correlation whatsoever
to whether the third result (C) matches the second one. However, this reasoning turns
out to be deceiving. The variables are independent, but are the events they dene
necessarily so? What if the die is not perfectly fair?
Here again we start with an intuitive discussion. Suppose the die is so biased that
one of its sides (say, 1) lands upwards in 90% of the throws. In this case, the event
B = A reveals that something unexpected happened, that is, A and B are not both 1.
Since it is highly probable that C is 1, we cannot say that our assessment of the prob-
ability associated to the event C = B was unaffected by the knowledge that B = A.
More formally, let be the range of possible values of three iid random variables
A, B, and C. For i , let p
i
be the probability Pr{A = i } = Pr{B = i } = Pr{C = i }
that a given variable takes value i .
Clearly,
Pr{C = B} =
i
p
2
i
(3)
and
Pr{C = B | B = A} =
Pr{C = B = A}
Pr{B = A}
=
i
[ p
2
i
(1 p
i
)]
i
[ p
i
(1 p
i
)]
. (4)
The following result relates the two probabilities above.
Claim 2 (Fonseca et al. [5], Lemma 6). Given three iid discrete random variables
A, B, and C, we have Pr{C = B | B = A} Pr{C = B}.
For an example where equality does not hold, let = {1, 2, 3}, p
1
= 0.8, p
2
=
p
3
= 0.1. In this case, Pr{C = B} = 0.66, whereas Pr{C = B | B = A} 0.429.
In the realm of dice, the probability that the third result matches the second one
given that the second result does not match the rst one in a sequence of three inde-
pendent throws of a die is less than or equal to the unconditional probability that the
third result matches the second one.
Perfectly fair dice. By replacing p
i
with 1/n in equations (3) and (4), it is straight-
forward to verify that both probabilities are equal to 1/n. In other words, if the die is
fair, then the probability of having a match between the third and the second results is
not at all affected by the fact that the second throw and the rst one do not match. The
events are in this case independent.
Perfectly loaded dice. If the die always lands with the same side upwards, then it
is impossible that two throws have different outcomes. Our comparison here is then
meaningless.
Coins. When n = 2, the unconditional probability that two tosses of the coin yield the
same result is clearly
Pr{C = B} = p
2
1
+ p
2
2
= p
2
+(1 p)
2
= 2p
2
2p +1, (5)
where p denotes, without loss of generality, the probability of heads. Such a probabil-
ity has minimum value 0.5 at p = 0.5 (i.e., a fair coin).
On the other hand, the conditional probability we are interested in is obtained from
equation (4) and translates to
Pr{C = B | B = A} =
p
2
(1 p) +(1 p)
2
p
2p 2p
2
=
p p
2
2p 2p
2
= 0.5,
regardless of p (of course, provided p is not 0 or 1).
A simpler way of seeing this is to note that, given that the rst two coins came up
differently, the second coin is uniformly distributed over heads and tails. Thus, the
probability that the third coin results the same as the second is 0.5p +0.5(1 p) =
0.5.
This interesting result, which motivates the whole next section, is summarized in
the corollary below. We recall that a Bernoulli random variable is such that it takes
value 1 with success probability p and value 0 with failure probability 1 p.
Corollary 3. Given three independent Bernoulli random variables A, B, and C with
success probability 0 < p < 1, we have Pr{C = B | B = A} = 0.5 regardless of p.
4. HAND CLAPS AND WHISTLES. The idea of playing a fair heads or tails with
a biased coin
2
is attributed to von Neumann [8] as follows. The coin is tossed twice
in a row. The rst player wins if the outcome is a heads-tails sequence, whereas the
second player wins with a tails-heads sequence. If two identical results are obtained,
another turn of two coin tosses starts from scratch. The probability of winning the
game at a certain turn is identical for both players, namely p(1 p), where p is the
2
In spite of the numerous references to such entities over the centuries, some recent evidence seems to show
that there is no such thing as a biased coin that is caught by the hand (i.e., the coin is allowed to spin in the air,
but not to bounce). See [6] for nding out why the biased coin may be considered the unicorn of probability
theory.
probability that a coin toss results heads. Thus, the probability that some player wins at
a certain turn is q = 2p(1 p), and the number of turns until the game has a winner is
a geometric random variable X whose expectation is E{X} = 1/q = 1/(2( p p
2
)).
Since each turn comprises exactly two coin tosses, the expected number of tosses until
a player wins using von Neumanns method is 2E{X} = 1/( p p
2
).
Nowsuppose we have a situation where the coin is concealed, that is, the contestants
are not able to see the outcome of each toss. Instead, they can only gure out, after each
toss, whether the result happened to match the one that just preceded it, which we refer
to as the base result. For the sake of illustration, consider that the coin is tossed by a
trusted third party who claps hands each time the coin toss results the same as in the
previous toss, and who whistles otherwise. Only after the very rst toss (the initial
toss, in each game), no sound is produced, since there is no base result to compare
with. We want to assure fairness, and we certainly cannot use von Neumanns original
method, since the actual coin results are not known to the players.
We rst discuss wrong ways of trying to obtain a fair game.
Unsuccessful attempts at fairness. In the proposed setting, the players cannot see the
actual results, but they hear something after each toss
3
: a hand clap (Cl) or a whistle
(Wh). Then why not simply regard a hand clap as heads, a whistle as tails, and play
the good old heads or tails? In other words, after the initial toss, toss the coin exactly
once more. Player 1 wins with a hand clap, Player 2 wins with a whistle. Of course
we can play such a game, but should not expect it to be fair if the coin itself is not
fair. Indeed, the greater the coin bias, the greater the advantage of playing for a hand
clap. Formally, if p is the probability that the coin lands heads, then the probability
that Player 1 wins is equal to the probability that two independent tosses of the coin
yield the same result, namely p
2
+(1 p)
2
= 2p
2
2p +1, whose minimum value
0.5 is obtained at p = 0.5, as in (5).
Aiming at neutralizing the coin bias, a second, natural attempt is to use a straight-
forward translation of von Neumanns idea based on the perceived sounds. After the
initial toss, which sets the rst base value, the coin is tossed twice in a row. Player
1 wins with a Cl-Wh sequence, Player 2 wins with a Wh-Cl sequence. If any other
sequence occurs, the game proceeds with another turn of two coin tosses. Is the game
now fair, regardless of the coin bias? We show that it is not, except again if p = 0.5.
To calculate the probability P
1
that Player 1 wins the game, let F
H
and F
T
denote
the events where the initial toss results heads (H) or tails (T), respectively, so that
P
1
= Pr{P
1
| F
H
} Pr{F
H
} +Pr{P
1
| F
T
} Pr{F
T
}
by the law of total probability. By letting P
H
1
= Pr{P
1
| F
H
} and P
T
1
= Pr{P
1
| F
T
},
we can write
P
1
= P
H
1
p + P
T
1
(1 p). (6)
We now obtain P
H
1
conditioned on the results of the two coin tosses that followed
the initial toss (whose result was heads, by denition):
[base H] H-Hthis sequence yields two hand claps, no player wins yet, and the
probability that Player 1 eventually wins is still P
H
1
due to the memorylessness of
the process, which now starts over with a new turn of two tosses still having heads
as base value;
3
Except for the initial toss, as mentioned above.
[base H] H-Tthis gives a Cl-Wh sequence, and victory is awarded to Player 1;
[base H] T-Hwe have two whistles and no winner, hence the probability of Player
1 winning is still P
H
1
, as in the case H-H;
[base H] T-Ta whistle followed by a hand clap: victory goes straight to Player 2.
Using again the law of total probability, we can write
P
H
1
=
SS
Pr{P
H
1
| S} Pr{S}
= P
H
1
p
2
+1 p(1 p) + P
H
1
(1 p) p +0 (1 p)
2
,
where S = {H-H, H-T, T-H, T-T} is the set of events associated to the two coin tosses
that follow the initial toss, as mentioned above. After some easy manipulations, we
obtain
P
H
1
= p. (7)
An analogous argument shows that
P
T
1
= 1 p. (8)
By plugging (7) and (8) into (6), we obtain P
1
= p
2
+(1 p)
2
= 2p
2
2p +1.
This is again the same expression as in (5), which has minimum value 0.5 at p = 0.5.
In short, the game is only fair when the coin itself is fair.
Before proceeding, it is interesting to notice that hereunlike the coin game pro-
posed in Section 2, where a probability of heads around 0.5 0.3536 would maxi-
mize a players winning oddsthe greater the coin bias, the greater the probability
that Cl-Wh wins over Wh-Cl, in spite of the apparent symmetry of both strategies.
We could try this game against an inadvertent opponent (perhaps someone who er-
roneously regards it as being equivalent to von Neumanns method). However, us-
ing a similar reasoning as in the calculation of P
1
, we see that the expected number
of tosses before the game nishes (disregarding the initial toss) is 1/( p p
2
), as in
von Neumanns method. This fact cannot be disregarded. For example, when the coin
is so biased that p = 0.999, though the Cl-Wh player has a winning probability of
98.02%, the coin needs to be tossed 1001 times on average before the game has a
winner!
In the next section, we look into a variation that successfully shields the players
from whatever bias the concealed coin might have (provided both sides appear with
non-zero probability).
Fair game with a concealed biased coin. We want to devise a coin game with the
following properties:
1. the coin is possibly biased, yet the exact bias is unknown to the players;
2. both players have equal chances of winning;
3. the actual coin results are not revealed, but it is possible to infer whether any two
consecutive coin tosses yielded the same result.
The two rst conditions are the usual ones. Ideally, we would like to cope with the
third, new condition without incurring any increase in the expected duration of the
game.
First, let us consider the natural solution of tossing the coin four times, considering
the sounds X and Y produced after the second and fourth tosses, respectively. Player
1 wins with a Cl-Wh sequence, Player 2 wins with Wh-Cl, and the other possible
sequences are discarded, triggering four fresh tosses of the coin. By doing so, X and
Y are clearly independent, and this approach corresponds exactly to von Neumanns
idea, therefore assuring a fair game. The drawback is that it may take twice as long.
Now consider the following variation. The main action consists of tossing the coin
two consecutive times, whose results we call respectively A and B, constituting a turn.
If a whistle is heard after the second toss (meaning B = A), then a third coin toss
whose result we call Cwill decide the game. Player 1 wins if C = B (indicated by
the subsequent whistle), Player 2 wins if C = B (indicated by the subsequent hand
clap). On the other hand, if the sound alert after the second coin toss is a hand clap
(meaning B = A), then the game continues with a fresh turn of two coin tosses that will
set the values of A and B from scratch, conguring a perfectly memoryless process.
Note that an initial toss that would set a base value prior to the rst turn is not even
necessary here, since the players only care about sound alerts that come with even
parity (after the second, fourth, sixth tosses, etc.) until a whistle with even parity is
heard for the rst time. When it eventually happens, a single, decisive extra toss closes
the game.
Theorem 4. The proposed coin game, where Player 1 wins with a whistle of even
parity immediately followed by another whistle, and Player 2 wins with a whistle of
even parity immediately followed by a hand clap, is a perfectly fair game, no matter
the probability of heads 0 < p < 1 of the employed coin.
Proof. Let A and B be the results of the two coin tosses that preceded the rst whistle
with even parity. That very whistle indicates B = A. The next coin toss, whose result
we call C, will decide the game in Player 1s favor if C = B, and in Player 2s favor
if C = B. Using a simple (tails 0, heads 1) mapping, those three results are
clearly independent Bernoulli random variables with success probability p. Thus, by
Corollary 3, both players have identical probabilities
Pr{C = B | B = A} = Pr{C = B | B = A} = 0.5
of winning the game.
As for the expected number of tosses, notice that the event that a whistle is heard
after the second toss of a turn corresponds to the event that the two consecutive tosses
of that turn produces either a H-T or a T-H sequence; hence, it occurs with probability
2p(1 p). The number of turns before the game nishes is therefore a geometric ran-
dom variable X, whose expectation is E{X} = 1/(2p(1 p)), plus one (correspond-
ing to the nal, decisive toss). Thus, the expected number of tosses in the game is
2 E{X} +1 =
1
p p
2
+1, (9)
which is only one coin toss greater than in the original von Neumanns method.
It is interesting to observe that, if in an attempt to reduce its expected duration, we
decided the game right after the rst whistle regardless of its parity, then two unfortu-
nate consequences would ensue:
(i) the intended improvement would not decrease the expected length of the
game at all; and, most importantly,
(ii) we would not be able to assure the fairness of the game anymore!
To show (i), let F
H
(and, respectively, F
T
) denote the event that the rst coin toss
in the game yields heads (respectively, tails), and let Y be the random variable corre-
sponding to the number of tosses before the rst whistle is heard. Using conditional ex-
pectations, we can write the overall expected number of tosses in the modied game as
1 +E{Y} = 1 +E{Y | F
H
} Pr{F
H
} +E{Y | F
T
} Pr{F
T
}
= 1 +
1 +
1
1 p
p +
1 +
1
p
(1 p)
= 1 +
1
p p
2
,
which is exactly the same as in (9).
To show (ii), we argue that Player 1who wins the game if the sound after the rst
whistle is another whistlewill win the game exactly if:
the rst coin toss yields H, and the toss immediately after the rst occurrence of
T (which produces the rst whistle of the game) yields H (producing the second
whistle in a row); or, analogously,
the rst coin toss yields T, and the toss immediately after the rst occurrence of H
yields T.
Thus, the probability that Player 1 wins the game is p
2
+(1 p)
2
= 2p
2
2p +1,
the same function of p seen in (5)which has minimum value 0.5 exactly at p =
0.5yet again.
ACKNOWLEDGEMENTS. The triple or two straight doubles question was proposed to us by Guilherme
Dias da Fonseca when he was studying randomized data structures. We are also grateful to Alexandre Stauffer,
to Luiz Henrique de Figueiredo, and to the anonymous referees for the numerous valuable suggestions.
REFERENCES
1. M. Blum, Independent unbiased coin ips from a correlated biased source: a nite state Markov chain,
Combinatorica 6 (1986) 97108.
2. A. Cauchy, Oeuvres 2, III (1821).
3. B. Chor, O. Goldreich, Unbiased bits from sources of weak randomness and probabilistic communication
complexity, in Proc. 26th IEEE Symp. on Foundations of Computer Science (1985) 429442.
4. P. Elias, The efcient construction of an unbiased random sequence, The Annals of Mathematical Statis-
tics 43 (1972) 865870.
5. G. D. da Fonseca, C. M. H. de Figueiredo, P. C. P. Carvalho, Kinetic hanger, Information Processing
Letters 89 (2004) 151157.
6. A. Gelman, D. Nolan, You can load a die, but you cant bias a coin, The American Statistician 56 (2002)
308311.
7. W. Hoeffding, G. Simons, Unbiased coin tossing with a biased coin, The Annals of Mathematical Statis-
tics 41 (1970) 341352.
8. J. von Neumann, Various techniques used in connection with random digits, J. Res. National Bureau of
Standards 12 (1951) 3638.
9. Y. Peres, Iterating von Neumanns procedure for extracting random bits, The Annals of Statistics 20
(1992) 590597.
10. P. A. Samuelson, Constructing an unbiased random sequence, Journal of the American Statistical Asso-
ciation 63 (1968) 15261527.
11. A. Srinivasan, D. Zuckerman, Computing with very weak random sources, SIAM Journal on Computing
28 (1999) 14331459.
12. Q. F. Stout, B. Warren, Tree algorithms for unbiased coin tossing with a biased coin, The Annals of
Probability 12 (1984) 212222.
13. S. Vembu, S. Verd u, Generating random bits from an arbitrary source: fundamental limits, IEEE Trans-
actions on Information Theory 41 (1995) 13221332.
14. D. Zuckerman, Simulating BPP using a general weak random source, in Proc. 32nd IEEE Symp. on
Foundations of Computer Science (1991) 7989.
VIN
ICIUS GUSM
AO PEREIRA DE S
A received his D.Sc. degree from Universidade Federal do Rio de

Janeiro, where he is currently an associate professor. When he is neither working with combinatorics or pro-
gramming computers, he can be seen bicycling or trying to play the piano.
Departamento de Ci encia da Computac ao, UFRJ, Brazil
vigusmao@dcc.ufrj.br
CELINA MIRAGLIA HERRERA DE FIGUEIREDO is a full professor at the Systems Engineering and
Computer Science Program of COPPE, Universidade Federal do Rio de Janeiro, where she has been a collab-
orator since 1991.
COPPE, UFRJ, Brazil
celina@cos.ufrj.br
A Drug-Induced Random Walk by Daniel Velleman (Amherst College) was
inspired by his cat, Natasha.
Abstract. The label on a bottle of pills says Take one half pill daily. A natural way
to proceed is as follows: Every day, remove a pill from the bottle at random. If it is a
whole pill, break it in half, take one half, and return the other half to the bottle; if it is a
half pill, take it. We analyze the history of such a pill bottle.
Here, Natasha is helping Dr. Velleman proofread his article that was featured in the April 2014
issue of the MONTHLY.
Zero Sums on Unit Square Vertex Sets
and Plane Colorings
Richard Katz, Mike Krebs, and Anthony Shaheen
Abstract. We prove that if a real-valued function of the plane sums to zero on the four vertices
of every unit square, then it must be the zero function. This fact implies a lower bound in a
coloring of the plane problem similar to the famous HadwigerNelson problem, which asks
for the smallest number of colors needed to assign every point in the plane a color so that no
two points of unit distance apart have the same color.
1. INTRODUCTION. The rst problem from the 2009 Putnam exam asks, Let f
be a real-valued function on the plane such that for every square ABCD in the plane,
f (A) + f (B) + f (C) + f (D) = 0. Does it follow that f (P) = 0 for all points P in
the plane? Spoiler alert: The answer is yes, and here is a proof. Let P be an arbitrary
point in the plane, and arrange eight other points A, B, C, D, E, F, G, H in the plane
as in Figure 1, so that ACHF is a square with center P and B, E, G, and D are the
midpoints of AC, CH, HF, and FA, respectively.
A C
D
P
E
F
B
G H
Figure 1. Nine points dening six squares
Each square produces an equation; the square ABPD, for example, tells us that
f (A) + f (B) + f (P) + f (D) = 0. Adding the four equations from the four small
squares, then using the fact that ACHF and BEGD are also squares, shows that
4 f (P) = 0 and so f (P) = 0.
Note that the various squares involved do not all have the same size. ABPD, for
example, is smaller than BEGD, which in turn is smaller than ACHF. Is that a necessary
feature of any solution, or can we nd a proof that makes use of squares all with
the same side lengths? In other words, suppose we are given only that the condition
f (A) + f (B) + f (C) + f (D) = 0 holds whenever ABCD is a unit square, that is, a
square with side length 1. Does it still follow that f must be the zero function?
MSC: Primary 00A05
The answer, it turns out, is again yes. Not surprisingly, arriving at this result is more
difcult than solving the Putnam problem, but our proof nevertheless uses only high
school algebra and geometry. In Section 2, we present our solution, which takes a
pleasant and meandering journey before reaching its nal destination. In Section 3, we
provide a brief history of the long-standing HadwigerNelson problem (which asks
for the smallest number of colors needed to color the plane so that no two points of
distance 1 from each other have the same color), and then show how our main theorem
sheds some light on this question and other closely related questions.
2. PROOF OF THE MAIN RESULT. In this section, we prove the following
theorem.
Theorem 2.1. Let f : R
2
R such that f (A) + f (B) + f (C) + f (D) = 0 holds
whenever ABCD is a unit square. Then f (P) = 0 for all P R
2
.
Henceforth, if a function satises the hypothesis of Theorem 2.1, we will say it
satises the unit square condition.
Figure 1 offers us a clue as to how to prove Theorem 2.1. Each point corresponds to
a variable, and each square corresponds to a linear equation those variables must sat-
isfy. The key is the rotated square BEGD, which produces an extra equation without
introducing any additional variables. The main idea of our proof, then, is to rotate a
collection of unit squares so that the resulting gure contains, on average, many unit
squares per vertex. Applying this approach and then following our noses leads to a
lemma (Lemma 2.2) that takes us from the unit square condition (a statement about
four noncollinear points) to a more useful statement about six collinear points.
Lemma 2.2. Suppose that f satises the unit square condition. Let r be any point in
the plane, and let u be any unit vector. Then
f (r +8u) + f (r +7u) + f (r +5u) = f (r +3u) + f (r +u) + f (r).
Proof. We can arrange a coordinate system so that r = (0, 0) and u = (1, 0). We begin
by observing that the value of f on integer lattice points is determined by its values
on the axes. More precisely, let X
a,b
= f (a, b), P
a
= X
a,0
, Q
b
= X
0,b
, and = X
0,0
.
It follows by repeatedly applying the unit square condition (i.e., using four double
inductions on a and b, one for each quadrant) that for any integers a, b, we have
X
a,b
= (1)
b
P
a
+(1)
a
Q
b
+(1)
a+b+1
. (1)
We remark that the integer lattice alone is not sufcient to force f to be the zero
function; a checkerboard of 1s and 1s shows otherwise. With that in mind, we
now rotate the integer lattice so that the new lattice has many points in common with
the old lattice. To do so, we take advantage of a 3-4-5 right triangle. Let
Y
a,b
= f
3
5
a +
4
5
b,
4
5
a +
3
5
b
and U
a
= Y
a,0
, V
b
= Y
0,b
.
Figure 2 shows some of the points evaluated to nd the values X
a,b
and Y
a,b
for inte-
gers a, b. Because (3a/5 +4b/5, 4a/5 +3b/5) is obtained by applying a rotation
matrix to (a, b), and because rotations map unit squares to unit squares, the values
AugustSeptember 2014] ZERO SUMS 611
Figure 2. The X
a,b
(resp. Y
a,b
) give the values of f at the solid (resp. hollow) dots. Note the 3-4-5 right
triangle that denes the angle of rotation.
Y
a,b
will satisfy the same relations as the values X
a,b
. So the following analogue of
equation (1) holds for any integers a, b:
Y
a,b
= (1)
b
U
a
+(1)
a
V
b
+(1)
a+b+1
. (2)
The intersection points of the two lattices give us relations between the values
P
a
, Q
b
, U
a
, V
b
. Specically, we have
X
4 j 5m,3 j 5m
= Y
m,5 j 7m
(3)
for any integers j, m. Combining equations (1), (2), and (3) and then simplifying, we
get
P
4 j 5m
+(1)
j
Q
3 j 5m
= U
m
+(1)
j
V
5 j 7m
(4)
for any integers m, j .
Our next goal is to eliminate variables one at a time from equation (4) until we have
only Ps. First we wipe out the Us. To do so, take j = 0 to get that
P
5m
+ Q
5m
V
7m
= U
m
. (5)
Substitute (5) back into (4) and collect variables of the same type to get that
P
4 j 5m
P
5m
+(1)
j
Q
3 j 5m
Q
5m
= V
7m
+(1)
j
V
5 j 7m
. (6)
Next, we take aim at the Vs. Substitute 7r for j in equation (6) to get
P
28r5m
P
5m
+(1)
r
Q
21r5m
Q
5m
= V
7m
+(1)
r
V
35r7m
(7)
for any integers r, m. Substitute j + 7r for j and m + 5r for m in equation (6) to
get
P
4 j 5m+3r
P
5m25r
+(1)
j +r
Q
3 j 5m4r
Q
5m25r
= V
7m35r
+(1)
j +r
V
5 j 7m
. (8)
Observe that the right-hand sides of the three equations (6), (7), and (8) con-
tain only three V variables, namely V
7m
, V
5 j 7m
, and V
7m35r
. (As you may have
suspected, that was the reason for all those substitutions.) So with the appropriate
manipulationsnamely, subtract (7) from (6), multiply the result by (1)
r
, and set it
equal to (8)we nd that
(1)
r
P
4 j 5m
+(1)
r+1
P
28r5m
P
4 j 5m+3r
+ P
5m25r
= (1)
j +r+1
Q
3 j 5m
+ Q
21r5m
+(1)
j +r
Q
3 j 5m4r
Q
5m25r
. (9)
Finally, we lay waste to the Qs. Equation (9) is a relation between the values
of f on certain points on the coordinate axes. Because this equation was derived
using only the unit square condition, we can shift all of those points to the right
one unit and obtain the following new relation for the values of f at the shifted
points:
(1)
r
P
4 j 5m+1
+(1)
r+1
P
28r5m+1
P
4 j 5m+3r+1
+ P
5m25r+1
= (1)
j +r+1
X
1,3 j 5m
+ X
1,21r5m
+(1)
j +r
X
1,3 j 5m4r
X
1,5m25r
. (10)
Using equation (1), this equation becomes
(1)
r
P
4 j 5m+1
+(1)
r+1
P
28r5m+1
P
4 j 5m+3r+1
+ P
5m25r+1
= (1)
j +r
Q
3 j 5m
Q
21r5m
+(1)
j +r+1
Q
3 j 5m4r
+ Q
5m25r
. (11)
Observe that the right-hand side of equation (11) is precisely the negative of the
right-hand side of equation (9). For the values r = 1, m = 4, j = 6, equations (9)
and (11) both hold, giving us
P
4
+ P
8
P
1
+ P
5
= P
5
P
9
+ P
2
P
6
. (12)
Equation (12) tells us about a relationship between values of f on certain points along
the horizontal coordinate axis. Employing our favorite trick, we can slide this relation-
ship back and forth at will, because it ultimately derives from the unit square condition,
which is invariant under rigid motions. So, by shifting to the left 1 unit, we nd that
P
8
+ P
7
+ P
5
= P
3
+ P
1
+ P
0
. (13)
Unwinding our variable denitions, we see that (13) is precisely the equation we set
out to prove.
With Lemma 2.2 rmly in hand, we now prove Theorem 2.1. Let
u
1
=
1
24
,
575
24
; u
2
=
1
24
,
575
24
;
u
3
=
575
24
,
1
24
; u
4
=
575
24
,
1
24
.
Observe that u
1
, u
2
, u
3
, and u
4
all have magnitude 1, and that 12u
1
+ 12u
2
= e
1
=
(1, 0) and 12u
3
+12u
4
= e
2
= (0, 1).
Let f be a function satisfying the unit square condition, and let n {0, 1, 2, 3, 4}.
We will prove the following claim: If f is invariant under translation by 12u
j
for all
integers j with 1 j n, then f is the zero function. Note that the case n = 0 is
precisely our theorem, so it sufces to prove this claim.
The proof will be by reverse induction. For the base case (n = 4), we have that f
is invariant under translation by 12u
1
, 12u
2
, 12u
3
, and 12u
4
. Therefore, f is invariant
under translation by e
1
and e
2
. Hence for any x, we have 4 f (x) = f (x) + f (x +e
1
) +
f (x +e
2
) + f (x +e
1
+e
2
) = 0. Therefore, f is the zero function.
Now we prove the claim for n = k 1, assuming it has been proven for n = k,
where k {1, 2, 3, 4}. Dene two functions g and h from R
2
to R by g(x) = f (x +
u
k
) + f (x) and h(x) = g(x +3u
k
) + g(x). Then g and h each satisfy the unit square
condition, and each are invariant under translation by 12u
j
for all integers j with
1 j k 1. Moreover, note that by Lemma 2.2, we have
h(x +4u
k
) = f (x +8u
k
) + f (x +7u
k
) + f (x +5u
k
) + f (x +4u
k
)
= f (x +4u
k
) + f (x +3u
k
) + f (x +u
k
) + f (x)
= h(x).
So h is invariant under translation by 4u
k
and hence also by 12u
k
. By the inductive
hypothesis, h is the zero function. It follows that g(x +3u
k
) = g(x) for all x. So g
also is invariant under translation by 12u
k
and therefore is the zero function. From this
we see that f (x +u
k
) = f (x) for all x, which by similar reasoning implies that f is
the zero function, as desired.
3. COLORINGS OF THE PLANE
`
A LA HADWIGERNELSON. The classic
HadwigerNelson problem asks the following. Suppose we wish to assign each point
in the plane a color so that no two points of distance 1 from each other have the same
color. What is the smallest number of colors needed? It has been known since 1961
that at least 4 colors are needed, as shown by the Moser spindle [3], which is depicted
in Figure 3. Moreover, a book [2] published in 1964 uses a hexagonal tiling as in
Figure 4 to meet the requirement using only 7 colors. So the smallest number of colors
needed is at least 4 and is no more than 7. For the past half-century, no improvement
has been made on these bounds. We remark that there is not universal agreement as
to who originally posed this problem or what it should be called. A more extensive
discussion of the problem (including a history of the tangled web of attributions) can
be found in [4].
The question can be formulated easily in the language of graph theory. Namely, let
G be the graph whose vertex set is R
2
so that two vertices are adjacent if and only if
they are distance 1 from each other. HadwigerNelson asks for the chromatic number
of G. As a corollary to Theorem 2.1, we obtain a lower bound on the chromatic number
of a similar graph.
Corollary 3.1. Suppose we wish to assign each point in the plane a color so that any
two points of distance 1 or
2 from each other have different colors. Then at least 5

colors are needed.
Figure 3. The Moser spindlea subset of the plane requiring at least four colors where each line segment
shown has length 1.
5 3 1 6
6 4 2 7
7 5 3 1
1 6 4 2
2 7 5 3
3 1 6 4
1 6 4
2 7 5
3 1 6
4 2 7
5 3 1
Figure 4. A portion of a paint-by-numbers 7-coloring of the plane using hexagons of diameter just less
than 1.
In other words, dene a graph whose vertex set is R
2
, where two vertices are adja-
cent if and only if the distance between them is either 1 or
2. We will show that the

chromatic number of this graph is greater than or equal to 5.
Proof. Temporarily assume that such a coloring is possible with only 4 colors, say,
orange, yellow, blue, and hot pink. Observe that all four colors are needed for the
four vertices of any unit square. Dene a function f : R
2
R by f (x) = 3 if x is
orange and f (x) = 1 otherwise. Then f (A) + f (B) + f (C) + f (D) = 0 whenever
ABCD is a unit square. So by Theorem 2.1, f is the zero function, which contradicts
the denition of f .
Remark 3.2. Figure 4 is not a proper 7-coloring of the plane with respect to the dis-
tance set {1,
2}. By carefully analyzing Figure 4, however, we can make the appro-

priate adjustments. The 7-coloring in Figure 4 is obtained as follows. Working modulo
7, for any hexagon with value n, assign the value n +1 to the adjacent hexagon im-
mediately below it, and assign the value n + 2 to the adjacent hexagon immediately
up and to the right of it. (On the boundary, assign the value of any adjoining hexagon.)
The result is that any hexagon H with value n has a bubble, two hexagons deep,
surrounding it, so that none of the hexagons in the bubble have value n. Moreover, two
hexagons deep is just deep enough so that given any point p in H, the circle of radius 1
centered at p lies entirely within that bubble, as illustrated in Figure 4. A line segment
of length
2 with one end at p, however, may extend outside the bubble. To preclude
this situation, we must nd new values that create a bubble three hexagons deep. To
do so, we work modulo 13. For any hexagon with value n, assign the value n +1 to
the adjacent hexagon immediately below it, and assign the value n +3 to the adjacent
hexagon immediately up and to the right of it. This time, the circles of radius 1 and
2
both lie in the bubble, as depicted in Figure 5. The result, then, is a 13-coloring of the
plane with the property that no two points of distance 1 or
2 from each other have

the same color. So the number of colors needed is at least 5 and no more than 13.
13 7 1 8
1 8 2 9
2 9 3 10
3 10 4 11
4 11 5 12
5 12 6 13
6 13 7 1
7 1 8 2
10 4 11 5 12
11 5 12 6 13
12 6 13 7 1
13 7 1 8 2
1 8 2 9 3
2 9 3 10 4
3 10 4 11 5
Figure 5. A portion of a 13-coloring of the plane using hexagons of diameter just less than 1.
Note that in Corollary 3.1, we have shown that at least 5 colors are needed without
nding any specic subset of the plane that requires at least 5 colors. Indeed, a similar
argument produces the lower bound of 4 for the original HadwigerNelson problem
without the need for the Moser spindle or any such gure. First, we establish that a
function f : R
2
R that sums to zero on the vertices of any equilateral triangle of
side length 1 must equal zero everywhere. Recursively draw circles of radius
3, rst
centered at the origin, then centered at all points previously drawn. These circles ll
up the plane, thereby showing that f must be the constant function, hence the zero
function. The lower bound of 4 now follows immediately from a proof just like that of
Corollary 3.1. (Indeed, using higher-dimensional analogues of regular tetrahedra with
side length 1, we can likewise show that at least n +2 colors are needed for R
n
.)
We move now into speculative territory. The preceding discussion suggests a pos-
sible two-step strategy for improving the current best lower bound in the Hadwiger
Nelson problem. The rst step would be to nd a set S of 4k points in the plane such
that any 4-coloring of S that forbids the distance 1 must use each color exactly k times.
In other words, in graph-theoretic language, nd a nite unit distance graph with chro-
matic number 4 so that any proper 4-coloring uses all four colors equitably. Assuming
4 colors sufce, dene a real-valued function f of the plane with value 3 on points
of the rst color and 1 everywhere else. Observe that f sums to zero on any set
congruent to S. The second step would be to argue that any such function must vanish
everywhere.
As for the second step, it is not unreasonable to hope that Theorem 2.1 can be ex-
tended to sets other than unit squares. In a recent MONTHLY article [1], de Groote
and Duerinckx prove that if E is any nite subset of R
2
, and f is a real-valued func-
tion of the plane that vanishes on all sets directly similar to E (that is, the image of
E under the composition of a translation, a dilation, and a rotation), then f must be
the zero function. In fact, consider the group AGL(2, R) of all afne transformations
of the plane. (An afne transformation is a translation composed with a linear trans-
formation.) Suppose that G is a subgroup of AGL(2, R) such that point stabilizers are
transitive and abelian. They show more generally that for any nite subset E of the
plane, if f sums to zero on g(E) for all g G, then f is the zero function. We note
that Theorem 2.1 is not a special case of their result, because for us the relevant group
is the group of rigid motions of the plane, so our point stabilizers are not transitive.
It is natural to ask, then, whether the intersection of the extensions of their result and
ours holds. In other words, let E be a nite subset of R
2
, and let f : R
2
R. Suppose
that whenever E
is an image of E under a rigid motion of the plane (i.e., whenever E
is congruent to E), then the sum of f over the elements of E
is zero. Does it follow

that f vanishes everywhere?
ACKNOWLEDGMENT. We would like to offer our profuse thanks to the reviewers, especially the eagle-
eyed referee who spotted a serious but deadly error in our original proof of Theorem 2.1.
REFERENCES
1. C. de Groote, M. Duerinckx, Functions with constant mean on similar countable subsets of R
2
, Amer.
Math. Monthly 119 (2012) 603605.
2. H. Hadwiger, H. Debrunner, Combinatorial Geometry in the Plane. Translated by V. Klee. With a new
chapter and additional material supplied by the translator. Holt, Rinehart and Winston, New York, 1964.
3. L. Moser, W. Moser, Solution to problem 10, Canad. Math. Bull. 4 (1961) 187189.
4. A. Soifer, Chromatic number of the plane & its relatives, history, problems and results: an essay in 11
parts, in Ramsey Theory. Progr. Math. Vol. 285. Birkh auser Springer, New York, 2011, available at http:
//dx.doi.org/10.1007/978-0-8176-8092-3_8.
RICHARD KATZ received a B.E.E. from City College of New York and, after working in the aerospace
industry for four years, returned to school and obtained a Ph.D. in mathematics from UCLA. He recently
retired from Cal State University, Los Angeles. His main research interest was (and still is) function theory on
Riemann surfaces and he has had a lifelong interest in the history of mathematics.
Department of Mathematics, California State UniversityLos Angeles, 5151 State University Drive,
Los Angeles, CA 90032
rkatz@calstatela.edu
MIKE KREBS received a bachelors degree from Pomona College and a Ph.D. in mathematics from Johns
Hopkins University. He is currently an Associate Professor of Mathematics at California State University, Los
Angeles. On a recent sixteen-hundred mile road trip, he and his six-year-old son rode square-wheeled tricycles
at a math-themed museum exhibit.
mkrebs@calstatela.edu
TONY SHAHEEN is an Associate Professor of Mathematics at California State University, Los Angeles.
When Pomp and Circumstance begins to play, Professor Shaheen makes a magical transformation into Sum-
mer Tony.
ashahee@calstatela.edu
A Renement of a Theorem of J. E. Littlewood
In [1], J. E. Littlewood raises the question of how short a doctoral dissertation in mathematics could
in principle be,
1
and proposes the precise answer: Two sentences long. He proves this assertion
by pointing out that two sentences is a lower bound, since any thesis in mathematics must contain
at least the statement of one theorem and its proof, and then goes on to show that this lower bound
could in theory be attained (though it presumably never has been) by exhibiting a one-sentence proof,
using only results and notations that were familiar to the mathematicians of the period, of Picards
small theorem stating that an entire function avoiding the values 0 and 1 is constant. But Littlewood
overlooked the fact that one can shorten the thesis by one whole sentence by placing the statement of
the theorem in the title rather than the body of the text (cf. [2]). Moreover, his proof of the upper bound
is also too complicated, since there is a theorem that is better-known and simpler to prove than Picards,
yet undoubtedly also important enough to justify conferring the title of Doctor of Philosophy on its
author. We therefore propose the following corrected and rened version of Littlewoods theorem.
Theorem. The shortest possible doctoral thesis in mathematics is one sentence long.
Proof. One sentence is clearly a lower bound, because a thesis in mathematics must contain a proof. We
show that this bound can be attained by the following explicit construction of a theoretically possible
doctoral thesis.
Bounded entire functions are constant
by Joseph Liouville
submitted to the Department of Mathematics of the University of Paris, 1844,
in partial fulllment of the requirement of the
Degree of Doctor of Philosophy in Mathematics
If f (z) is bounded and entire, then Mr. Cauchys theorem implies that
f
(a) = lim
R
1
2i
|z|=R
f (a + z)
z
2
dz
= 0
for every a in the complex plane, so f (z) is constant.
REFERENCES
1. J. E. Littlewood, A Mathematicians Miscellany. Methuen & Co., London, 1953.
2. D. Zagier, A one-sentence proof that every prime p 1 (mod 4) is a sum of two squares, Amer.
Math. Monthly 97 (1990) 144.
1
Actually, Littlewood asks whether a dissertation of 2 lines could deserve and get a Fellowship,
but I have rephrased this in a less Cambridge-oriented way.
by D. Zagier, Max Planck Institute, Bonn, Germany
MSC: Primary 63B10
Positive Linear Maps and
Spreads of Matrices
Rajendra Bhatia and Rajesh Sharma
Abstract. The farther a normal matrix is from being a scalar, the more dispersed its eigenval-
ues should be. There are several inequalities in matrix analysis that render this principle more
precise. Here it is shown how positive unital linear maps can be used to derive many of these
inequalities.
An interesting theorem in linear algebra says that the eigenvalues of a normal matrix
are more spread out than its diagonal entries; i.e., if A =
_
a
i j
_
is an n n normal
matrix with eigenvalues
1
(A), . . . ,
n
(A), then
max
i, j
a
i i
a
j j
max
i, j
i
(A)
j
(A)
. (1)
It is customary to call the quantity on the right-hand side of (1) the spread of A, and
denote it by spd (A). Then the inequality (1) can be stated as
spd (diag (A)) spd (A). (2)
One proof of this goes as follows. Let x, y be the standard inner product on C
n
dened as x, y =
n
i =1
x
i
y
i
, and let x = x, x
1/2
be the associated norm. The set
W(A) = {x, Ax : x = 1} , (3)
is called the numerical range of the matrix A. If A is normal, then using the spectral
theorem, we can see that W(A) is the convex polygon spanned by the eigenvalues of
A. So spd (A) is equal to the diameter diam W(A). The diagonal entry a
i i
= e
i
, Ae
i
evidently is in W(A). So, we have the inequality (1).

The ToeplitzHausdorff Theorem is the statement that for every matrix A, the nu-
merical range W(A) is a convex set. It contains all the eigenvalues of A (in (3) choose
x to be an eigenvector of A). So, we always have diam W(A) spd (A). Chapter 1
of [4] contains a comprehensive discussion of the numerical range, and all these facts
can be found there.
In the special case when A is Hermitian, we can arrange its eigenvalues in decreas-
ing order as
1
(A)
n
(A). Then W(A) is the interval
_
n
(A),
1
(A)
_
, and the
inequality (1) says
max
i, j
a
i i
a
j j
1
(A)
n
(A). (4)
The inequality (2) is not always true for arbitrary matrices. For example, the 2 2
matrix
MSC: Primary 15A60, Secondary 15A42
AugustSeptember 2014] LINEAR MAPS AND MATRICES 619
_
1 1/4
1 0
_
has eigenvalue 1/2 with multiplicity 2. Here, spd (A) = 0, but spd (diag (A)) = 1.
It is not always easy to nd the eigenvalues of a matrix, and the importance of
relations like (1) lies in the information they give about eigenvalues in terms of matrix
entries. Many authors have found different lower bounds for spd (A) in which the left-
hand side of (1) is replaced by a larger quantity or by some other function of entries of
A. The aim of this note is to propose a method by which many of the known results,
and some new ones, can be obtained.
Let M(n) be the space of all n n complex matrices. A linear map from M(n)
to M(k) is said to be positive if (A) is positive semidenite whenever A is. It is said
to be unital if (I ) = I . In the special case when k = 1, such a is called a positive,
unital, linear functional, and it is customary to represent it by the lower case letter .
We refer the reader to [1] for properties of such maps.
The space M(n) is a Hilbert space with the inner product A, B = tr A
B. As a
consequence, every linear functional on M(n) has the form (A) = tr AX for some
matrix X. This functional is positive if and only if X is positive semidenite, and
unital if and only if tr X = 1. (Positive semidenite matrices with trace 1 are called
density matrices in the physics literature.) Let
1
, . . . ,
n
be the (necessarily real and
nonnegative) eigenvalues of X and let u
1
, . . . , u
n
be a corresponding orthonormal set
of eigenvectors. If T is any n n matrix, and u
1
, . . . , u
n
is an orthonormal basis of
C
n
, then tr T =
n
j =1
u
j
, Tu
j
. Hence,
(A) = tr AX =
n
j =1
u
j
, AXu
j
=
n
j =1
j
u
j
, Au
j
.
Since
j
= 1, this shows that (A) is a convex combination of the complex num-
bers u
j
, Au
j
, each of which is in W(A). So by the ToeplitzHausdorff Theorem,
(A) is also in W(A). There exists a unit vector y (depending on A) such that (A) =
y, Ay. Thus, the numerical range W(A) is also the collection of all complex numbers
(A) as varies over positive unital linear functionals. So if
1
and
2
are any two
such functionals, then
|
1
(A)
2
(A)| diam W(A). (5)
The following theorem, which is of independent interest, is an extension of this obser-
vation.
We use the notation A for the operator norm of A dened as
A = sup {Ax : x = 1} .
If s
1
(A) s
n
(A) are the decreasingly ordered singular values of A, then A =
s
1
(A). If A is normal, then A = max
j
j
(A)
. If A is Hermitian, then
A = max
x=1
|x, Ax| .
These facts about the norm are used in the following discussion. If A and B are
two Hermitian matrices, we say that A B if A B is positive semidenite.
Theorem 1. If
1
and
2
are any two positive unital linear maps from M(n) into
M(k), then
(i) for every Hermitian A in M(n),
1
(A)
2
(A) diamW(A), (6)
(ii) if n = 2, then the inequality (6) holds also for all normal matrices A.
Proof. If A is an n n Hermitian matrix, then
n
(A)I A
1
(A)I . The linear
maps
j
, for j = 1, 2, preserve order and take the identity I in M(n) to I in M(k). So
we have
n
(A)I
j
(A)
1
(A)I , for j = 1, 2. From this we obtain
1
(A)
2
(A)
_
1
(A)
n
(A)
_
I
and
2
(A)
1
(A)
_
1
(A)
n
(A)
_
I.
Now, if X is a Hermitian matrix and X I , then
j
(X)
for all j , and hence

X . So we have the inequality (6).
Now, suppose that n = 2. If A is a 2 2 normal matrix, then A = P +Q, where
, are the eigenvalues of A and P, Q are the corresponding eigenprojections. We
have P + Q = I , and hence
j
(P) +
j
(Q) = I . Hence,
1
(A)
2
(A) =
1
(P) +
1
(Q)
2
(P)
2
(Q)
= (I
1
(Q)) +
1
(Q) (I
2
(Q))
2
(Q)
= ( )(
2
(Q)
1
(Q)).
Taking norms of both sides, we get
1
(A)
2
(A) = | |
2
(Q)
1
(Q). (7)
Since 0 Q I , we have 0
j
(Q) I , and hence
j
(Q) 1 for j = 1, 2. If
X, Y are positive semidenite, then X Y max(X, Y). So, from equation
(7) it follows that
1
(A)
2
(A) | |. This proves part (ii) of the Theorem.
When n = 3, the inequality (6) is not valid for all normal matrices. For nonnormal
matrices, it need not hold even when n = 2. Let
1
be the map that takes a 3 3
matrix A to its top left 2 2 block, and let
2
be the map that takes A to its bottom
right 2 2 block. Then
1
,
2
are positive unital linear maps from M(3) into M(2).
Let
A =
_
_
0 1 0
0 0 1
1 0 0
_
_
.
Then A is normal, and its eigenvalues are the three cube roots of 1. So diam W(A) =
3, but
1
(A)
2
(A) =
_
0 2
0 0
_
= 2,
and the inequality (6) breaks down. Let
1
: M(2) M(2) be the map dened as
1
(A) =
_
_
1
2
i, j
a
i j
_
_
I,
and let
2
(A) = A. If X is a positive semidenite matrix of any order n and e the
all-ones n-vector, then
i, j
x
i j
= e, Xe 0.
So
1
dened above is a positive unital linear map. Choose
A =
_
0 1
0 0
_
.
A little calculation shows that W(A) is the disk of radius 1/2 centered at the origin.
So diam W(A) = 1. On the other hand, the matrix
1
(A)
2
(A) =
_
1/2 1
0 1/2
_
and its norm is bigger than 1. (The norm X cannot be smaller than the Euclidean
norm of any column of X.)
Interesting lower bounds for spd (A) of normal and Hermitian matrices can be ob-
tained from (5) and (6). We illustrate this with a few examples.
Let
1
,
2
be linear functionals on M(n) dened for i = j as
1
(A) =
1
2
_
a
i i
+a
j j
+a
i j
e
i
+a
j i
e
i
_
and
2
(A) =
1
2
_
a
i i
+a
j j
a
i j
e
i
a
j i
e
i
_
.
Both
1
and
2
are positive and unital. (Positivity is a consequence of the fact that if A
is positive semidenite, then
a
i j

a
i i
a
j j

1
2
(a
i i
+a
j j
)). So from (5), we see that
for every normal matrix A,
spd (A)
a
i j
e
i
+a
j i
e
i
.
This is true for every . The maximum value of the right-hand side over is |a
i j
| +
|a
j i
|. Thus, for every normal matrix A we have
spd (A) max
i =j
_
|a
i j
| +|a
j i
|
_
. (8)
This was rst proved by L. Mirsky (see Theorem 3 (iii) in [7]). When A is Hermitian,
this says
spd (A) 2max
i =j
|a
i j
|. (9)
Another result of Mirsky, Theorem 2 in [7], subsumes both the inequalities (4) and (9).
It says that for every Hermitian matrix A, we have
spd (A)
2
max
i =j
_
(a
i i
a
j j
)
2
+4|a
i j
|
2
_
. (10)
This can be obtained from (6) as follows. Let
1
(A) =
_
a
i i
a
i j
a
i j
a
j j
_
and
2
(A) =
_
a
j j
a
i j
a
i j
a
i i
_
.
Then
1
and
2
are positive, unital, linear maps, and
1
(A)
2
(A) =
_
a
i i
a
j j
2a
i j
2a
i j
a
j j
a
i i
_
.
This is a Hermitian matrix with trace 0. Its eigenvalues are , where = [(a
i i

a
j j
)
2
+ 4|a
i j
|
2
]
1/2
. So
1
(A)
2
(A) = . The inequality (10) then follows
from (6).
Next let
1
(A) =
1
n
i, j
a
i j
and
2
(A) =
1
n
_
_
tr A
1
n 1
i =j
a
i j
_
_
.
Both are unital linear functionals. We have already observed that
1
is positive. We
claim that
2
is also positive. If A is any Hermitian matrix, then
i =j
a
i j
= 2Re
i <j
a
i j
2
i <j
|a
i j
|.
If further A is positive semidenite, then we have

a
i j

1
2
(a
i i
+ a
j j
). Combining
these two inequalities, we see that
i =j
a
i j
(n 1)tr A.
So the linear functional
2
is positive. The inequality (5) now shows that for every
normal matrix A, we have
spd (A)
1
n 1
i =j
a
i j
. (11)
This inequality is stated as Theorem 2.1 in [5] and as Theorem 5 in [6], and is proved
there by other arguments.
Many more inequalities, some of them stronger and more intricate than the ones
we have discussed, can be obtained by choosing other positive maps. Enhancing this
technique, we have the inequality of Bhatia and Davis [2]. This says that if is a
positive unital linear map, then for every Hermitian matrix A,
(A
2
) (A)
2
1
4
spd (A)
2
. (12)
Again, choosing different one can obtain a variety of inequalities. This is demon-
strated in [3].
REFERENCES
1. R. Bhatia, Positive Denite Matrices. Princeton University Press, Princeton, NJ, 2007.
2. R. Bhatia, C. Davis, A better bound on the variance, Amer. Math. Monthly 107 (2000) 353357.
3. R. Bhatia, R. Sharma, Some inequalities for positive linear maps, Linear Algebra Appl. 436 (2012) 1562
1571.
4. R. A. Horn, C. R. Johnson, Topics in Matrix Analysis. Cambridge University Press, Cambridge, UK, 1991.
5. C. R. Johnson, R. Kumar, H. Wolkowicz, Lower bounds for the spread of a matrix, Linear Algebra Appl.
71 (1985) 161173.
6. J. K. Merikoski, R. Kumar, Characterizations and lower bounds for the spread of a normal matrix, Linear
Algebra Appl. 364 (2003) 1331.
7. L. Mirsky, Inequalities for normal and Hermitian matrices, Duke Math. J. 24 (1957) 591599.
RAJENDRA BHATIA has been a Professor at the Indian Statistical Institute since 1984. In between, he has
held several visiting appointments in Asia, Europe, and North America, the last of these as a Fellow Professor
at Sungkyunkwan University in Korea. He is the author of ve books, three of which are close to the topics of
this article.
Theoretical Statistics and Mathematics Unit, Indian Statistical Institute, 7, S. J. S. Sansanwal Marg,
New Delhi - 110016, India
Department of Mathematics, Sungkyunkwan University, Suwon 440-746, Republic of Korea
rbh@isid.ac.in
RAJESH SHARMA received his Ph.D. in mathematics from the Himachal Pradesh University in 1992. He
is presently a Professor at the same university, and frequently visits the Indian Statistical Institute, New Delhi.
His research work is focussed on inequalities and matrix analysis.
Department of Mathematics & Statistics, Himachal Pradesh University, Shimla 171005, India
rajesh hpu math@yahoo.co.in
NOTES
Edited by Sergei Tabachnikov
Trisecting Angles in Pythagorean Triangles
Wen D. Chang and Russell A. Gordon
Abstract. Suppose that a
2
+ b
2
= c
2
, where a, b, and c are relatively prime positive integers,
and consider the right triangle T with sides a, b, and c. We prove that both of the acute angles
in T can be trisected with a compass and unmarked straightedge if and only if c is a perfect
cube.
1. INTRODUCTION. It has been known for about two centuries that some angles
cannot be trisected using only a compass and straightedge (assuming the rules es-
tablished by the classic Greek geometers). Most sources mention that certain angles
such as 90
, 45
, and 9
can be trisected with only these tools and then prove that a
60
angle cannot be trisected in this way. In this note, we determine necessary and
sufcient conditions for both of the acute angles in right triangles corresponding to
Pythagorean triples to be trisectible with only a compass and straightedge.
Since both the trisection of angles problem and the properties of Pythagorean triples
have been studied extensively, we leave most of the background details (which are
tedious and distract from our primary goal) to the reader. There are a plethora of
references for these subjects; the short list provided at the end of this note barely
scratches the surface on these topics. A primitive Pythagorean triple (a, b, c) consists
of three relatively prime positive integers such that a
2
+ b
2
= c
2
, while a primitive
Pythagorean triangle refers to the corresponding right triangle having side lengths of
a, b, and c. It is easy to verify that c must be odd and that a and b must have oppo-
site parity; we will always assume that the rst integer in the triple is even. Note that
primitive Pythagorean triangles (and thus their acute angles) can be constructed with
a compass and straightedge, since the sides of the triangle have integer lengths.
2. PRELIMINARIES. To prove our main result (Theorem 3), we need two theo-
retical results. We encourage the reader to look into the details behind each of these
facts. The rst result involves the theory of constructible numbers. A number is con-
structible if, given a line segment of length x, a line segment of length ||x can be
constructed with a compass and straightedge. It can be shown that every constructible
number is an algebraic number for which the degree of its minimal polynomial is a
power of two. See [1] or [6] for a proof of this fact. The geometry text [8] does a nice
job presenting an informal discussion of constructible numbers and explaining why a
60
angle cannot be trisected.

Theorem 1. Suppose that cos = r, where r is a rational number. The angle can be
trisected with a compass and straightedge if and only if the polynomial 4x
3
3x r
has a rational root.
MSC: Primary 51M04, Secondary 11A99
AugustSeptember 2014] NOTES 625
Proof. This result is a consequence of the triple angle formula for cosine and the fact
that the degree of a constructible number must be a power of two. The details can be
found in the texts mentioned above or in a variety of other sources.
In order to prove the next result, we will make use of Gaussian integers. A Gaussian
integer is a number of the form p + qi , where p and q are integers and i =
1.
Unique factorization (up to multiplication by units) is valid for the set of Gaussian
integers (see [3], [7], or [9]) but there is the complicating feature that there are four
units, namely 1, 1, i , and i . It turns out that integer primes of the form 4k + 3 are
prime in the Gaussian integers, while integer primes of the form 4k + 1 can be written
as ww, where both w and w are distinct prime Gaussian integers. (For the record, the
integer 2 is not a Gaussian prime, since 2 = (1 + i )(1 i ). However, the numbers
1 + i and 1 i are not distinct Gaussian primes, since each is the product of a unit
with the other; for example, 1 i = i (1 + i )). Equations such as
37 = (1 + 6i )(1 6i ), (1 + 6i )
2
= 35 + 12i, 12
2
+ 35
2
= 37
2
hint at a connection between Gaussian integers and primitive Pythagorean triples.
Theorem 2. If (A, B, C
3
) is a primitive Pythagorean triple, then there exists a primi-
tive Pythagorean triple (a, b, C) such that A = |a
3
3ab
2
| and B = |b
3
3a
2
b|.
Proof. By properties of Pythagorean triples (see [7], [9], or most any elementary num-
ber theory text), there exist relatively prime positive integers s and t such that s < t ,
t s is odd, and A = 2st , B = t
2
s
2
, C
3
= t
2
+ s
2
. It follows that
C
3
= (t + si )(t si ) and (t + si )
2
= B + Ai.
It is not too difcult to show that the Gaussian integers t + si and t si are relatively
prime (the fact that t s is odd is necessary here, since 2 is not a Gaussian prime) and
thus each must be a perfect cube. Choose a Gaussian integer v + ui so that (v + ui )
3
=
t + si and let a = |2uv| and b = |v
2
u
2
|. We claim that a and b have the desired
properties. Noting that
(a
2
+ b
2
)
3
=
(v + ui )
2
(v ui )
2
3
= ((t + si )(t si ))
2
= C
6
,
we nd that (a, b, C) is a Pythagorean triple. In addition, the equation
B + Ai = (t + si )
2
=
(v + ui )
2
3
=
(v
2
u
2
) + 2uvi
3
=
(v
2
u
2
)
3
3(v
2
u
2
)(2uv)
2
3(v
2
u
2
)
2
(2uv) (2uv)
3
i
= (v
2
u
2
)(b
2
3a
2
) + 2uv(3b
2
a
2
)i
reveals that B = |b(b
2
3a
2
)| and A = |a(a
2
3b
2
)|. Finally, the fact that A and B
are relatively prime guarantees that a and b are relatively prime. This completes the
proof.
3. MAIN RESULTS. We are now able to prove our main result. We mention in pass-
ing that if one of the acute angles in a right triangle can be trisected with a compass
and straightedge, then the other acute angle can also be trisected in this way. To see
this, let and be the two acute angles in a right triangle and suppose that /3 can
be constructed with a compass and straightedge. Since it is easy to construct a 30
angle (just bisect one of the angles in an equilateral triangle), it follows that the angle
/3 = 30
/3 can be constructed and thus can be trisected with a compass and

straightedge.
Theorem 3. The acute angles of a primitive Pythagorean triangle (a, b, c) can be
trisected with a compass and straightedge if and only if c is a perfect cube.
Proof. Suppose rst that the acute angles of the primitive Pythagorean triangle
(a, b, c) can be trisected with a compass and straightedge. By Theorem 1, the polyno-
mial 4x
3
3x (b/c) has a rational root. Let y/z be such a root, where y and z are
relatively prime integers with z > 0. It follows that
4
y
z
3
3
y
z
=
b
c
or
y(4y
2
3z
2
)
z
3
=
b
c
.
If z is odd, then both of the fractions in the second equality are in reduced form. To see
this, suppose that p is a prime that divides both y(4y
2
3z
2
) and z
3
. Then p is odd
and p divides both z and 4y
2
3z
2
. But this implies that p divides 4y
2
and thus y, a
contradiction to the fact that y and z are relatively prime. It follows that c = z
3
. Now,
suppose that z is even (and thus that y is odd) and let z = 2w. In this case,
b
c
=
y(4y
2
12w
2
)
8w
3
=
y(y
2
3w
2
)
2w
3
.
If w is even, then (as above) the last fraction is in reduced form and c = 2w
3
, a con-
tradiction to the fact that c is odd. If w is odd, then similar reasoning shows that the
only common factor of y(y
2
3w
2
) and 2w
3
is 2, and we nd that 2b = y(y
2
3w
2
)
and c = w
3
. We conclude that c is a perfect cube.
Now suppose that c = C
3
is a perfect cube and represent the corresponding
primitive Pythagorean triple by (A, B, C
3
). By Theorem 2, there exists a primitive
Pythagorean triple (a, b, C) such that
A = |a
3
3ab
2
| and B = |b
3
3a
2
b|.
Noting that
4
b
C
3
3
b
C
=
b(4b
2
3C
2
)
C
3
=
b
4b
2
3(a
2
+ b
2
)
C
3
=
b(b
2
3a
2
)
C
3
=
B
C
3
,
we nd that either b/C or b/C is a rational root of 4x
3
3x (B/C
3
). It follows
from Theorem 1 that the acute angles of the primitive Pythagorean triangle (A, B, C
3
)
can be trisected with a compass and straightedge.
The primitive Pythagorean triangle (44, 117, 125) is the smallest such triangle that
has angles that can be trisected with a compass and straightedge. When c is the cube
of a prime, there is only one possible triple, so these types of examples are relatively
simple. In general, if c is the product of n distinct primes of the form 4k + 1, then there
are 2
n1
distinct primitive Pythagorean triples that have c as hypotenuse; see Fassler
[5]. This fact indicates why c and c
3
appear as the third term in the same number
of primitive triples. To give a nontrivial example, consider the number (5 13)
3
. The
two primitive triples corresponding to this value for c are (186416, 201663, 274625)
and (7336, 274527, 274625). Referring to the proof of the theorem, these triples are
generated by the triples (16, 63, 65) and (56, 33, 65), respectively. We leave the details
and other examples to the reader.
We next consider a somewhat related result. By choosing t = s + 1 in the
Pythagorean triple (2st, t
2
s
2
, t
2
+ s
2
), we easily see that there are an innite num-
ber of primitive Pythagorean triples (a, b, c) for which c a = 1. With a little more
effort (Pell equations play a role; see [5]), it can be shown that there are an innite
number of primitive Pythagorean triples (a, b, c) for which |b a| = 1. It turns out
that in each of these cases, the integer c cannot be a perfect cube and thus the acute
angles in the corresponding right triangle cannot be trisected. Although there may be
simpler ways to verify this (for instance, the result follows from the fact that a cube
cannot be expressed as the sum of two consecutive nonzero squares), the proofs of
the following theorem and its corollary provide an interesting perspective and indicate
other algebraic techniques related to the problem of trisecting angles.
Theorem 4. Suppose that b > 1 is a positive integer. If tan = b or tan = 1/b, then
cannot be trisected with a compass and straightedge.
Proof. Since the angles in question are complements of each other, it is sufcient to
consider the case in which tan = b. Using the triple angle formula for tangent, we
nd that
3 tan(/3) tan
3
(/3)
1 3 tan
2
(/3)
= tan = b or x
3
3bx
2
3x + b = 0,
where x = tan(/3). Appealing to the theory of constructible numbers once again, it is
sufcient to prove that this equation has no rational roots. Since the leading coefcient
of the polynomial is 1, the only possible rational roots are integers. Suppose that n is
a nonzero integral root of this equation. Since n divides b, we may write b = j n for
some integer j . We then have
n
3
3 j n
3
3n + j n = 0 or n
2
=
j 3
3 j 1
.
The only choices for j that give an integral perfect square are 1 and 3. These give b
values of 1 and 0, respectively, which are not allowed.
Corollary 5. If the acute angles of a primitive Pythagorean triangle can be trisected
with a compass and straightedge, then no two side lengths of the triangle differ by 1.
Proof. We will prove the contrapositive of this statement. Let (a, b, c) be a primitive
Pythagorean triangle and let be the acute angle opposite the side b. Suppose rst that
c a = 1. We then have
tan(/2) =
sin
1 + cos
=
b/c
1 + (a/c)
=
b
c + a
=
b(c a)
c
2
a
2
=
c a
b
=
1
b
.
It follows that /2 and thus cannot be trisected with a compass and straightedge.
Now, suppose that |b a| = 1. We will assume that b = a + 1; the case a = b + 1
is similar. Since tan = (a + 1)/a, the difference formula for tangent yields
tan( 45
) =
tan 1
1 + tan
=
1/a
2 + (1/a)
=
1
2a + 1
.
By Theorem 4, the angle 45
cannot be trisected. Since the angle 45
can be
trisected, we know that cannot be trisected with a compass and straightedge.
4. CONCLUDING REMARKS. In addition to the references we have given, there
are many sources (both online and text) for the number theory results in this paper.
The same is true for the problem of trisecting angles. A good place to start for the
latter is Dudley [4], since he provides an elementary discussion of the problem and
presents myriad ill-fated attempts to prove that angles can be trisected. The article [2]
gives an elementary introduction to the trisection problem and makes some interesting
observations about which rational numbers can appear as the cosine or tangent of an
angle that can be trisected.
The reader intrigued by the ideas and results presented thus far may enjoy working
through the details of the following related result. This result indicates that Mordell
curves (y
2
= x
3
+ n) play a role in determining which angles can be trisected. The
website for Wolfram MathWorld is a good source for information about these curves.
The text [10] gives an elementary approach for nding integer solutions to these equa-
tions for a few values of n, but it does not make mention of the term Mordell curve.
Theorem 6. Suppose that tan = a/b, where a and b are relatively prime positive
integers.
1. If a +b is odd, then the angle can be trisected with a compass and straightedge
if and only if a
2
+ b
2
is a perfect cube.
2. If a + b is even, then the angle can be trisected with a compass and straight-
edge if and only if (a
2
+ b
2
)/2 is a perfect cube.
Proof. To prove (1), assume that a + b is odd and suppose rst that can be trisected
with a compass and straightedge. By the triple angle formula for tangent and the theory
of constructible numbers, the polynomial bt
3
3at
2
3bt + a, where t = tan(/3),
has a rational root. Let m/n be such a root, where m and n are relatively prime integers
with n > 0. We then have
b
m
n
3
3a
m
n
2
3b
m
n
+ a = 0 or
a
b
=
m
3
3mn
2
3m
2
n n
3
.
We assert that the integers m(m
2
3n
2
) and n(3m
2
n
2
) are relatively prime, leaving
a proof of this fact to the reader. (This is where the hypothesis that a + b is odd is
used.) It follows that a = |m(m
2
3n
2
)| and b = |n(3m
2
n
2
)|, and hence
a
2
+ b
2
=
m
6
6m
4
n
2
+ 9m
2
n
4
9m
4
n
2
6m
2
n
4
+ n
6
= m
6
+ 3m
4
n
2
+ 3m
2
n
4
+ n
6
= (m
2
+ n
2
)
3
.
This shows that a
2
+ b
2
is a perfect cube.
Now suppose that a
2
+ b
2
= c
3
is a perfect cube. Since c
3
= (a + bi )(a bi )
expresses a perfect cube as the product of two relatively prime Gaussian integers,
each of the factors must be a perfect cube as well. Let a + bi = (m + ni )
3
. Then
a = m
3
3mn
2
and b = 3m
2
n n
3
, and it is not difcult to show that m/n is a ra-
tional root of the polynomial bt
3
3at
2
3bt + a. It follows that the angle can be
trisected with a compass and straightedge.
For (2), suppose that both a and b are odd and, without loss of generality, that
0 < a < b. Since a 45
angle can be trisected with a compass and straightedge, it

follows that can be trisected with a compass and straightedge if and only if + 45
can be trisected in this way. Using the sum formula for tangent, we nd that
tan( + 45
) =
tan + 1
1 tan
=
(b + a)/2
(b a)/2
.
Since
1
2
(b +a) and
1
2
(b a) are relatively prime positive integers with opposite parity,
part (1) of the theorem reveals that can be trisected with a compass and straightedge
if and only if
b + a
2
2
+
b a
2
2
=
a
2
+ b
2
2
is a perfect cube. This completes the proof.
To illustrate this theorem, the equation 2
2
+ 11
2
= 5
3
reveals that the angle satis-
fying tan = 2/11 can be trisected with a compass and straightedge. The same is true
for the angle satisfying tan = 9/13, since 9
2
+ 13
2
= 2 5
3
. We note in passing
that the odd integer that is cubed must be a product of primes of the form 4k + 1; this
follows from the fact that the number can be represented as a sum of two squares.
Part (1) of Theorem 6 can be used to give a different proof of Theorem 3. For
primitive Pythagorean triangles corresponding to the triple (a, b, c), the integers a and
b have opposite parity. We thus know that
the acute angles in the triangle can be trisected if and only if a

2
+ b
2
is an odd
perfect cube;
the triangle is Pythagorean if and only if a

2
+ b
2
is an odd perfect square.
Thus, we need Pythagorean triples of the form (a, b, c
3
) in order to guarantee that the
angles can be trisected. This restatement of our main result is a tting place to end this
note.
REFERENCES
1. G. Birkhoff, S. MacLane, A Survey of Modern Algebra. Fourth edition. Macmillan, New York, 1977.
2. K. Chew, On the trisection of an angle, Math. Medley 6 (1978) 1018.
3. J. Conway, R. Guy, The Book of Numbers. Springer-Verlag, New York, 1996.
4. U. Dudley, The Trisectors. Mathematical Association of America, Washington, DC, 1994.
5. A. Fassler, Multiple Pythagorean number triples, Amer. Math. Monthly 98 (1991) 505517.
6. J. Gallian, Contemporary Abstract Algebra. Fifth edition. Houghton Mifin, Boston, 2002.
7. G. Hardy, E. Wright, An Introduction to the Theory of Numbers. Third edition. Clarendon Press, Oxford,
1954.
8. I. Isaacs, Geometry for College Students. American Mathematical Society, Providence, RI, 2001.
9. I. Niven, H. Zuckerman, An Introduction to the Theory of Numbers. Third edition. Wiley, New York,
1972.
10. J. Uspensky, M. Heaslet, Elementary Number Theory. McGraw-Hill, New York, 1939.
Department of Mathematics and Computer Science, Alabama State University, Montgomery, AL 36104
wchang@alasu.edu
Department of Mathematics, Whitman College, Walla Walla, WA 99362
gordon@whitman.edu
Yet Another Direct Proof of the Uncountability of the
Transcendental Numbers
The most known proof of uncountability of the transcendental numbers is based on proving
that A is countable and concluding that R\A is uncountable, since R is. Very recently, J. Gas-
par [1] gave a nice direct proof that the set of transcendental numbers is uncountable. In this
context, the word direct means a proof that does not follow the previous steps. However, we
point out that his proof is based on the transcendence of , which is, to the best of the authors
knowledge, proved by an indirect argument. In this note, in the spirit of Gaspar, we present a
direct proof of the following stronger result.
Theorem. There exist uncountable many algebraically independent real numbers. In partic-
ular, the set of the transcendental real numbers is uncountable.
Proof. Let B be a transcendence basis (which exists by Zorns lemma) of the eld extension
R/Q. If B were countable, then Q(B) would be countable (because its elements are of the
form P(
b)/Q( c) with n N, P, Q Q[X

1
, . . . , X
n
], and

b, c B
n
), so R would be count-
able (because its elements are roots of polynomials in Q(B)[X]\{0} and there would be only
countable roots), which is false. The elements of B are uncountable many algebraically inde-
pendent real numbers.
ACKNOWLEDGMENT. The author would like to thank the anonymous referees for carefully exam-
ining this paper. He is also grateful to Matheus Bernardini for pointing out the Gaspar paper.
REFERENCES
1. J. Gaspar, Direct proof of the uncountability of the transcendental numbers, Amer. Math. Monthly
121 (2014) p. 80.
Submitted by Diego Marques*, University of Brasilia
MSC: Primary 11J81
*Supported by FAP-DF and CPNq-Brazil.
Points Covered an Odd Number of
Times by Translates
Rom Pinchasi
Abstract. Given an odd number of axis-aligned unit squares in the plane, it is known that the
area of the set whose points in the plane that belong to an odd number of unit squares cannot
exceed the area of one unit square, that is, 1. In this paper, we consider the same problem for
other shapes. Let T be a xed triangle and consider an odd number of translated copies of T in
the plane. We show that the set of points in the plane that belong to an odd number of triangles
has an area of at least half of the area of T. This result is best possible. We resolve also the
more general case of a trapezoid and discuss related problems.
1. INTRODUCTION. While preparing puzzles for my puzzles course that I gave
at the Technion during 2009 and 2010, I found the following puzzle in the falls contest
of Tournament of Towns for the year 2009 [2]:
On an innite chessboard are placed 2009 n n cardboard pieces such that
each of them covers exactly n
2
cells of the chessboard. Prove that the number of
cells of the chessboard which are covered by odd numbers of cardboard pieces
is at least n
2
.
As for the history of the problem (as far as I could trace it back), a one-dimensional
version of this puzzle was suggested by Uri Rabinovich and communicated to Igor
Pak. Pak added one more dimension and communicated it to Arseniy V. Akopyan, and
then through some Russian network contacts to the organizers of the Tournament of
Towns in Moscow.
Perhaps the shortest solution to this puzzle is an elegant use of the coloring tech-
nique: Color a grid square (a, b) by the color ((a mod n), (b mod n)). We thus have
n
2
different colors and the crucial observation is that each n n cardboard contains
precisely one grid square of each color. It follows that there must be at least one grid
square of each color class that is covered an odd number of times (because the total
number of grid squares from each color is 2009 with multiple counting).
It is almost an immediate consequence of this puzzle that the continuous version
of this problem is true. That is, given an odd number of axes-parallel unit squares in
the plane, then the total area of all points covered by an odd number of squares is
at least 1 (see Figure 1 for an illustration). Equality is possible, for example, if all
squares coincide. We can also give a direct proof: Color each point (x, y) of the plane
by the color ((x mod 1), (y mod 1)). Then every axes-parallel unit square contains
precisely one point of each color. The rest of the proof is identical to the discrete case,
keeping in mind that any (measurable) set that contains at least one point of each color
has an area that is greater than or equal to 1. We will use variants of this simple idea in
the sequel.
As a nice exercise, suggested by anonymous referee, based on this simple argument
we can show the following. Given an even number of axes-parallel unit squares in the
MSC: Primary 52C20
plane, denote by D the set of points that belong to an odd number of those squares.
Then there is no axes-parallel unit square in the plane that contains more than half of
the area of D.
Figure 1. Five unit squares. The area in black is covered an odd number of times.
We then tried to consider the continuous version with respect to other planar shapes.
The most natural that comes to mind is perhaps the unit disc, and also regular n-gons.
Problem. Let F be a family of an odd number of unit discs in the plane. Is it true that
the total area of all points in the plane that are contained in an odd number of discs in
F is at least ?
As it turns out, Igor Pak preceded me in considering this question, as it appears as
an (unsolved) Exercise 15.14 in [1].
Denition 1. Let T be any given measurable compact set in the plane with positive
measure. We dene f
odd
(T) to be the maximum number such that for any collection
F of an odd number of translates of T, the set of all points in the plane that belong to
an odd number of members from F has area of at least f
odd
(T). We denote by A(T)
the area of the set T.
Clearly, f
odd
(T)/A(T) [0, 1] for any T. As we observed earlier, if T is a square,
then f
odd
(T) = A(T). In fact, it can be shown for any shape T which can tile the plane
that we have f
odd
(T) = A(T). At this point, we could be tempted to conjecture that
f
odd
(T) = A(T) for any T. This is perhaps the rst guess after several failed attempts
to construct T with f
odd
(T) < A(T). The truth, however, is different, as is shown in
the following theorem.
Theorem 1. For any triangle T, we have f
odd
(T) =
1
2
A(T).
We will prove Theorem 1 as a special case of the following more general result
about trapezoids.
Theorem 2. Let T be a trapezoid that is not a parallelogram. Let h denote the height
of T and let x
1
and x
2
denote the lengths of the two parallel sides of T, respectively.
Then f
odd
(T) =
1
4
h|x
2
x
1
|.
Theorem 1 follows from Theorem 2 if we take x
1
= 0 in Theorem 2.
Notice that f
odd
(T)/A(T) can be arbitrarily small when T is a trapezoid. This
happens when the difference between the lengths of the two parallel sides of T
is small compared to their sum. Somewhat surprisingly, this implies that the func-
tion f
odd
(T)/A(T) is far from being continuous. If T is a trapezoid that is very
close to a square, then f
odd
(T)/A(T) is very close to 0, while if T is a square, then
f
odd
(T)/A(T) = 1.
It is interesting to note, however, that the value of
1
4
h|x
2
x
1
| in Theorem 2 is never
actually attained.
Theorem 3. Let T be a proper trapezoid that is not a parallelogram. Let h denote
the height of T and let x
1
and x
2
denote the lengths of the two parallel sides of T,
respectively. Let F be a nite collection of an odd number of translates of T. Then
the area of all points in the plane that belong to an odd number of trapezoids in F is
strictly greater than
1
4
h|x
2
x
1
|.
The proof of Theorem 2 is presented in Section 2. We will provide the proof of
Theorem 3 in Section 3, leaving some of the details to the reader.
2. PROOF OF THEOREM 2. We assume without loss of generality that x
2
> x
1
.
Because T is not a parallelogram, it is enough to show the theorem for the trapezoid
T
whose vertices are A = (0, 0), B = (1, 1), C = (t, 1) and D = (t, 0), where t =
x
2
x
2
x
1
1 (the case t = 1 is where x
1
= 0 and T is a triangle). This is because T
is
the image of T under a linear transformation that we denote by g. Notice that when
x
1
> 0, the ratio between the parallel sides of T and T
is the same as
t
t 1
=
x
2
x
1
. If
x
1
= 0, then both T and T
are triangles and therefore linearly equivalent. Applying

a linear transformation g changes the value of f
odd
(T) by a factor of |det g|. Here, we
have |det g| =
1
h(x
2
x
1
)
, because this is the ratio between the area of T
= g(T) and the

area of T. After applying the linear transformation g, we have h = 1, x
1
= t 1, and
x
2
= t . Hence, if we show the theorem for T
(namely, f
odd
(T
)
1
4
), it will imply that
f
odd
(T) =
1
|det g|
f
odd
(T
) =
1
4
h(x
2
x
1
).
We color each point of the plane as follows. We color a point (a, b) by the color
((a mod
1
2
), (b mod
1
2
)). Unlike our coloring of the plane in the case of dealing with
a square, it is not true in this case that every translated copy of T
contains precisely
one point of each color. However, it is true that every translated copy of T
contains
an odd number of points of each color. To see this, observe that, up to sets of measure
0, T
is the disjoint union of ve regions: two triangles D

1
, D
2
, a square Q, and two
rectangles R
1
and R
2
, as shown in Figure 2.
Consider a translated copy of T
, that is, T
+ (x, y) for some vector (x, y). As

we have already seen, Q + (x, y) contains exactly one point of each color. The two
rectangles R
1
+ (x, y) and R
2
+ (x, y) contain the same multi-set of colors and the
same is true for the two triangles D
1
and D
2
. We conclude that every translated copy
of T
contains an odd number of points from each color.

As in the case of the square, it follows that for each color there must be a point of
this color that belongs to an odd number of translated copies of T
. Because the area

of the basic square of colors is
1
4
, this shows that f
odd
(T
)
1
4
.
To complete the proof of Theorem 2, we will show, by construction, that f
odd
(T
)
is not greater than
1
4
. Let S = T
(, ) for some small > 0. The set of points

that belong to just one of T
and S consists of the rectangle whose vertices are (t, 1

), (t , 1 ), (t, 0), and (t , 0), together with another set of points contained in
2 1
1
0
(t, 1) (1, 1)
t
D
1
D
2
Q
R
2
R
1
1
2
(
1
2
,
1
2
)
(t,
1
2
)
1
2
Figure 2. T
is the disjoint union of ve regions

the two stripes Z
1
= { y 0} and Z
2
= {1 y 1} (see Figure 3(b)). Take
N =
t 1/2
and assume is chosen so that N is an integer. For every i = 1, . . . , N, let

T
i
= T
+(i , 0) and S
i
= S +(i , 0).
Consider now the family F of 2N translated copies of T
, which consists of
T
1
, . . . , T
N
and S
1
, . . . , S
N
. The set of points that belong to an odd number of the sets
in F consists of an axes-parallel rectangle B and an additional set of points that we
denote by W, contained in the stripes Z
1
and Z
2
(see Figure 3(c)). B is the rectangle
whose vertices are (t, 0), (t, 1 ), (2t
1
2
, 1 ), and (2t
1
2
, 0). Hence, B is a
rectangle of dimensions (t 1/2) (1 ).
(a) The trapezoid T
B
Y
1
Y
2
(b) T
and S
(d) 2N+1 translates of T
(c) 2N translates of T
. The black area in Z

1
Z
2
is W.
Z
1
Z
2
Figure 3. Illustrating the construction in the proof of Theorem 2
We nowadd to F the trapezoid T +(t 1/2, 0). Now, F is a family of an odd num-
ber of translated copies of T. The set of those points in the plane that belong to an odd
number of sets fromF is the union of the two triangles Y
1
and Y
2
, together with points
that belong to Z
1
Z
2
, as illustrated in Figure 3(d). Y
1
is the triangle whose vertices
are (t
1
2
, 0), (t, 0), and (t,
1
2
). Y
2
is the triangle whose vertices are (t,
1
2
), (t, 1 ),
and (t +
1
2
, 1 ). The area of Y
1
Y
2
is equal to
1
8
+
1
2
(
1
2
)
2
=
1
4

2
+

2
2
.
The area of those points that belong to Z
1
Z
2
and to one of the sets in F is bounded
fromabove by 4t . It follows that f
odd
(T)
1
4
+4t . Since this is true for every > 0,
we conclude that f
odd
(T)
1
4
.
3. PROOF OF THEOREM 3. By applying a suitable linear transformation in the
plane, as in the proof of Theorem 2, we may assume that the vertices of T are A =
(0, 0), B = (1, 1), C = (t, 1), and D = (t, 0) for t =
x
2
x
2
x
1
> 1 (here we assume that
x
1
> 0; that is, T is a proper trapezoid). Then h = 1 and |x
2
x
1
| = 1, and we need
to show that, given a collection F of odd number of translated copies of T, the area
of those points in the plane that belong to odd number of trapezoids from F is strictly
greater than
1
4
.
Let F be such a collection of an odd number of translated copies of T. We may
assume that no two trapezoids in F coincide, or else we can remove such a pair from
F, as this will not change the set of points in the plane that is covered an odd number
of times by trapezoids in F. Let be the vertical line supporting an edge of at least one
trapezoid from F, such that all trapezoids in F lie in the closed half-plane bounded to
the left of .
Let > 0 be a small constant to be determined later. Let
be the vertical line that

lies to the left of at distance from . We assume that is small enough so that
does not meet any trapezoid not supported by .

We color the points of the plane in the same way as we did in the proof of Theorem
2. Namely, we color the point (x, y) by the color ((x mod
1
2
), (y mod
1
2
)). We know,
as we argued in the proof of Theorem 2, that every trapezoid in F contains an odd
number of points of each color. The crucial observation is that if we cut off the picture
of the region H bounded between and
, then it is still true that every trapezoid

contains an odd number points of each color. This is because every trapezoid is either
disjoint from H or intersects H in a rectangle of dimensions 1. In the latter case,
notice that the color of a point (x, y) in the plane is the same as the color of (x, y +
1
2
)
and therefore a rectangle of dimensions 1 contains either zero or two points of
each color.
It follows now that the total area of all points not in H that are covered an odd
number of times is at least
1
4
, as in the proof of Theorem 2. The crucial point is that
the region H contains points that belong to exactly one trapezoid (and 1 is an odd
number); namely, the area that belongs to the highest trapezoid supported by , but not
to the second highest trapezoid supported by .
We note that the assumption in Theorem 3 that T is a proper trapezoid is not crucial
for the proof, and with small modications, the same idea works if T is a triangle.
ACKNOWLEDGMENTS. The author was supported by ISF grant (grant No. 1357/12) and by BSF grant
(grant No. 2008290).
REFERENCES
1. I. Pak, Lectures on Discrete Geometry and Convex Polyhedra. Cambridge U. Press (to appear), available
at http://www.math.ucla.edu/
~
pak/geompol8.pdf.
2. The International Mathematics Tournament of the Towns, fall of 2009, available at http://www.math.
toronto.edu/oz/turgor/archives/TT2009F_JAproblems.pdf.
Mathematics Department, TechnionIsrael Institute of Technology, Haifa 32000, Israel,
room@math.technion.ac.il
On the Characterization
of Galois Extensions
Meinolf Geck
Abstract. We present a shortcut to the familiar characterizations of nite Galois extensions,
based on an idea from an earlier MONTHLY note by Sonn and Zassenhaus.
Let L K be a eld extension and G := Aut(L, K). We assume throughout that
[L : K] < . Then, automatically, |G| < . The following result is an essential part
of the usual development (and teaching) of Galois theory.
Theorem 1. The following conditions are equivalent.
(a) |G| = [L : K].
(b) L is a splitting eld for a polynomial 0 = f K[X], which does not have
multiple roots in L.
(c) K = {y L | (y) = y for all G}.
If these conditions hold, then L K is called a Galois extension. For example,
condition (b) yields a simple criterion for an extension to be a Galois extension, and
condition (c) is crucial for many applications of Galois theory. Combining all three
immediately shows that, if L K is a Galois extension and M is an intermediate
eld, then L M is also a Galois extension, and so
M = {y L | (y) = y for all H} where H := Aut(L, M) G,
which is a signicant part of the Main Theorem of Galois Theory.
Proofs of Theorem 1 often rely on Dedekinds Lemma on group characters and
Artins Theorem [1, Chap. II, F], and some results on normal and separable exten-
sions. Some textbooks (e.g., [2], [4]) use the Theorem on primitive elements at an
early stage to obtain a shortcut. It is the purpose of this note to point out that there is a
different shortcut, which avoids using the existence of primitive elements, and actually
establishes this existence as a by-product (at least for Galois extensions). This relies
on only a few basic results about elds (e.g., the degree formula and the uniqueness of
splitting elds); no assumptions on the characteristic are required. The starting point
is the following observation.
Lemma 2. If K L, then L is not the union of nitely many elds M such that
K M L.
Proof. If K is innite, then each M as above is a proper subspace of the K-vector
space L, and it is well known and easy to prove that a nite-dimensional vector space
over an innite eld is not the union of nitely many proper subspaces. Now, assume
that K is nite. Then L is also nite, and so |L| = p
n
for some prime p. In this case,
MSC: Primary 12F10
it is not enough just to argue with the vector space structure. We could use the fact that
the multiplicative group of L is cyclic. Or we can argue as follows. Again, it is well
known and easy to prove that, for every m n, there is at most one subeld M L
such that |M| = p
m
. (The elements of such a subeld are roots of the polynomial
X
p
m
X L[X]). So the total number of elements in L that lie in proper subelds is
at most 1 + p + + p
n1
< p
n
= |L|, as desired.
Corollary 3. There exists an element z L such that Stab
G
(z) = {id}.
Proof. If G = {id}, there is nothing to prove. Now, assume that G = {id}. For id =
G, we set M
:= {y L | (y) = y}. Then M
is a eld such that K M
L.
Now apply Lemma 2.
Corollary 4. We always have |G| [L : K]. If equality holds, then there is some
z L such that L = K(z) and the minimal polynomial
z
K[X] has only simple
roots in L; furthermore, L is a splitting eld for
z
.
Proof. Let z L be as in Corollary 3. Let G = {
1
, . . . ,
m
}. Then {
i
(z) | 1 i
m} has precisely m elements. Now [L : K] [K(z) : K] = deg(
z
). Since
z
has
coefcients in K, we have
z
(
i
(z)) =
i
(
z
(z)) = 0 for all i . So
z
has at least
m distinct roots; in particular, deg(
z
) m = |G|. This shows that [L : K] |G|.
If [L : K] = |G|, then all of the above inequalities must be equalities. This yields
L = K(z) and deg(
z
) = m; in particular,
z
=

m
i =1
(X
i
(z)) has only simple
roots and L is a splitting eld for
z
.
Corollary 4 shows the implication (a) (b) in Theorem 1 and also establishes
the existence of a primitive element. Then the remaining implications in Theorem 1
are proved by standard arguments, which we briey sketch.
Proof of (a) (c). Let M := {y L | (y) = y for all G}. Then M is a eld
such that K M L and it is clear from the denitions that G = Aut(L, M). Hence,
Corollary 4 shows that |G| [L : M] [L : K]. Since (a) holds, this implies that
[L : M] = [L : K] and so M = K.
Proof of (c) (b). Let L = K(z
1
, . . . , z
m
) and form the set B := {(z
i
) | 1 i
m, G}. Since (c) holds, we have f :=
zB
(X z) K[X]; furthermore, L is a
splitting eld for f , and f has no multiple roots.
Proof of (b) (a). This relies on a standard result on extending eld isomorphisms
(which is also used to prove that any two splitting elds of a polynomial are isomor-
phic; see, e.g., [1, Theorem 10]). Using this result and induction on [L : K], it is a
simple matter of bookkeeping (no further theory required) to construct [L : K] dis-
tinct elements of G; the details can be found, for example, in [2, Ch. 14, (5.4)]. This
shows that |G| [L : K], and Corollary 4 yields equality.
Once Theorem 1 and Corollary 4 are established, the Main Theorem of Galois
Theory now follows rather quickly; see [2, Ch. 14, 5] or [4, 9.3].
Remark 5. The idea of looking at elements of L that do not lie in proper subelds is
taken from [3].
REFERENCES
1. E. Artin, Galois Theory. Edited and with a supplemental chapter by A. N. Milgram. Reprint of the 1944
second edition. Dover, Mineola, NY, 1998.
2. M. Artin, Algebra. Prentice Hall, Englewood Cliffs, NJ, 1991.
3. J. Sonn, H. Zassenhaus, On the theorem on the primitive element, Amer. Math. Monthly 74 (1967) 407
410.
4. J. Stillwell, Elements of Algebra: Geometry, Numbers, Equations. Springer-Verlag, New York, 1994.
IAZ Lehrstuhl f ur Algebra, Universit at Stuttgart, Pfaffenwaldring 57, 70569 Stuttgart, Germany
meinolf.geck@mathematik.uni-stuttgart.de
Simpsons Rule and Sums of Powers
Our aim is to show that Simpsons Rule yields the well-known closed formulas for the rst Sums
of Powers S
k
(n) =

n
i =1
i
k
= 1
k
+ 2
k
+ + n
k
. For sure, most of us have handled closed-form
expressions for S
k
(n) in order to calculate denite integrals through the limit of some of its Riemann
sums, a simple example being
1
0
x
2
dx = lim
n
n
i =1
1
n
i
n
2
= lim
n
S
2
(n)
n
3
= lim
n
n(n +1)(2n +1)
6n
3
=
1
3
.
Our result moves in the opposite direction: starting from a formula of numerical integration, we calcu-
late S
k
(n). We expect our derivation to give new insight in these old, but always alive, subjects.
First recall the celebrated Compound Simpsons Rule. It reads
b
a
f (x) dx =
ba
6n
f (a)+4
n1
i =0
f (x
2i +1
)+2
n1
i =1
f (x
2i
)+ f (b)
f
(4)
()
(ba)
5
2880n
4
, (1)
where f C
4
[a, b], n is a positive integer, x
i
= a +i
ba
2n
for i = 0, 1, . . . , 2n, and a < < b. What
if we use (1) with f (x) = f
k
(x) = x
k
and with either a = 0, b = 1 or a = 1/(2n), b = 1 +(1/(2n))?
For the cases k = 1, 2, 3, 4, for which f
(4)
k
() = 24
4k
, we easily get
(6n +k +1)(2n)
k
2(k +1)
+
4k
4n
10
= 2 T
k
(n) +2
k
S
k
(n), (2)
3((2n +1)
k+1
1)
(k +1)
+1 (2n +1)
k
+
4k
4n
5
= 2 T
k
(n) +4 2
k
S
k
(n), (3)
where T
k
(n) =
n
i =1
(2i 1)
k
= 1
k
+3
k
+ +(2n 1)
k
.
Now combine (2) and (3) to obtain, for k = 1, 2, 3, 4,
S
k
(n) =
1
3 2
k
3((2n +1)
k+1
1)
(k +1)
+1 (2n +1)
k
+
4k
4n
10

(6n +k +1)(2n)
k
2(k +1)
,
and hence S
1
(n) = n(n + 1)/2, S
2
(n) = n(n + 1)(2n + 1)/6, S
3
(n) = n
2
(n + 1)
2
/4 and S
4
(n) =
n(n +1)(2n +1)(3n
2
+3n 1)/30.
Maybe the use of different composite rules provides new formulas. We leave the interested reader
the task of hunting them.
Submitted by Samuel G. Moreno* and Esther M. Garca-Caballero*,
Universidad de Ja en
MSC: Primary 65D32
*This work was partially supported by Junta de Andaluca, Research Group FQM 0178.
Continuous Images of Cantors Ternary Set
F. Dreher and T. Samuel
Abstract. The HausdorffAlexandroff Theorem states that any compact metric space is the
continuous image of Cantors ternary set C. It is well known that there are compact Hausdorff
spaces of cardinality equal to that of C that are not continuous images of Cantors ternary
set. On the other hand, every compact countably innite Hausdorff space is a continuous
image of C. Here, we present a compact countably innite non-Hausdorff space that is not the
continuous image of Cantors ternary set.
1. INTRODUCTION. One of the simplest sets that is widely studied by and most
important to many mathematicians, in particular analysts and topologists, is Cantors
ternary set (also referred to as the middle third Cantor set), introduced by H. Smith [6]
and by G. Cantor [2]. Cantors ternary set is generated by the simple recipe of dividing
the unit interval [0, 1] into three parts, removing the open middle interval, and then
continuing the process so that at each stage, each remaining subinterval is similarly
subdivided into three and the middle open interval removed. Continuing this process
ad innitum, we obtain a nonempty set consisting of an innite number of points. We
now formally dene Cantors ternary set in arithmetic terms.
Denition 1.1. Cantors ternary set is dened to be the set
C :=
nN
n
3
n
:
n
{0, 2} for all n N
.
Throughout, C will denote Cantors ternary set, equipped with the subspace
topology induced by the Euclidean metric. Below, we list some of the remarkable
properties of Cantors ternary set. For a proof of Property (1), more commonly known
as the HausdorffAlexandroff Theorem, we refer the reader to [8, Theorem 30.7].
Both F. Hausdorff [4] and P. S. Alexandroff [1] published proofs of this result in 1927.
For Properties (2) to (8), we refer the reader to [3] or [7, Counterexample 29]; for a
proof of Properties (9) and (10), the denition of the Lebesgue measure, Hausdorff
dimension, and that of a self-similar set, we refer the reader to [3]. For basic denitions
of topological concepts, see [5] or [8].
1. Any compact metric space is the continuous image of C.
2. The set C is totally disconnected.
3. The set C is perfect.
4. The set C is compact.
5. The set C is nowhere dense in the closed unit interval [0, 1].
6. The set C is Hausdorff.
7. The set C is normal.
8. The cardinality of C is equal to that of the continuum.
MSC: Primary 28A80, Secondary 54A25
9. The one-dimensional Lebesgue measure and outer Jordan content of C are both
equal to zero.
10. The set C is a self-similar set and has Hausdorff dimension equal to
log(2)/ log(3).
In this note, we are interested in whether the HausdorffAlexandroff Theorem can
be strengthened. To avoid any misunderstandings regarding the compactness condition
that might stem from different naming traditions, we shall explicitly dene what we
mean by compact. Note that a compact space does not have to be Hausdorff.
Denition 1.2 (Compact). Given a subset A X of a topological space (X, ), an
open cover of A is a collection of open sets whose union contains A. An open subcover
is a sub-collection of an open cover whose union still contains A. We call a subset A
of X compact if every open cover has a nite open subcover.
It is known that for compact Hausdorff spaces, the properties (i) metrizability, (ii)
second-countability, and (iii) being a continuous image of Cantors ternary set, are
equivalent. This follows from the fact that a compact Hausdorff space is metrizable
if and only if it is second-countable (see, for instance, [5, p. 218]), the Hausdorff
Alexandroff Theorem, and the fact that the continuous image of a compact metric
space (in our case the Cantor set C) in a compact Hausdorff space is again a compact
metrizable space ([8, Corollary 23.2]).
Obviously, the cardinality of a space that is the continuous image of C cannot ex-
ceed that of the continuum. This restriction on cardinality is a necessary but not a suf-
cient condition; there are compact Hausdorff spaces with cardinality equal to that of
the continuum which are not second-countable, for instance, the Alexandroff one-point
compactication of the discrete topological space (R, P(R)), where P(R) denotes the
power set of R.
Restricting the cardinality even further leads to a sufcient condition. If we only
look at countably innite target spaces, we can deduce that for compact countably
innite spaces, the Hausdorff property already implies metrizability. In a countably
innite Hausdorff space, every point is a G
point, which, together with compactness,

implies that the space is rst-countable [8, Problem 16.A.4]; since the space is count-
ably innite, it follows that it is also second-countable. Hence, we have a space that
is compact Hausdorff and second-countable and so, using the above-mentioned equiv-
alence, metrizable. Therefore, any compact countably innite Hausdorff space is the
continuous image of C. Is this strong restriction on the spaces cardinality a sufcient
condition for the HausdorffAlexandroff Theorem, that is, is every compact countably
innite topological space the continuous image of C? This is precisely the question
we address and, in fact, show that the answer is no, by exhibiting a counterexample.
Theorem 1.3. There exists a compact countably innite topological space (T, ) that
is not the continuous image of C.
To prove this result, we require an auxiliary result, Lemma 2.1. In the proof of this
result, we rectify an error in [7, Counterexample 99].
The main idea behind the proof of Theorem 1.3 is to choose a specic space that
is not Hausdorff, and to show that if there exists a continuous map from the Cantor
set into this space, then the continuous map must push forward the Hausdorff property
of Cantors ternary set, which will be a contradiction to how the target space was
originally chosen.
2. PROOF OF THEOREM 1.3.
Lemma 2.1. There exists a countable topological space (T, ) with the following
properties:
(a) (T, ) is compact,
(b) (T, ) is non-Hausdorff, and
(c) every compact subset of T is closed with respect to .
Proof. This proof is based on [7, Counterexample 99]. We dene
T := (N N) {x, y} ,
namely, the Cartesian product of the set of natural numbers N with itself unioned with
two distinct arbitrary points x and y. We equip the set T with the topology whose
base consists of all sets of the form:
(i) {(m, n)}, where (m, n) N N,
(ii) T \ A, where A (N N) {y} contains y and is such that the cardinality
of the set A {(m, n) : n N} is nite for all m N; that is, the set A con-
tains at most nitely many points on each row (these sets T \ A are the open
neighbourhoods of x), and
(iii) T \ B, where B (N N) {x} contains x and is such that there exists an
M N, so that if (m, n) B (N N), then m M; that is, B contains
only points from at most nitely many rows (these sets T \ B are the open
neighborhoods of y).
Property (a) follows from the observation that any open cover of T contains at least
one open neighborhood U T \ A of x and one open neighborhood V T \ B of
y with A and B as given above. The points not already contained in these two open
sets are contained in T \ (U V) T \ ((T \ A) (T \ B)) = A B, which, by
construction, is a nite set. In this way, a nite open subcover can be chosen and hence
the topological space (T, ) is compact.
To see why (T, ) has Property (b), consider open neighborhoods of x and y. An
open neighborhood of x contains countably innitely many points on each row of the
lattice NN; an open neighborhood of y contains countably innitely many full rows.
It follows that there are no disjoint open neighborhoods U x and V y, and thus T
is non-Hausdorff.
We use contraposition to prove Property (c). Suppose that E T is not closed. We
may assume that E is a strict subset of T, since T itself is closed by the fact that .
By construction of the topology, a set that is not closed cannot contain both x and y.
Also, there needs to be at least one point in the closure E of E, but not already in E;
this has to be one of the points x or y, because singletons {(m, n)} that are subsets of
the lattice N N are open. Thus, the point (m, n) cannot be a limit point of E. We
shall now check both cases, that is, (i) if x E \ E, and (ii) if y E \ E.
(i) If x E \ E, then every open neighborhood of x has a nonempty intersection
with E. It follows that there is at least one row in N N that shares innitely
many points with E. Denote this row by B. Then the open cover
{T \ (B {})} {{b} : b B E}
of E cannot be reduced to a nite open subcover, and therefore, E is not com-
pact.
(ii) If y E \ E, then, similar to (i), we have that E contains points from innitely
many rows. Take one point from each of these rows and call the resulting set
A. Then the open cover {T \ (A {y})} {{a} : a A E} of E cannot be
reduced to a nite open subcover, and hence E is not compact.
Proof of 1.3. Assume that there exists a surjective continuous map f : C T. We
will show that this implies that (T, ) is Hausdorff, which contradicts Lemma 2.1(b).
Choose two distinct points u, v T. Since singletons are compact, Lemma 2.1(c)
implies that the sets {u} and {v} are closed in T with respect to . Therefore, their
pre-images under f are nonempty closed subsets of C and have disjoint open neigh-
borhoods U(u) and U(v), as C is normal. The complements U(u)
c
and U(v)
c
of these
open neighborhoods are compact subsets of C. Thus, their images under f are com-
pact in T because f is continuous, and they are closed because of Lemma 2.1(c).
Therefore, V (u) := f (U (u)
c
)
c
and V (v) := f (U (v)
c
)
c
are open neighborhoods of
u and v, respectively. We claim that these sets are disjoint:
V (u) V (v) = f (U (u)
c
)
c
f (U (v)
c
)
c
= ( f (U (u)
c
) f (U (v)
c
))
c
= ( f (U (u)
c
U (v)
c
))
c
= ( f ((U (u) U (v))
c
))
c
= ( f (
c
))
c
= .
Hence, we have separated the points u and v by open neighborhoods. Since both u and
v T were chosen arbitrarily, we conclude that (T, ) is Hausdorff, giving the desired
contradiction.
REFERENCES
1. P. S. Alexandroff,

Uber stetige Abbildungen kompakter R aume, Math. Annalen 96 (1927) 55571.
2. G. Cantor, Grundlagen einer allgemeinen Mannigfaltigkeitslehre, Math. Annalen 21 (1883) 545591.
3. K. J. Falconer, Fractal Geometry. Second edition. John Wiley and Sons, Chichester, West Sussex, 2003.
4. F. Hausdorff, Mengenlehre, Zweite, neubearbeitete Auage. Walter de Gruyter, 1927.
5. J. R. Munkres, Topology. Second edition. Prentice-Hall, Upper Saddle River, NJ, 2000.
6. H. J. S. Smith, On the integration of discontinuous functions, Proc. London Math. Soc. 6 (1875) 140153.
7. L. R. Steen, J. A. Seebach (Jr.), Counterexamples in Topology. Dover Publications, New York, 1978.
8. S. Willard, General Topology. Addison-Wesley, Reading, MA, 1970.
Fachbereich 3 - Mathematik, Universit at Bremen, 28359 Bremen, Germany
fdreher@uni-bremen.de
Fachbereich 3 - Mathematik, Universit at Bremen, 28359 Bremen, Germany
tony@math.uni-bremen.de
Extended Echelon Form and Four Subspaces
Robert A. Beezer
Abstract. Associated with any matrix, there are four fundamental subspaces: the column
space, row space, (right) null space, and left null space. We describe a single computation
that makes readily apparent bases for all four of these subspaces. Proofs of these results rely
only on matrix algebra, not properties of dimension. A corollary is the equality of column rank
and row rank.
Bringing a matrix to reduced row-echelon form via row operations is one of the most
important computational processes taught in an introductory linear algebra course. For
example, if we augment a square nonsingular matrix with an identity matrix of the
same size and row-reduce the resulting n 2n matrix, we obtain the inverse in the
nal n columns:
[A | I
n
]
RREF
I
n
| A
1
.
What happens if A is singular, or perhaps not even square?
We depict this process for an arbitrary m n matrix A:
M = [A | I
m
]
RREF
N = [B | J] =
C K
0 L
.
The columns are partitioned into the rst n columns, followed by the last m columns.
The rows are partitioned so that C has a leading one in each row. We refer to N as
the extended echelon form of A. While it is transparent that C contains information
about A, it is less obvious that L also contains signicant information about A.
Given a matrix A, four associated subspaces are of special interest: the column
space C (A), the (right) null space N (A) = {x | Ax = 0}, the row space R(A), and
the left nullspace N
A
t
. These are Strangs four subspaces, whose interesting proper-

ties and important relationships are described in the Fundamental Theorem of Linear
Algebra (see [4] and [5]). Bases for all four of these subspaces can be obtained easily
from C and L. Additionally, we obtain the result that row rank and column rank are
equal. We only know of one other textbook [3] besides our own [1], that describes
this procedure. (See also Lays paper [2].) So, one purpose of this note is to make this
approach better known.
Several informal observations about extended echelon form are key. As we perform
the row operations necessary to bring M to reduced row-echelon form, the conversion
of I
m
to J records all of these row operations, as it is the product of the elementary
matrices associated with the row operations. Indeed, B = J A (see Lemma 1). Second,
J is nonsingular, which we can see by its row-equivalence with I
m
, or recognized as
a product of elementary matrices, each of which is nonsingular. Third, the entries of
a row of L are the scalars in a relation of linear dependence on the rows of A, so
each row of L is an element of N
A
t
. We will see that the rows of L are a basis for

N
A
t
.
MSC: Primary 15A03, Secondary 97H60; 15A23
We begin our last observation by considering that there can be many different se-
quences of row operations that bring a matrix to reduced row-echelon form. If A does
not have rank m, then part of this variety is due to the many ways row operations
can produce zero rows. However, because the reduced row-echelon form of a matrix
is unique, we conclude that J is unique. Thus, J can be interpreted as a canonical
transformation of the rows of A to reduced row-echelon form.
Lemma 1. For any matrix A, with the notation as above, B = J A and J is nonsingu-
lar. For vectors x and y, Ax = y if and only if Bx = Jy.
Proof. Since N is obtained from M via row operations, there is a nonsingular matrix
R (a product of elementary matrices) such that N = RM. The last m columns of this
matrix equality give J = R, and hence J is nonsingular. If we replace R by J in the
matrix equality, then the rst n columns give B = J A.
The equivalence follows from substitutions and the invertibility of J:
Ax = y J Ax = Jy Bx = Jy.
Informally, the equivalence simply says that if we solve the system Ax = y for x
by augmenting the coefcient matrix A with the vector y and row-reducing, then we
obtain a matrix in reduced row-echelon form representing an equivalent system having
B as coefcient matrix and Jy as the vector of constants.
Because A and B are row-equivalent, and because B and C differ only in the zero
rows of B, it should be apparent that N (A) = N (C) and R(A) = R(C). While the
individual rows of L are easily explained as elements of the the left null space of A,
together they form a basis for the left null space. Less obvious to a student is that the
null space of L is the column space of A!
We now establish these two results about L carefully. Our second purpose is to give
proofs that establish these set equalities without ever exploiting the properties of the
subspaces as vector spaces, in contrast to the arguments on dimension used in [2]. This
gives us a fundamental result about the dimensions of these subspaces as a corollary.
The equivalence of Lemma 1 is our primary tool in the proofs of the next two lemmas.
Lemma 2. If A is a matrix with L as above, then C (A) = N (L).
Proof. Suppose that y C (A). Then there exists a vector x such that Ax = y. There-
fore,
Cx
0
Cx
0x
= Bx = Jy =
Ky
Ly
.
If C has r rows, then the last m r entries of this vector equality imply that y N (L).
Conversely, suppose that y N (L). Because C is in reduced row-echelon form,
there is a vector x such that Cx = Ky. Then
Jy =
Ky
Ly
Cx
0
Cx
0x
= Bx.
Thus, Ax = y and y C (A).
Our next lemma can be obtained by taking orthogonal complements of the sub-
spaces in the conclusion of Lemma 2. Instead, we provide a direct proof.
Lemma 3. If A is a matrix with L as above, then N
A
t
= R(L).
Proof. Suppose that y N
A
t
. Since J is nonsingular, there is a vector x such that

J
t
x = y. Suppose C has r rows, and write x as a partitioned vector with parts of size
r and size m r, x =
u
v
. Then
C
t
u =
C
t
0

u
v
= B
t
x = (J A)
t
x = A
t
J
t
x = A
t
y = 0.
Since C is in reduced row-echelon form, the rows of C are linearly independent, so in
turn the columns of C
t
are linearly independent. This implies that u = 0. Now
y = J
t
x = [K
t
| L
t
]
0
v
= L
t
v.
So y C
L
t
, thus implying y R(L).

Conversely, suppose that y R(L). Then there is a vector v such that L
t
v = y.
Then
A
t
y = A
t
L
t
v
= A
t

K
t
L
t

0
v
= A
t
J
t
0
v
= (J A)
t
0
v
= B
t
0
v
C
t
0
t

0
v
= 0. (1)
Thus y N
A
t
.
So far, we have only employed denitions and matrix operations in the proofs of
these results. Now, consider the vector space structure of these subspaces, specically
their dimensions. We choose to dene the rank of a matrix, r, as the dimension of the
row space. From this denition, it follows that the matrix C must have r rows, and
consequently L has m r rows. Recall that if a matrix with n columns has reduced
row-echelon form with r pivot columns, then there is a natural basis of the null space
with n r vectors. Notice that L is in reduced row-echelon form with no zero rows.
So the dimension of N (L) is m (m r) = r, and by Lemma 2, the dimension of
C (A) is r. So we have an argument based only on denitions and matrix operations
that says the two types of rank, the row rank and the column rank, are equal. In a
similar vein, the dimension of R(L) is m r, and so by Lemma 3 the dimension of
N
A
t
is m r.
We summarize our results as a theorem and corollary.
Theorem. Suppose that A is an m n matrix of rank r. Let C and L be matrices
dened as above. Then we have the following.
1. The row space of A is the row space of C, with dimension r.
2. The null space of A is the null space of C, with dimension n r.
3. The column space of A is the null space of L, with dimension r.
4. The left null space of A is the row space of L, with dimension m r.
So a student can easily obtain all four fundamental subspaces from extended ech-
elon form as row spaces or null spaces of the matrices C and L. Bases for these sub-
spaces are easy to enumerate, since both C and L are in reduced row-echelon form.
From this theorem, we have the very important corollary about dimension.
Corollary. If A is a matrix, then dimR(A) = dimC (A).
ACKNOWLEDGMENTS. Jeff Stuart provided a literature search of similar results in introductory textbooks,
of which there was only one. Helpful suggestions from a referee have improved this note, and are greatly
appreciated.
REFERENCES
1. R. A. Beezer, A First Course in Linear Algebra. Congruent Press, Gig Harbor, WA, 2012, available at
http://linear.pugetsound.edu.
2. D. C. Lay, Subspaces and echelon forms, College Math. J. 24 (1993) 5762.
3. T. Shifrin, M. R. Adams, Linear Algebra: A Geometric Approach. W. H. Freeman, New York, 2011.
4. G. Strang, The fundamental theorem of linear algebra, Amer. Math. Monthly 100 (1993) 848855.
5. , Linear Algebra and Its Applications. Fourth edition. Thomson Brooks/Cole, Belmont, CA, 2005.
Department of Mathematics and Computer Science, University of Puget Sound, Tacoma WA 98416
beezer@pugetsound.edu
PROBLEMS AND SOLUTIONS
Edited by Gerald A. Edgar, Doug Hensley, Douglas B. West
with the collaboration of Itshak Borosh, Paul Bracken, Ezra A. Brown, Randall
Dougherty, Tam as Erd elyi, Zachary Franco, Christian Friesen, Ira M. Gessel, L aszl o
Lipt ak, Frederick W. Luttmann, Vania Mascioni, Frank B. Miles, Richard Pefer,
Dave Renfro, Cecil C. Rousseau, Leonard Smiley, Kenneth Stolarsky, Richard Stong,
Walter Stromquist, Daniel Ullman, Charles Vanden Eynden, Sam Vandervelde, and
Fuzhen Zhang.
Proposed problems and solutions should be sent in duplicate to the MONTHLY
problems address on the back of the title page. Proposed problems should never
be under submission concurrently to more than one journal. Submitted solutions
should arrive before December 31, 2014. Additional information, such as gen-
eralizations and references, is welcome. The problem number and the solvers
name and address should appear on each solution. An asterisk (*) after the num-
ber of a problem or a part of a problem indicates that no solution is currently
available.
PROBLEMS
11789. Proposed by Gregory Galperin, Eastern Illinois University, Charleston, IL,
and Yury J. Ionin, Central Michigan University, Mount Pleasant, MI. Let a and k be
positive integers. Prove that for every positive integer d there exists a positive integer
n such that d divides ka
n
n.
11790. Proposed by Arkady Alt, San Jose, CA and Konstantin Knop, St. Petersburg,
Russia. Given a triangle with semiperimeter s, inradius r, and medians of length m
a
,
m
b
, and m
c
, prove that m
a
m
b
m
c
2s 3(2
3 3)r.
11791. Proposed by Mari an

Stofka, Slovak University of Technology, Bratislava, Slo-
vakia. Show that for r 1,
r
s=1
_
6r 1
6s 2
_
B
6s2
=
6r 1
6
,
where B
n
denotes the nth Bernoulli number.
11792. Proposed by Stephen Scheinberg, Corona del Mar, CA. Show that every innite
dimensional Banach space contains a closed subspace of innite dimension and innite
codimension.
11793. Proposed by Istv an Mez o, Nanjing University of Information Science and Tech-
nology, Nanjing, China. Prove that
n=1
log(n 1)
n
2
=
/
(2)
n=3
(1)
n1
(n)
n 2
,
where denotes the Riemann zeta function and
/
denotes its derivative.
648 c _ THE MATHEMATICAL ASSOCIATION OF AMERICA [Monthly 121
11794. Proposed by George Stoica, University of New Brunswick, Saint John, Canada.
Find every twice differentiable function f on R such that for all nonzero x and y,
x f ( f (y)/x) = y f ( f (x)/y).
11795. Proposed by Mircea Merca, University of Craiova, Craiova, Romania. Let p
be the partition counting function on the set Z
of positive integers, and let g be the

function on Z
given by g(n) =
1
2
{n/2 {(3n 1)/2. Let A(n) be the set of positive
integer triples (i, j, k) such that g(i ) j k = n. Prove for n 1 that
p(n) =
1
n
(i, j,k)A(n)
(1)
{i /21
g(i ) p( j ) p(k).
SOLUTIONS
Inequality for a Convex Quadrilateral
11655 [2012, 523]. Proposed by P al P eter D alyay, Szeged, Hungary. Let ABCD be
a convex quadrilateral, and let , , , and be the radian measures of angles DAB,
ABC, BCD, and CDA, respectively. Suppose > and > , and let
= and = . Let a, b, c, d, e, f be real numbers with ac =
bd = ef . Show that if abe > 0, then
a cos b cos c cos d cos e cos f cos
be
2a

cf
2b

de
2c

af
2d
,
while for abe < 0 the inequality is reversed.
Solution by Richard Stong, Center for Communications Research, San Diego, CA. It
sufces to consider the case abe > 0; the other case follows by negating a, b, c, d, e, f .
Write
AB
,
BC
,
CD
, and
DA
for outward unit normal vectors to the sides AB, BC,
CD, and DA, respectively. We have
DA

AB
= cos ,
AB

BC
= cos ,
BC

CD
= cos ,
CD

DA
= cos ,
AB

CD
= cos ,
BC

DA
= cos .
Let
r =
2abe
2e
, s =
2abe
2a
, t =
ac
2abe
, u =
2abe
2b
.
We compute
a = 2ru, b = 2rs, c = 2st, d = 2t u, e = 2su, f = 2rt,
r
2
s
2
t
2
u
2
=
ab
2e

be
2a

ac
2
2be

ae
2b
=
af
2d

be
2a

cf
2b

de
2c
.
Thus the right side of the desired inequality is r
2
t
2
s
2
u
2
, and the left side is
the negation of
2ru
DA

AB
2rs
AB

BC
2st
BC

CD
2t u
CD

DA
2su
BC

DA
2rt
AB

CD
.
AugustSeptember 2014] PROBLEMS AND SOLUTIONS 649
Thus the desired inequality is equivalent to
_
_
r
AB
s
BC
t
CD
u
DA
_
_
2
0,
which holds trivially.
Also solved by GCHQ Problem Solving Group (U. K.), and the proposer.
How Many Polynomial Shapes Are There?
11656 [2012, 608]. Proposed by Valerio De Angelis, Xavier University of Louisiana,
New Orleans, LA. The sign chart of a polynomial f with real coefcients is the list of
successive pairs (, ) of signs of ( f
/
, f ) on the intervals separating real zeros of f f
/
,
together with the signs at the zeros of f f
/
themselves, read from left to right. Thus, for
x
3
3x
2
, the sign chart is ((1, 1), (0, 0), (1, 1), (0, 1), (1, 1), (1, 0), (1, 1)).
As a function of n, how many distinct sign charts occur for polynomials of degree n?
Solution by Ronald E. Prather, Oakland CA. We count the sign charts, but do not prove
that they can all be achieved by polynomials of the required degree.
Let n 1. The polynomials f of degree n satisfying lim
x
f (x) = pro-
duce half of the sign charts, so it sufces to count the sign charts for them.
Instead of sign charts, we rst consider a related enumeration of a set D(n) of
shapes of polynomials f of degree n. A shape will be a nite sequence chosen from
a set of twelve symbols. We write m when, scanning from the left, we encounter a
local minimum of f (that is, f
/
has a zero of odd order, f
/
< 0 to the left, and f
/
> 0
to the right). We write M on meeting a local maximum (that is, a point at which f
/
has a zero of odd order, f
/
> 0 to the left, and f
/
< 0 to the right). We write I for
a decreasing stationary point (that is, f
/
has a zero of even order and f
/
< 0 on both
sides). We write J for an increasing stationary point (that is, f
/
has a zero of even
order and f
/
> 0 on both sides). Each of these symbols will have subscript , 0, or ,
according as f > 0, f = 0, or f < 0 at the point. This yields twelve symbols: three
ways to subscript each of m, M, I , and J.
The shape of a polynomial f is the sequence, from left to right, of the symbols cor-
responding to the zeros of f
/
. With the restriction that f is positive and f
/
is negative
far to the left, we will be able to recover the sign chart from the shape of a polynomial
and vice versa. In particular, from the data in the shape, we can discover the intervals
where there is a point with f = 0 but f
/
,= 0.
The polynomial x
3
3x
2
, the negative of the example in the problem statement,
has shape m
0
M
. Setting f (x) = x
3
2x
2
, let us recover the corresponding sign
chart. Left of the point with symbol m
0
(or just m
0
) f is positive and decreasing.
Between m
0
and M
, f is positive and increasing. Just right of M
, f is positive and
decreasing. Since f at the far right, there is a zero of f where it changes sign.
This yields
_
(1, 1), (0, 0), (1, 1), (0, 1), (1, 1), (1, 0), (1, 1)
_
.
Next we dene weight. The weight of a subscripted m or M is 1, the weight of
a subscripted I or J is 2, and the weight of a shape is the sum of the weights of
its components. The case of weight 0 with no components is allowed. The weight of
the shape of a polynomial f is equal to the degree of f
/
, possibly reduced by an even
number, so the weight is the degree of f minus an odd positive number. Write S(w) for
the set of all shapes of polynomials f with weight w for which lim
x
f (x) = .
We have [D(n)[ = [S(n 1)[ [S(n 3)[ [S(n 5)[ . . . , where we end at [S(0)[
or [S(1)[.
We must now compute [S(w)[. The difculty is that not every sequence of the
twelve symbols can occur. For example, following a zero of f
/
of type m
the func-
tion is positive and increasing, so the next symbol must be either M
or J
. Another
example: m
can only be followed by M
, M
0
, M
, J
, J
0
, or J
.
We next provide a code for our shapes, symbol by symbol, according to the con-
vention
m
a m
0
b m
c M
c M
0
a M
b
I
bb I
0
ba I
aa J
aa J
0
ba J bb.
The coding of a shape of weight w is a string of w symbols from the alphabet {a, b, c]
such that the substring ab never appears. All such w-strings occur as encodings.
Given a string (with no ab), here is the method to obtain the corresponding shape.
A string of w letters (as, bs, and cs) arising in this fashion from a polynomial is
the concatenation of some k batches of strings b
u
a
v
c, with the nal c perhaps absent:
b
u
1
a
v
1
cb
u
2
a
v
2
c . . . b
u
k
a
v
k
c
where k 0, the u
j
and v
j
are nonnegative integers, and
{0, 1].
Suppose we are turning a string into a shape, moving left to right, and we come
to a batch a
u
b
v
c of letters knowing that any corresponding underlying function f is
positive and decreasing. (The case of negative and increasing will be similar. These
are the only two possibilities following a c, and we have enough information about the
shape to know the local sign of any f that might underlie the shape we are building.)
There are four cases for the upcoming batch, depending on the parities of the upcoming
u and v.
If the batch has the form b
2i 1
a
2 j 1
c, then the corresponding next part of the shape
is I
i
I
0
I
j
m
. If it has the form b

2i 1
a
2 j
c, then the next part is I
i
m
0
J
j
M
. If it has
the form b
2i
a
2 j
c, then the next part is I
i
I
j
m
. If it has the form b

2i
a
2 j 1
c, then the
next part is I
i
m
j
M
. If there is no nal c (which can only happen if the current

batch is the last), then drop the nal shape item (an m or M). Note that i = 0 and
j = 0 are allowed in all cases.
Let s(w) = [S(w)[. Because the coding is a bijection, s(w) is the number of strings
of length w with no ab. It can be computed recursively: s(0) = 1, s(1) = 3, s(w) =
3s(w 1) s(w 2). The solution is
s(w) =

2w2
2w2
5
= F
2w2
,
where = (1
5 )/2 is the golden ratio and F

n
is the nth Fibonacci number.
Finally, the solution to the problem itself is
2
_
s(n 1) s(n 3) s(n 5) . . .
_
=
2
_
2n2
2n2
_
(1)
n1
5
5
=
2L
2n2
(1)
n1
5
5
,
in terms of Lucas numbers L
n
.
Editorial comment. As noted, this solution does not show that every shape of weight w
can be realized by a polynomial of degree w 1, another of degree w 3, and so on.
The proposer also omits this consideration, but to be fair the determination by degree
was added by the editors. Stong found that all of the sign charts enumerated can be
achieved by polynomials of the required degree, but his proof would have been too
long to present with all details lled in.
Also (partially) solved by R. Stong and the proposer.
Partitioning Segments Into Triples
11657 [2012, 608]. Proposed by Gregory Galperin, Eastern Illinois University,
Charleston, IL, and Yury Ionin, Central Michigan University, Mount Pleasant, MI.
Given a set V of n points in R
2
, no three of them collinear, let E be the set of
_
n
2
_
line
segments joining distinct elements of V.
(a) Prove that if n , 2 (mod 3), then E can be partitioned into triples in which the
length of each segment is smaller than the sum of the other two.
(b) Prove that if n 2 (mod 3) and e is an element of E, then E {e] can be so
partitioned.
Solution by the proposers. Part (a) is immediate for n = 3; we prove (a) for n = 4 and
(b) for n = 5 and proceed by induction on n. For x, y V, let [xy[ denote the length
of the segment xy. Since E is the edge set of the complete graph with vertex set V, for
x, y, z, w V we evoke terminology from graph theory by letting wxyz denote the
graph with vertex set {w, x, y, z] and edge set {wx, wy, wz] (a claw), letting .xyzw
denote the graph with vertex set {x, y, z, w] and edge set {xy, yz, zw] (a path), and
letting .xyz denote the graph with vertex set {x, y, z] and edge set {xy, xz, yz] (a
triangle).
When three segments satisfy the condition that the length of each is smaller than
the sum of the other two, we say that the triple is triangular; this holds for any triangle
.xyz and also for other triples of edges, including some paths and claws.
For n = 4, let V = {x, y, z, w]. Without loss of generality, let xy be a shortest
segment in E, and label z and w so that [xz[ [yz[ [xw[ [yw[. Using this in-
equality and the triangles containing zw, we have 2([xw[ [yw[) ([xw[ [yw[)
([xz[ [yz[) = ([xw[ [xz[) ([yw[ [yz[) > 2[zw[, so [xw[ [yw[ > [zw[. Also,
[xw[ [zw[ [xw[ [xy[ > [yw[ and [yw[ [zw[ [yw[ [xy[ > [xw[. Hence
the triangular triples .xyz and wxyz sufce.
For n = 5, let V = {x
1
, x
2
, y
1
, y
2
, y
3
] and E
/
= E {x
1
x
2
]. We partition E
/
into
three triangular triples. If . x
1
y
i
y
j
x
2
is triangular, where {i, j, k] = {1, 2, 3], then we
use {. x
1
y
i
y
j
x
2
, .x
1
y
j
y
k
, .x
2
y
i
y
k
]. If x
i
y
1
y
2
y
3
is triangular, where {i, j ] = {1, 2],
then we apply the case n = 4 to partition the segments formed by {x
j
, y
1
, y
2
, y
3
] into
two triangular triples and use x
i
y
1
y
2
y
3
as the third.
In the remaining case for n = 5, no path of the form . x
1
y
i
y
j
x
2
and no claw of the
form x
i
y
1
y
2
y
3
is triangular. If y
i
y
j
is a longest segment in E
/
, then [y
i
y
j
[ [x
1
y
i
[
[x
2
y
j
[ and [y
i
y
j
[ [x
1
y
j
[ [x
2
y
i
[; otherwise, . x
1
y
i
y
j
x
2
or . x
1
y
j
y
i
x
2
is triangular.
This yields the contradiction
2[y
i
y
j
[ ([x
1
y
i
[ [x
1
y
j
[) ([x
2
y
j
[ [x
2
y
i
[) > 2[y
i
y
j
[.
Therefore, any longest segment in E
/
has the form x
i
y
j
. Index the points so x
1
y
1
is a longest segment and [x
1
y
1
[ [x
1
y
2
[ [x
1
y
3
[. Now [x
1
y
1
[ [x
1
y
2
[ [x
1
y
3
[ (oth-
erwise x
1
y
1
y
2
y
3
is triangular), and [x
1
y
3
[ [y
3
y
1
[ > [x
1
y
1
[ (since .x
1
y
1
y
3
is tri-
angular), so [y
1
y
3
[ > [x
1
y
2
[. Since . x
1
y
1
y
3
x
2
is not triangular and x
1
y
1
is a longest
segment, [x
1
y
1
[ [y
1
y
3
[ [y
3
x
2
[. If y
1
y
3
is a longest segment in . x
1
y
3
y
1
x
2
, then
[y
1
y
3
[ [x
1
y
3
[ [y
1
x
2
[, yielding the contradiction [x
1
y
1
[ [x
1
y
3
[ [y
3
x
2
[ [x
2
y
1
[
(combining two triangles). Since [y
1
y
3
[ > [x
1
y
2
[ [x
1
y
3
[, we conclude that x
2
y
1
is
longest in . x
1
y
3
y
1
x
2
. Since . x
1
y
3
y
1
x
2
is not triangular, we obtain the contradiction
[x
2
y
1
[ [y
1
y
3
[ [y
3
x
1
[ > [x
1
y
1
[, nishing the case n = 5.
Now consider n 6. In the case n , 2 (mod 3), let xy be a shortest segment in
E, and let z be a point in V such that [xy[ [yz[ = min
w/ {x,y]
{[xw[ [yw[]. The
proof given for n = 4 shows that wxyz is triangular for all w / {x, y, z]. Hence
.xyz and the claws wxyz for w / {x, y, z] can be combined with a partition of the
pairs in V {x, y, z] (obtained by the induction hypothesis) to complete the desired
partition.
Finally, in the case n , 2 (mod 3), let e = uv. Let xy be a shortest segment de-
termined by the points in V {u, v], and let z be a point in V {x, y, u, v] such
that [xy[ [yz[ is the minimum of [xw yw[ over all w / {x, y, u, v] such that
xw, yw E. Again, every claw wxyz is triangular when w / {x, y, z, u, v]. Ap-
ply the case n = 5 for the pairs in {x, y, z, u, v], and apply the induction hypothesis to
the n 3 points in V {x, y, z], in both cases choosing e = uv as the deleted line seg-
ment. Combine the resulting partitions with the claws wxyz for w / {x, y, z, u, v].
Editorial comment. The result applies to any metric space, since the proof given needs
only the property that every actual triangle is triangular. The problem was printed with
an unfortunate typo in the rewording of the triangle inequality, with the word greater
appearing instead of smaller.
Also solved by R. Stong.
Pentagonal Series as Limit
11659 [2012, 608]. Proposed by Albert Stadler, Herrliberg, Switzerland. Let x be real
with 0 < x < 1, and consider the sequence a
n
) given by a
0
= 0, a
1
= 1, and, for
n > 1,
a
n
=
a
2
n1
xa
n2
(1 x)a
n1
.
Show that
lim
n
1
a
n
=
k=
(1)
k
x
k(3k1)/2
.
Solution by Greg Martin, University of British Columbia, Vancouver, BC, Canada.
Take the reciprocal of the recurrence relation and multiply both sides by a
n1
to get
a
n1
a
n
= x
a
n2
a
n1
(1 x), (n 2).
Let r
n
= a
n
/a
n1
. We now have r
0
= 0 and the recurrence r
n
= xr
n
1 x for n 1.
The solution to this is r
n
= 1 x
n
. Therefore
lim
n
1
a
n
= lim
n
n1
k=1
a
k
a
k1
= lim
n
n1
k=1
r
k
= lim
n
n1
k=1
(1 x
k
) =
k=1
(1 x
k
).
By Eulers pentagonal number theorem, for [x[ < 1 this innite product is equal to
k=
(1)
k
x
k(3k1)/2
.
Also solved by M. Bataille (France), D. Beckwith, P. Bracken, B. Bradie, N. Caro (Brazil), R. Chapman (U. K.),
H. Chen, J. E. Cooper III, P. P. D alyay (Hungary), E. S. Eyeson, S. M. Gagola Jr., C. Georghiou (Greece), O.
Geupel (Germany), A. Habil (Syria), E. A. Herman, R. Howard, M. Kim (Korea), O. Kouba (Syria), W. C.
Lang, J. Li, J. C. Linders (Netherlands), J. H. Lindsey II, O. P. Lossers (Netherlands), R. Martin (Germany),
T. L. McCoy, R. Nandan, M. Omarjee (France), P. Perfetti (Italy), C. R. Pranesachar (India), M. A. Prasad
(India), D. N. Sanyasi (India), R. K. Schwartz, C. R. Selvaraj & S. Selvaraj, N. C. Singer, A. Stenger, R. Stong,
R. Tauraso (Italy), D. B. Tyler, E. I. Verriest, J. Vinuesa (Spain), T. Viteam (Chile), M. Vowe (Switzerland), M.
Wildon (U. K.), GCHQ Problem Solving Group (U. K.), TCDmath Problem Group (Ireland), and the proposer.
An Unusual Differential Equation
11660 [2012, 609]. Proposed by Stefano Siboni, University of Trento, Trento, Italy.
Consider the following differential equation: s
//
(t ) = s(t ) s(t )
2
sgn(s
/
(t )), where
sgn(u) denotes the sign of u. Show that if s(0) = a and s
/
(0) = b with ab ,= 0, then
(s, s
/
) tends to (0, 0) with
s
2
s
/2
C/t as t , for some C > 0.
Solution by GCHQ Problem Solving Group, Cheltenham, UK. The claim does not
always hold. Suppose that s(t ) is the position of a particle at time t , and denote s
/
(t )
by v(t ). If we let s
1
(t ) denote s when s(0) = a and s
/
(0) = b and s
2
(t ) denote s when
s(0) = a and s
/
(0) = b, then s
2
(t ) = s
1
(t ) for all t and the claim holds for s
1
if
and only if it holds for s
2
. Without loss of generality, we may assume a 0. Whenever
v > 0, the equation is v
dv
ds
= s s
2
. Hence integration yields
1
2
(v
2
V
2
) =
1
2
(S
2
s
2
)
1
3
(S
3
s
3
),
where (s, v) = (S, V) at some point of the motion. Whenever v < 0, the equation is
v
dv
ds
= s s
2
, so
1
2
(v
2
U
2
) =
1
2
(R
2
s
2
)
1
3
(s
3
R
3
),
where (s, v) = (R, U) at some point of the motion. Thus, if a, b > 0 such that 3a
2
3b
2
2a
3
5, then v
2
= a
2
b
2
s
2
2
3
(a
3
s
3
), so the particle moves to the right
until v = 0 or s
2
2
3
s
3
= a
2
b
2
2
3
a
3
5
3
.
Letting denote the smallest solution, we have 1 and
2
> . When s reaches
, the particle stops instantaneously and reverses its direction. As the particle attempts
to go left, sgn(s
/
(t )) ips; since
2
, the net acceleration then acts to the right and
motion is instantaneously reversed again. Therefore (s, v) = (, 0) for all future time.
It is not true that s
2
s
/2
C/t as t , but it is trivially true that (s )
2
s
/2
C/t . Note also s
//
(t ) , 0 as t , even when = 1. Similarly, if a > 0 and
b < 0 with 3a
2
3b
2
2a
3
0, then v
2
= a
2
b
2
s
2
2
3
(s
3
a
3
). The particle
moves to the left until s
2
2
3
s
3
= a
2
b
2
2
3
a
3
5
3
. Letting denote the largest
negative solution, we have 1. Therefore, once s reaches , we have (s, v) =
(, 0) for all future time.
Now we show that the claim holds for the opposite inequality, and that C = 10
will sufce. Consider the case in which a 0, b > 0, and 3a
2
2a
3
3b
2
< 5. The
particle moves to the right until s = , where
2
2
3
3
= a
2
b
2
2
3
a
3
. Since < 1,
the particle then accelerates left with motion governed by v
2
=
2
s
2
2
3
(s
3
3
),
stopping instantaneously when 0 = 3(
2
s
2
) 2(s
3

3
) = ( s)(3( s)
2(
2
s s
2
)) = ( s) f (s), say. With f (s) dened this way, we have f (0) =
(3 2) > 0 and f () = 6(1 ) > 0, so f (s) has one negative root and one
root larger than . The particle will stop when s is the smaller of these roots, which
we denote by , where
=
1
4
_
3 2
_
(3 2)(3 6)
_
=
1
4
3 2
_
3 6
3 2
_
.
(1)
Since f () = 2
2
< 0, we have > or [[ < [[. The particle now accel-
erates to the right, obeying v
2
=
2
s
2
2
3
(
3
s
3
), and stops when 0 = 3(
2
s
2
) 2(
3
s
3
) = ( s)(3( s) 2(
2
s s
2
)) = ( s)g(s), say. With
g(s) dened this way, we have g() = 6(1 ) < 0 and g(0) = (3 2) < 0.
Thus g(s) has one positive root and one root smaller than . The particle will stop
when s is the larger of these roots, which we denote . We have
=
1
4
_
(3 2)
_
(3 2)(3 6)
_
=
1
4
_
3 2
_
_
3 6
_
3 2
_
.
Since g(s) = 2
2
> 0, we know that s
< or [s
[ < [[. The sequence of

distances from the origin to the stationary points is decreasing and bounded below by
0 and must converge. If it converges to a positive limit P, then by (1) it would have
to satisfy 4P = 3 2P
(3 2P)(3 6P), which yields P = 0. Hence s 0

and also v 0. Dene T
0
to be the time that the particle comes to rest following
its rst rightward motion. Dene T
1
, T
2
, T
3
, . . . to be the times for subsequent rests,
with odd (even) indices following leftward (rightward) motion. Let s = S
j
at time T
j
and let U
j
= [S
j
[. Consider the motion of the particle during the cycle from s = S
2 j
through s = S
2 j 1
to s = S
2 j 2
. We have the following bounds: v < 0 implies v
2
=
S
2
2 j
s
2
2
3
(s
3
S
3
2 j
), which implies

s
2
s
/
2
U
2 j
; if v > 0 then v
2
= S
2
2 j 1

s
2
2
3
(S
3
2 j 1
s
3
), which implies
s
2
s
/
2
U
2 j 1
.
We now show that
s
2
s
/
2
C/t in four steps (a)(d).
(a) T
n
< T
0
n. If v < 0, then s S
2 j 1
. Hence v
2
= S
2
2 j

2
3
S
3
2 j
s
2
2
3
s
3
S
2
2 j

2
3
S
3
2 j
s
2
(1
2
3
S
2 j 1
) so the velocity is larger and travel time less than when
travelling under
ds
dt
=
_
S
2
2 j

2
3
S
3
2 j
s
2
(1
2
3
S
2 j 1
). This implies
t =
_
ds
_
S
2
2 j

2
3
S
3
2 j
s
2
(1
2
3
S
2 j 1
)
=
1
_
1
2
3
S
2 j 1
arcsin
_
_
s
_
1
2
3
S
2 j 1
S
2
2 j

2
3
S
3
2 j
_
_
.
Therefore,
T
2 j 1
T
2 j

1
_
1
2
3
S
2 j 1
_
2

_
2
__
=

_
1
2
3
U
2 j 1
< .
When v > 0, we have s S
2 j 2
, so v
2
= S
2
2 j 1

2
3
S
2
2 j 1
s
2
2
3
s
3
S
2
2 j 1

2
3
S
3
2 j 1
s
2
(1
2
3
S
2 j 1
), which yields T
2 j 2
T
2 j 1
< .
(b) U
j 1
< U
j

1
3
U
2
j
. From above, we have 0 < U
j
3/2 and
U
j 1
=
1
4
_
3 2U
j
(
_
3 6U
j

_
3 2U
j
). Now 0 < 9 6U
j
2U
2
j
, so 0 <
8u
2
j

16
3
U
3
j

16
9
U
4
j
. Together with our range for U
j
, this implies 0 < 9 12U
j

12U
2
j
< (3 2U
j

4
3
U
2
j
)
2
. Taking the square root now yields
_
(3 6U
j
)(3 2U
j
)
< 3 2U
j

4
3
U
2
j
= (3 2U
j
) (4U
j

4
3
U
2
j
). This implies (b).
(c) If n 0, then 1/U
n
1/U
0

n
3
. Clearly this holds for n = 0, so assume it
holds up to some n = N. By factoring, we have
_
N
3
U
0
_
2
1 <
_
N
3
U
0
_
2
=
N 3/U
0
1
(N 3/U
0
)
2
<
1
N 3/U
0
1
.
The expression x
1
3
x
2
is increasing on the range of interest, so
U
N1
< U
N

1
3
U
2
N
<
3
N 3/U
0
3
(N 3/U
0
)
2
=
3(N 3/U
0
1)
(N 3/U
0
)
2
<
1
N 3/U
0
1
.
The result holds for n = N 1 and the induction is complete. Equality holds only
when n = 0.
(d)
s
2
s
/
2
C/t . For T
n
t T
n1
, we have
1
s
2
s
/
2

1
U
n
1
U
0
n
3
>
1
U
0
T
n
T
0
3
1
U
0
(t ) T
0
3
>
t
C
for C = 10 and for sufciently large t . Any choice of C with C > 3 will work.
If a 0 and b 0, then we obtain the same asymptotic result, and C = 10 again
sufces.
Also solved by E. A. Herman, O. P. Lossers (Netherlands), R. Stong, D. B. Tyler, E. I. Verriest, TCDmath
Problem Group (Ireland), and the proposer.
More Triangle Inequalities
11664 [2012, 699]. Proposed by Cosmin Pohoata, Princeton University, Princeton, NJ,
and Darij Grinberg, Massachusetts Institute of Technology, Cambridge, MA. Let a, b,
and c be the side lengths of a triangle. Let s denote the semiperimeter, r the inradius,
and R the circumradius of that triangle. Let a
/
= s a, b
/
= s b, and c
/
= s c.
(a) Prove that
ar
R

b
/
c
/
.
(b) Prove that
r(a b c)
R
_
1
R 2r
4R r
_
2
_
b
/
c
/
a

c
/
a
/
b

a
/
b
/
c
_
.
Solution for (a) by Alper Ercan, Istanbul, Turkey. Recall these formulas, involving the
area of the triangle:
= rs =
abc
4R
,
2
= sa
/
b
/
c
/
.
The inequality to be proved becomes 4a
/
b
/
c
/
(a
/
b
/
)(a
/
c
/
), because bc =
(a
/
b
/
)(a
/
c
/
). By the AMGM inequality, 2
a
/
b
/
a
/
b
/
and 2
a
/
c
/
a
/
c
/
.
The desired inequality follows.
Solution for (b) by Paolo Perfetti, Universit` a degli Studi di Roma Tor Vergata, Rome,
Italy. Recall
R =
abc
(a b c)(a b c)(b c a)(c a b)

,
r =
abc
4Rs
=
(a b c)(a b c)(b c a)(c a b)

s
.
For convenience, write x = a
/
, y = b
/
, z = c
/
, and D = (x y)(y z)(z x). The
inequality to be proved becomes
8(x y z)xyz
D
5D 4xyz
4D 4xyz
2
_
xy
x y

yz
y z

zx
z x
_
.
Clearing denominators, we see that this is equivalent to
(xy)
3
(yz)
3
(zx)
3
3(xyz)
2
x
3
y
2
z y
3
z
2
x z
3
x
2
y,
which, with = xy, = yz, and = zx, becomes Schurs third-degree inequality
3
3
2

2

2
.
Editorial comment. Some solvers noted that part (a) is related to Problem 11306, this
Monthly 116 (2009) 8889. Part (b) was also solved using various other geometric
inequalities, such as Koois inequality.
Also solved by R. Boukharfane (Canada), M. Can, R. Chapman (U. K.), P. P. D alyay (Hungary), J. Fabrykowski
& T. Smotzer, O. Geupel (Germany), M. Goldenberg & M. Kaplan, W. Janous (Austria), O. Kouba (Syria),
K.-W. Lau (China), P. N uesch (Switzerland), V. Pambuccian, N. Stanciu & Z. Zvonaru (Romania), R. Stong,
M. Vowe (Switzerland), J. Zacharias, GCHQ Problem Solving Group (U. K.), and the proposer.
Inequalities for Inner Product Space
11667 [2012, 700]. Proposed by Cezar Lupu, University of Pittsburgh, Pittsburgh, PA,
and Dan Schwarz, Softwin Co., Bucharest, Romania. Let f , g, and h be elements of
an inner product space over R, with f, g) = 0.
(a) Show that
f, f )g, g)h, h)
2
4g, h)
2
h, f )
2
.
(b) Show that
( f, f )h, h))h, f )
2
(g, g)h, h))g, h)
2
4g, h)
2
h, f )
2
.
Solution I by P al P eter D alyay, Szeged, Hungary. If f , g, or h is zero, then the inequal-
ities clearly hold. Since f, g) = 0, note that
e =
h, f )
f, f )
f
h, g)
g, g)
g
is the orthogonal projection of h onto the space spanned by { f, g], and therefore
|h|
2
= |e|
2
|h e|
2
. Thus, |h|
2
|e|
2
= h, f )
2
/| f |
2
h, g)
2
/|g|
2
2h, f )h, g)/(| f | |g|). Squaring both sides gives (a). By the AMGM inequality,
f, f )h,h)h, f )
2
g,g)h,h)g,h)
2
2
_
f, f )g,g)(h,h))
2
(h, f ))
2
(g,h))
2
= 2
_
_
f, f )g,g)h,h)
_
h, f )g,h).
By (a), the bracketed term is at least 2[g, h)h, f )[, so (b) follows.
Solution II by Paolo Perfetti, Dipartimento di Matematica, Universit` a Degli Studi di
Roma, Rome, Italy. If f , g, or h is zero, then the result clearly holds, so we may dene
F = f /| f |, G = g/|g|, and H = h/|h|. Now (a) reads h, h)
2
4G, h)
2
F, h)
2
.
By Bessels inequality, |h|
2
G, h)
2
F, h)
2
2G, h)F, h), and (a) follows
by squaring this result. Similarly, part (b) reads | f |
2
H, f )
2
|g|
2
H, g)
2
4H, f )
2
H, g)
2
. Again by AMGM, | f |
2
H, f )
2
|g|
2
H, g)
2
2| f | |g|
[H, f )H, g)[, and it remains to show | f | |g| 2[H, f )H, g)[. When we mul-
tiply both sides by
|h|
| f ||g|
, this becomes |h| 2[h, F)h, G)[, which is essentially
(a).
Also solved by K. Andersen (Canada), G. Apostolopoulos (Greece), R. Boukharfane (Canada), P. Bracken, R.
Chapman (U. K.), A. Ercan (Turkey), D. Fleischman, C. Georghiou (Greece), O. Geupel (Germany), K. Hanes,
E. A. Herman, F. Holland, B. Karaivanov, O. Kouba (Syria), J. H. Lindsey II, O. P. Lossers (Netherlands), M.
Omarjee (France), M. A. Prasad (India), N. C. Singer, R. Stong, R. Tauraso (Italy), T. Trif (Romania), D. B.
Tyler, E. I. Verriest, J. Vinuesa (Spain), R. Wyant & T. Smotzer, GCHQ Problem Solving Group (U. K.), and
the proposers.
REVIEWS
Edited by Jeffrey Nunemacher
Mathematics and Computer Science, Ohio Wesleyan University, Delaware, OH 43015
The Outer Limits of Reason. Noson Yanofsky. MIT Press, Cambridge, MA and London, Eng-
land, 2013, xiv + 403 pp., ISBN 978-0-262-01935-4, $29.95.
Reviewed by Michael Barr
This beautifully-written book explores a variety of topics that are either impossible to
resolve or unfeasible (and likely permanently so). It is written in such a way that most
(although probably not all) of the material will be accessible to non-mathematicians.
Both my wife (who studied no mathematics at the college level) and my daughter
(whose mathematical training stopped with calculus) read early versions of this book
and agree with that assessment. (Full disclosure: both of them, along with me, are
thanked in the Acknowledgments). However, there is much in here that was new and
interesting to me, so its appeal is also not limited to non-mathematicians.
The best way to appreciate the breadth of this book is to give the chapter headings,
followed by a brief discussion of what is in each.
Chapter 1, Introduction, is an overview of the book.
Chapter 2, Language Paradoxes, discusses problems ranging from the well-known
Cretans, the paradoxes of self reference (This sentence is false.), and the paradoxes
of naming numbers (The rst number that cannot be described with fewer than thir-
teen words). No resolution is suggested for these paradoxes, although the author does
point out that there is no village in the real world in which the barber shaves everyone
who doesnt shave himself. There are no logical contradictions in the real world.
Chapter 3, Philosophical Conundrums, starts with the identity problem. Assuming
the ship of Theseus has had all its planks changed, is it the same ship? If not, when
did it cease? What if someone saved all the discarded planks and built a new ship
with it? Would the old one or the new one be most properly described as the ship
of Theseus? Are you the same person you were a year ago? Twenty years ago? You
have probably changed quite a bit. If not, when did the twenty-year-ago person change
to the present you? Further questions in this chapter include a very nice discussion
of Zenos paradoxes. Then the chapter veers into such questions as who is the tallest
midget or the hairiest bald man and the Monty Hall problem. It ends with an interesting
discussion of He knows that she knows that he knows that she knows, . . . . How deep
is the recursion?
Chapter 4, Innity Puzzles, covers ground, such as Hilberts hotel, that is thoroughly
familiar to mathematicians but not to other readers. It includes the Russell paradox and
a discussion of ZF(C) and the continuum hypothesis.
Chapter 5, Computing Complexities, is mostly a discussion of P versus NP and NP
complete. One criticism is that it leaves the impression that all NP problems are NP
complete. The author does not, for example, mention that the factorization of num-
bers is not known to be NP complete. There is a nice illustration showing how much
faster exponential functions grow compared to polynomial functions. There is a brief
discussion of quantum computing (Yanofsky is the coauthor of a book about quantum
computing) that takes a very pessimistic view of the likelihood that any NP complete
function will ever succumb to a quantum computer.
Chapter 6, Computing Impossibilities, begins with a brief discussion of pragmatic
impossibilities (a computer cannot fall in love, or decide that a poem or a ower is
beautiful), but is really about fundamental issues such as the halting problem. First,
he introduces a kind of pseudocode that is essentially BASIC. Looking at it reminds
me that, all the computer snobs notwithstanding, BASIC is really a very nice language
for beginners. You can read a simple program with (almost) no explanation. Even so,
he gives a simple program in which it is nearly impossible to follow through nested
loops and decide whether it halts. If I had a BASIC interpreter, I would enter it and
see. This leads directly to a discussion of the halting problem. Since this impossibility
is shown by having a program that can reason about itself, he relates it to the linguistic
paradoxes of Chapter 2, in which the problems all seem to be related to self-reference.
One thing mentioned was new to me, that even if you have an oracle to solve the
halting problem, Turing showed that there are problems beyond that that still could
not be solved. In fact, there is a hierarchy of such levels of impossibility. There is even
a theorem that says that there are three levels of impossibility A, B, C such that for an
oracle of level A, P
A
= NP
A
, while for an oracle of level B, P
B
= NP
B
, and for C, the
question P
C
= NP
C
is formally undecidable. While this doesnt prove that P = NP is
undecidable, it surely suggests that. The chapter ends with an inconclusive discussion
of whether the human mind can, as Kurt G odel and Roger Penrose appeared to believe,
transcend what computers can do. Or conversely, can humans keep up with computers,
having lost their mastery of chess and Jeopardy?
Chapter 7, Scientic Limitations, begins with a discussion of chaos theory and sen-
sitive dependence on initial conditions (the buttery effect). One of the important ques-
tions he raises here is the essential impossibility of solving the 3-body (or n-body)
problem. Yes, solutions exist, but they cannot predict whether the solar system is sta-
ble in the long run. Thus, this is a limitation on what we can know. He mentions
the Laplace computer that could, in principle, calculate the future state of the uni-
verse by starting with the current position and momentum of every particle. Laplace,
of course, did not know of quantum mechanics. Nonetheless, the author shows that a
Laplace computer is logically impossible. More precisely, he shows that having two
Laplace computers in the same universe is logically impossible by programming them
to take inconsistent actions in response to each other. This is followed by a long sec-
tion giving the essentials of quantum mechanics and uncertainty. Since that is already
a summary, I will not try to summarize it further. It is worth reading even if you al-
ready know a good deal about quantum mechanics, as you may still nd surprises.
The chapter ends with a discussion of relativity theory, both the special and the gen-
eral theory. Although the discussion is interesting, it is not clear what it has to with
unknowability.
Chapter 8, Metascientic Perplexities, begins with a long discussion of scientic
induction. How can we know that All swans are white unless we have seen all
swans? For me, that question was answered in the simplest possible way, by ac-
tually seeing some black swans. (They are native to Australia, but I saw my rst
ones swimming in the Z urichsee). In the end, he gives a probabilistic interpretation
to distinguish between the logically equivalent claims that all crows are black and
that all non-black things are non-crows. There are just so many fewer crows than
non-black things that each instance of a black crow has more persuasive power than
each instance of a non-black thing. That is correct, but somehow I remain convinced
AugustSeptember 2014] REVIEWS 659
that crow is a natural class (whatever that means) while non-black thing is not.
1
This is followed by a long, interesting section on the unreasonable effectiveness of
mathematics. Various explanations, none entirely satisfactory, are explored. Some,
such as the assertion that mathematics follows physics are clearly wrong (Rieman-
nian geometry, matrix theory, and group theory all preceded their applications in
physics by a half-century or more). The chapter continues with a fascinating discus-
sion of the fact that the physical constants of the universe seem to be just perfect
for life and even intelligent life to form. This leads to the anthropic principle that
states, essentially, that if the laws werent what they are, we wouldnt be here to
observe them. This leads to further unknowable questions. There is a serious error
involving the cosmological constant, which does not have to be correct to the 120th
decimal place in order for human life to have formed. It could perfectly well be
0 (as was widely assumed until about 15 years ago) without impinging on human
existence.
Chapter 9, Mathematical Obstructions, begins with familiar observations such as
the Greek discovery that

2 is irrational, including a pretty geometric proof that
makes no mention of divisibility. He then turns to Galois theory and the impossibility
of giving solutions by radicals. Then he considers a couple of problems (can a set of
tiles ll space; can you discover if a polynomial in several variables has an integer
solution?) which can encode the halting problem and therefore are undecidable. He
ends the chapter by describing G odels incompleteness theorem in some detail.
Chapter 10, Beyond Reason, begins by summing up the four main categories of lim-
itations: physical limitations, mental limitations, practical limitations, and limitations
based on failed intuition (such as generalizing familiar facts from nite to innite).
This is followed by a discussion of what reason is. One mark of reason is that it not
lead to a contradiction. For example, why, when you are ill, is it reasonable to measure
your temperature, but not to consult your horoscope? He then discusses ideas that used
to be considered reasonable, but have been left behind (phlogiston, ether, . . . ) and con-
versely what was long considered nonsense and turned out to be true (the germ theory
of disease, negative and imaginary numbers, . . . ). This last motivates me to ask a ques-
tion. Classical physics was formulated entirely with real numbers. Could quantum me-
chanics have even been formulated without complex numbers? What could it possibly
look like? Isnt the fact that complex numbers were created long before quantum me-
chanics required it, another example of the unreasonable effectiveness of mathematics?
What is the intended audience for this book? Any mathematician will nd it in-
teresting. But, as illustrated by my wife and daughter, most interested laymen would
enjoy at least parts of it.
Dept. Math. and Stats., McGill University, Montreal, QC, H3A2K6
barr@math.mcgill.ca
1
However, I discovered endnote 2 on page 367. The trouble is that it is an endnote, not a footnote and no
one looks at endnotes, while most people look at footnotesjust as you are looking at this. So he does know
about black swans and even rarer white ravens.
Illustrated Special Relativity through Its Paradoxes:
A Fusion of Linear Algebra, Graphics, and Reality
By John dePillis and Jos Wudka
Spectrum Series
The text illustrates and resolves several apparent
paradoxes of Special Relativity including the twin
paradox and train-and-tunnel paradox. Assuming
a minimum of technical prerequisites the authors
introduce inertial frames and use them to explain
a variety of phenomena: the nature of simultaneity,
the proper way to add velocities, and why faster-
than-light travel is impossible. Most of these
explanations are contained in the resolution of
apparent paradoxes, including some lesser-known
ones: the pea-shooter paradox, the bug-and-rivet
paradox, and the accommodating universe paradox.
The explanation of time and length contraction is
especially clear and illuminating.
The roots of Einsteins work in Maxwells lead the authors to devote several
chapters to an exposition of Maxwells equations. The authors establish that
those equations predict a frame-independent speed for the propagation of
electromagnetic radiation, a speed that equals that of light. Several chapters are
devoted to experiments of Roemer(SYMBOL!), Fizeau, and de Sitter to measure the
speed of light and the Michelson-Morley experiment abolishing the aether.
Throughout the exposition is thorough, but not overly technical, and often
illustrated by cartoons. The volume might be suitable for a one-semester general-
education introduction to Special Relativity. It is especially well-suited to self-study
by interested laypersons or use as a supplement to a more traditional text.
eISBN 978-1-61444-517-3
2013, 478 pp.
Catalog Code: ISR
PDF Price: $33.00
New in the
MAA eBooks Store
B
S
T
p
p
a
i
a
t
t
e
a
o
p
T
Illustrated
Special
Relativity
through its
Paradoxes
IIIIIIIIIIIIIIIIllllll
Spectrum
John dePillis & Jos Wudka
Illustrations and animations by John dePillis
To order, visit www.maa.org/ebooks/ISR.
1529 Eighteenth St., NW
Washington, DC 20036
Willoughbys essay is a gem. It should be in the hands of every young teacher. I wish that I had read it
many years ago. I have no doubt that many of his observations and the information he imparts will re-
main with me for a while. I certainly hope so. A collection of reminiscences from other teachers with their
valuable insights and experiences (who could write with such expertise as he does) would make a fne
addition to the education literature. James Tattersall, Providence College
A thought provoking
must read
for all teachers.
New
in the
MAA eBooks Store
ebook: $11.00
Print on demand (paperbound): $18.00
To order go to www.maa.org/ebooks/TTT
Textbooks, Testing, Training
How We Discourage Tinking
Stephen S. Willoughby

Amm Vol121n07

Diunggah oleh

Informasi Dokumen

Hak Cipta

Format Tersedia

Bagikan dokumen Ini

Bagikan atau Tanam Dokumen

Opsi Berbagi

Apakah menurut Anda dokumen ini bermanfaat?

Apakah konten ini tidak pantas?

Hak Cipta:

Format Tersedia

Amm Vol121n07

Diunggah oleh

Hak Cipta:

Format Tersedia

New

Figure 2. The graphs of some tropical polynomials

Figure 3. Some tropical curves

those dual to a vertex of C

those dual to a vertex of C

those dual to an intersection point of C

. Our choice must satisfy the following rules:

for each vertex v of C adjacent to the edges e

(x, y) when t tends

consists of a single point p, then

is given by the tropical polynomial 1 x + y + 0, so that the

is as in Figure 9(a) with L in blue and L

in red. Then the stable

, that is to say the point (1, 0). Note that

in the modied plane

as dened in Section 2. The

E (brugalle@math.cnrs.fr) got his Ph.D. in Rennes in 2004. After holding a position at

A} denote the corresponding quotient set (sometimes called

) for some integer . If j < k, then

1 belongs to A, then setting t = s yields

1 belongs to B, we now let t = s. Proceeding

for some > j so that

for some > j , from which it follows that

2) 0.5 0.3536. In other

[base H] H-Tthis gives a Cl-Wh sequence, and victory is awarded to Player 1;

A received his D.Sc. degree from Universidade Federal do Rio de

2 from each other have different colors. Then at least 5

2. We will show that the

2}. By carefully analyzing Figure 4, however, we can make the appro-

2 from each other have

is an image of E under a rigid motion of the plane (i.e., whenever E

is congruent to E), then the sum of f over the elements of E

is zero. Does it follow

evidently is in W(A). So, we have the inequality (1).

for all j , and hence

angle cannot be trisected.

/3 can be constructed and thus can be trisected with a compass and

cannot be trisected. Since the angle 45

angle can be trisected with a compass and straightedge, it

the acute angles in the triangle can be trisected if and only if a

the triangle is Pythagorean if and only if a

b)/Q( c) with n N, P, Q Q[X

are triangles and therefore linearly equivalent. Applying

= g(T) and the

is the disjoint union of ve regions: two triangles D

+ (x, y) for some vector (x, y). As

contains an odd number of points from each color.

. Because the area

(, ) for some small > 0. The set of points

and S consists of the rectangle whose vertices are (t, 1

is the disjoint union of ve regions

and assume is chosen so that N is an integer. For every i = 1, . . . , N, let

. The black area in Z

be the vertical line that

does not meet any trapezoid not supported by .

, then it is still true that every trapezoid

:= {y L | (y) = y}. Then M

is a eld such that K M

point, which, together with compactness,

. These are Strangs four subspaces, whose interesting proper-

. We will see that the rows of L are a basis for

. Since J is nonsingular, there is a vector x such that

, thus implying y R(L).

of positive integers, and let g be the