Anda di halaman 1dari 18

Barrett, the Honors College at Arizona State University

Tiling 𝒎×𝒎 Parallelogram Boards with Polyominoes


Derek Fermaint, Computer Science, B.S., Computational Mathematics, B.S., Statistics, minor

Dr. Kawski, Ph.D., President’s Professor, School of Mathematical and Statistical Sciences at Arizona State University

Abstract
Mathematical induction is a powerful technique used to
prove a statement about a well-ordered set. Induction uses
the particular and rule of inference to prove validity of
general statements. This paper extensively uses this method
to determine which 𝑚×𝑚 (m “by” m) parallelogram
boards can be tiled by L-shaped trominoes.

I. Introduction
The objective of this work is proving in what cases an 𝑚×𝑚 parallelogram board
can be tiled with bent trominoes. Polyominoes were invented by Solomon Golomb, then
a student at Harvard University, in 1954. A polyomino is a "rook"-connected set of equal
squares. In Computer Science, the kind of connectedness where neighboring squares are
required to share an edge is also known as the 4-connectedness. If two squares that share
a vertex are also considered neighbors, its known as the 8-connectedness. Trominoes are
polyominoes that consist of three squares. There are two kinds of trominoes: a 3×1
rectangle, also known as a “straight” tromino, and an L-shaped piece, also known as a
“bent” tromino. Dominoes are polyominoes of two squares, forming a straight 2×1 piece.
The focus of this paper is ultimately to answer which 𝑚×𝑚 parallelogram boards, in
which one square piece has been removed, can be tiled with L-shaped. Each
parallelogram board to be considered will be constructed with equal square pieces. Tiling
in this context consists of overlaying a board with polyomino “tiles” such that each
square on the grid of the board is covered. During mathematical induction, a base case is
generated, proving the first solution that exists. The second step is the induction
hypothesis, where it is assumed if the statement is true for any arbitrary natural number,
then it must be true for the succeeding natural number as well. Induction relies on the rule
of inference to establish validity. First it will be demonstrated whether a 2×𝑚 board can
be tiled with 2×1 dominoes. Then, to see whether a 5×5 board can be tiled with
Fermaint

dominoes, followed by removing a square piece of the 5×5 board. Further, straight
trominoes are used to tile an 8×8 board. Golomb’s theorem is considered to prove if
2) ×2) boards can be tiled with bent trominoes. Finally, with the aggregation of the
findings, the central question of interest is answered: whether a 𝑚×𝑚 parallelogram
board can be tiled using bent trominoes. Some methods used are parity and coloring
proofs, however the focus of this paper is on induction proofs.

II. 𝟐 × 𝒎 parallelogram board tiled with dominoes

Figure 1 Figure 2

The first case study considered is whether a 2×𝑚 parallelogram board


constructed with square pieces can be tiled with 2×1 dominos. This will be the
first deploy of the induction technique. Consider the cases of where 𝑚 equals 1,
2, 3, and 4 respectively…

𝑚=1 𝑚=2 𝑚=3 𝑚=4

Given the pattern, since a domino is equivalent to the size of each new
added set of squares, a new domino can always tile the board regardless of the
value of 𝑚. Thus, the following conjecture can be made…

𝑪𝒐𝒏𝒋𝒆𝒄𝒕𝒖𝒓𝒆:

A 2×𝑚 square board can be tiled with 2𝑥1 dominoes for any 𝑚.

𝑰𝒏𝒅𝒖𝒄𝒕𝒊𝒐𝒏 𝒑𝒓𝒐𝒐𝒇 𝒇𝒐𝒓 𝒕𝒊𝒍𝒊𝒏𝒈 𝒂𝒏𝒚 𝟐 × 𝒎 𝒃𝒐𝒂𝒓𝒅 𝒘𝒊𝒕𝒉 𝒅𝒐𝒎𝒊𝒏𝒐𝒆𝒔:

Let 𝑆(2, 𝑚) be a 2×𝑚 parallelogram board where 𝑚 𝜖 ℕ1.

𝑩𝒂𝒔𝒆 𝑪𝒂𝒔𝒆:

Let 𝑚 = 1. Then the board 𝑆 2, 𝑚 = 𝑆(2,1) ( fig.1 ) is a 2×1 square


board. Then, 𝑆(2,1) can be tiled completely by a 2×1 domino ( fig. 2 ), as every


The convention for ℕ is the set of all positive integers, so ℕ = {1,2,3,4 … }. This is a distinction from the 19
1 th

century convention where zero is included in the set.

2
Fermaint

square on the domino covers each square on 𝑆(2, 𝑚) completely. This is true
whether the domino is inverted or not.

Vertical Domino Inverted Vertical Domino

𝑰𝒏𝒅𝒖𝒄𝒕𝒊𝒐𝒏 𝑯𝒚𝒑𝒐𝒕𝒉𝒆𝒔𝒊𝒔:

Assume any parallelogram board 𝑆(2, 𝑚) can be tiled by 2𝑥1 dominos for
all 𝑚 = 𝑘, where 𝑘 𝜖 ℕ is arbitrary but fixed, then 𝑚 = 𝑘 + 1 must also allow
𝑆(2, 𝑚) to be tiled.

𝑰𝒏𝒅𝒖𝒄𝒕𝒊𝒐𝒏 𝑺𝒕𝒆𝒑:

𝑆(2, 𝑘) 𝑆(2, 𝑘 + 1)

Since this is true for 𝑚 = 𝑘 + 1, then the assumption for 𝑚 = 𝑘 implies


𝑆(2, 𝑚) can be tiled by 2×1 dominoes for all 𝑚.

III. 𝟓×𝟓 parallelogram board tiled with dominoes


It has been proven that a 𝑆(2, 𝑚) = 2×𝑚 parallelogram board 𝑆 can be
tiled by 2×1 dominoes for any 𝑚 𝜖 ℕ, to expand this problem consider a 5×5
board 𝑆. Since the 𝑆(2, 𝑚) boards ensured that for any 𝑚, 𝑆(2, 𝑚) would be
divisible by 2, in other words, always even and since dominoes are an object
consisting of 2 squares, it follows that the area of the board must be even to be
tillable by dominoes. In this case, the 𝑆(5,5) board is unable to be tiled as
𝑆(5,5)UVWU = 25 is an odd natural number. 25 divided by 2 leaves a remainder of
1, so there should be a remaining uncovered square, which is seen as the vacant
square in the bottom right in fig 3.

3
Fermaint

Figure 3

To satisfy the condition, the bottom right square will need to be removed.
Once removed, the area of the board is 24 squares, which means the area of the
board is divisible by the area of the dominoes, or there exist some 𝑐 ∈ ℕ such that
24 = 2𝑐. This implies that there exist some square such that 𝑆 can be tiled by
dominoes, as shown in fig 4.

Figure 4

In this example, it is true for when the square is removed from 𝑆(Z[,Z\) . The
inquiry that follows is if S is tillable regardless of the location the square piece is
removed. To determine this, tile tests were performed. That is, one square piece
was removed, then 𝑆 was tiled by dominoes to determine if it could be completely
tiled. If so, the square location was marked black. If not, the square location was
marked gray. Each square piece at every cell location on the board was tested
which resulted in fig. 5.

4
Fermaint

0,0 0,1 0,2 0,3 0,4

1,0 1,1 1,2 1,3 1,4

2,0 2,1 2,2 2,3 2,4

3,0 3,1 3,2 3,3 3,4

4,0 4,1 4,2 4,3 4,4

Figure 5

Notice in fig. 5 that each square is indexed beginning at (0,0) at 𝑆_[,_\ and
ending in (4,4) at 𝑆`[,`\ .

A formal definition of parity is required to go further. A parity of a


number is the property of whether it is even or odd. That is, given 𝑛 ∈ 𝕫c , if
𝑛 𝑚𝑜𝑑 2 = 0 then 𝑛 is even. If 𝑛 𝑚𝑜𝑑 2 ≠ 0 then 𝑛 is odd.

The necessary mathematical structures are now constructed to formulate a


proof.

𝑪𝒐𝒏𝒋𝒆𝒄𝒕𝒖𝒓𝒆:

If a square piece is removed from a 5×5 parallelogram board then at


every cell it can be tiled given the parity of its index is equal and unable to be
tiled if the parity is unequal.

0,0 0,2 0,4 0,1 0,3

1,1 1,3 1,0 1,2 1,4

2,0 2,2 2,4 2,1 2,3

3,1 3,3 3,0 3,2 3,4

4,0 4,2 4,4 4,1 4,3

5
Fermaint

Figure 6 Figure 7

𝑰𝒏𝒔𝒑𝒆𝒄𝒕𝒊𝒐𝒏 𝒑𝒓𝒐𝒐𝒇 𝒇𝒐𝒓 𝒕𝒊𝒍𝒊𝒏𝒈 𝟓×𝟓 𝒃𝒐𝒂𝒓𝒅𝒔 𝒘𝒊𝒕𝒉 𝒅𝒐𝒎𝒊𝒏𝒐𝒆𝒔:

Let 𝑆 5,5 be the 5×5 parallelogram board in fig. 5 where 𝑆 [,\


designates a cell location, 𝑖, 𝑗 ∈ 𝕫: 0 ≤ 𝑖 ≤ 𝑗 ≤ 42. If an arbitrary square is
removed at 𝑆 [,\ then 𝑆 can be tiled by 2×1 dominoes. By inspection, refer to fig.
5 that shows a black square if 𝑆 can be tiled and gray if 𝑆 cannot be tiled. See that
fig. 6 shows only tiles where 𝑖 and 𝑖 have equal parity. Every square in this case is
black, hence 𝑆 can be tiled for all cells that have equal parity. Fig. 7 shows only
tiles where 𝑖 and 𝑗 have unequal parity. Every square in this case is gray, hence 𝑆
cannot be tiled for all cells that have unequal parity.

IV. 𝟖 × 𝟖 parallelogram board tiled with straight trominoes

Figure 8 Figure 9

Consider an 8×8 parallelogram board (fig. 9), and replace dominoes with
straight trominoes (fig. 8). Notice that a straight tromino is a stack of 3×1 squares
if vertical and 1×3 squares if horizontal. In either case, the area of a straight
tromino is 3 squares. Recall the 5×5 board in fig. 3. Because the area of 𝑆 was
not divisible by the area of the dominoes, the board was unable to be tiled.

This paper will refer to this concept as the polyomino divisibility theorem.

𝑷𝒐𝒍𝒚𝒐𝒎𝒊𝒏𝒐 𝑫𝒊𝒗𝒊𝒔𝒊𝒃𝒊𝒍𝒊𝒕𝒚 𝑻𝒉𝒆𝒐𝒓𝒆𝒎3


2
That is, S is indexed starting at zero.
3
This “theorem” was created for readability, the author does not take credit for this theorem nor is aware whether it
is recognized in the mathematics community or not.

6
Fermaint

For any 𝒎×𝒏 parallelogram board to


be tiled by a 𝒒×𝒑 polyomino tile,
there must exist some 𝒄 such that the
area of the board is equal the area of
the tile times 𝒄.

The area of straight trominoes is 3×1 = 1×3 = 3. The area of an 8×8


board is 64. Since there does not exist a natural number 𝑐 such that 3𝑐 = 64, it
can be deduced that any attempt to tile the 8×8 board will be unsuccessful. If one
square is removed from the 8×8 board the new area will be 63, then 𝑐 = 21 for
𝑐 in 3𝑐 = 63. It can be deduced that if one square is removed from an 8×8 board,
then the board can be tiled by 21 straight trominoes.

Figure 10

𝑪𝒐𝒏𝒋𝒆𝒄𝒕𝒖𝒓𝒆:

There exists at least one square tile which, if removed, will allow a 8×8
parallelogram board to be completely tiled by straight trominoes.

𝑰𝒏𝒔𝒑𝒆𝒄𝒕𝒊𝒐𝒏 𝒑𝒓𝒐𝒐𝒇 𝒇𝒐𝒓 𝒕𝒊𝒍𝒊𝒏𝒈 𝟖×𝟖 𝒃𝒐𝒂𝒓𝒅𝒔 𝒘𝒊𝒕𝒉 𝒕𝒓𝒐𝒎𝒊𝒏𝒐𝒔:

Let 𝑆 be the 8×8 parallelogram board in fig. 10. Let 𝑆([,\) be the location
of an arbitrary cell in 𝑆 where 𝑖, 𝑗 ∈ 𝕫: 0 ≤ 𝑖 ≤ 𝑗 ≤ 74. Let each square be


4
That is, S is indexed starting at zero.

7
Fermaint

colored by order of black, gray, and white from right to left diagonally. Remove
one arbitrary square from 𝑆 so that 𝑆UVWU = 63. Since there exists some 𝑐 ∈ ℕ
such that 3𝑐 = 63, there exists some square at 𝑆([,\) such that when removed
allows 𝑆 to be tiled by straight trominoes by the polyomino divisibility theorem.

Consider the fig. 10: there are 21 black squares, 21 white squares, and 22
gray squares. That means that the arbitrary square must be removed from the gray
cells, since 22 − 1 = 21 which is divisible by 3.

Since each straight tromino consists of 3 squares in a linear structure, and


𝑆 is colored in an order of threes, each tromino will cover a distinctly colored
square, regardless the orientation of the tromino. Removing a gray square at 𝑆(_,_)
does not result in a solution, since symmetrically if 𝑆 is rotated 90 degrees
clockwise, 𝑆(_,_) will be white, which is known not to be a solution. It follows
then that the solutions are the gray squares which, if 𝑆 is rotated, remain gray. By
inspection, the only squares that remain gray after rotation are 𝑆(v,v) , 𝑆(v,Z) , 𝑆(Z,v) ,
and 𝑆(Z,Z) , therefore there exists four cells which, if a square is removed its
location will allow a 8×8 board to be tiled by straight trominoes.

V. Tiling 𝟐𝒎 ×𝟐𝒎 parallelogram boards with bent trominos

Figure 11

Consider fig. 11. Bent trominoes, or L-shaped trominoes, are 2×2 square
structures that have 1 square removed, producing 4 different variations. Notice
that in either variation, the area of a bent tromino is 3 squares. That means the
area of a 2w ×2w board must be divisible by 3 by the polyomino divisibility
theorem. So 2w ×2w − 1 must be divisible by 3. Induction on 𝑚 is used to
determine whether 2w ×2w − 1 is divisible by 3.

𝑰𝒏𝒅𝒖𝒄𝒕𝒊𝒐𝒏 𝒑𝒓𝒐𝒐𝒇 𝒇𝒐𝒓 𝒅𝒊𝒗𝒊𝒔𝒊𝒃𝒊𝒍𝒊𝒕𝒚 𝒐𝒇 𝒂𝒓𝒆𝒂 𝒐𝒇 𝟐𝒎 ×𝟐𝒎 𝒃𝒐𝒂𝒓𝒅:

8
Fermaint

Let 𝑚 ∈ ℕ be arbitrary but fixed. Let 𝑆 𝑚, 𝑚 be a parallelogram board


constructed with 1𝑥1 squares with an area of 2w ×2w − 1.

𝑩𝒂𝒔𝒆 𝑪𝒂𝒔𝒆:

Let 𝑚 = 1, then 𝑆UVWU = 2x ×2x − 1 = 3. Since 𝑐 = 1 when 3𝑐 = 3


there exist some arbitrary square such that when removed from 𝑆(1,1) allows 𝑆 to
be tillable by bent trominoes.

𝑰𝒏𝒅𝒖𝒄𝒕𝒊𝒐𝒏 𝑯𝒚𝒑𝒐𝒕𝒉𝒆𝒔𝒊𝒔:

Suppose there exists some 𝑐 such that 𝑚 = 𝑘 holds for all 𝑘 ∈ ℕ such that
vy
3𝑐 = 2 − 1, it follows then that 𝑚 = 𝑘 + 1 must also be true.

𝑰𝒏𝒅𝒖𝒄𝒕𝒊𝒐𝒏 𝑺𝒕𝒆𝒑:

Let 𝑚 = 𝑘 + 1, then 𝑆UVWU = 2 ycx ×2 ycx − 1


→ 2×2y ×2×2y − 1 = 3𝑐
→ 4×2vy − 1 = 3𝑐
→ 2v ycx − 1 = 3𝑐
Therefore, 𝑆UVWU is divisible by 3 for all 𝑚 ∈ ℕ.

𝑪𝒐𝒏𝒋𝒆𝒄𝒕𝒖𝒓𝒆 𝒇𝒐𝒓 𝒕𝒊𝒍𝒊𝒏𝒈 𝟐𝒎 ×𝟐𝒎 𝒑𝒂𝒓𝒂𝒍𝒍𝒆𝒍𝒐𝒈𝒓𝒂𝒎 𝒃𝒐𝒂𝒓𝒅𝒔:

It is always possible to tile a 2w ×2w parallelogram board with bent


trominoes given one square is removed.

𝑰𝒏𝒅𝒖𝒄𝒕𝒊𝒐𝒏 𝒑𝒓𝒐𝒐𝒇 𝒇𝒐𝒓 𝒕𝒊𝒍𝒊𝒏𝒈 𝟐𝒎 ×𝟐𝒎 𝒑𝒂𝒓𝒂𝒍𝒍𝒆𝒍𝒐𝒈𝒓𝒂𝒎 𝒃𝒐𝒂𝒓𝒅𝒔:

Let 𝑚 ∈ ℕ be arbitrary but fixed. Let 𝑆 be a parallelogram board


constructed with 1𝑥1 squares where one square is removed such that the size of 𝑆
is 2w ×2w − 1. Let 𝑆([,\) be the location of an arbitrary cell in 𝑆
where 𝑖, 𝑗 ∈ 𝕫: 0 ≤ 𝑖 ≤ 𝑗 ≤ 2w − 1.

𝑩𝒂𝒔𝒆 𝑪𝒂𝒔𝒆:

Let 𝑚 = 1, then the area of 𝑆 is 𝑆 𝑚, 𝑚 − 1 = 2x ×2x − 1 = 3. So 𝑆 is


tillable for all arbitrary squares removed at any cell location by definition of a
bent tromino.

9
Fermaint

𝑰𝒏𝒅𝒖𝒄𝒕𝒊𝒐𝒏 𝑯𝒚𝒑𝒐𝒕𝒉𝒆𝒔𝒊𝒔:

Suppose 𝑚 = 𝑘 holds for all 𝑘 ∈ ℕ such that 𝑆 𝑘, 𝑘 and 𝑆 is tillable by


bent trominoes for all arbitrary removed squares, then 𝑚 = 𝑘 + 1 must also be
true.

𝑰𝒏𝒅𝒖𝒄𝒕𝒊𝒐𝒏 𝑺𝒕𝒆𝒑:

Let 𝑚 = 𝑘 + 1. Fig. 12 shows 𝑆 𝑘 + 1, 𝑘 + 1 , fig. 13 shows 𝑆(𝑘 +


1, 𝑘 + 1) divided into 4 quadrants, each with size 2y .

2ycx 2y

2y
𝑄x 𝑄_
2ycx

𝑄v 𝑄|
Figure 12 Figure 13

Without loss of generality, by the induction hypothesis, quadrant 𝑄_ can


be tiled with the square removed shown in figure 13. By the same reasoning, 𝑄x ,
𝑄v , and 𝑄| can be completely tiled with one arbitrary square removed. If these
three quadrants are rotated such that their removed squares form a bent tromino
then it is possible to place a bent tromino in the vacant spot.

Therefore, it is always possible to tile a 2w ×2w parallelogram board with


bent trominoes given one square is removed.

VI. Tiling 𝐦×𝐦 parallelogram boards with bent trominos


Finally, the accumulation of previous effort can be combined to prove
whether 𝑚×𝑚 square boards can be tiled with bent trominoes. Again, since the
area of the bent tromino is 3, 𝑚×𝑚 − 1 must be divisible by 3 by the polyomino
divisibility theorem. Only 𝑚 greater than 1 will be considered since the area of
the board must be greater than or equal to the area of the bent tromino to be tiled
by it. The following data will be considered to tease out a pattern:

10
Fermaint

𝑚 2 3 4 5 6 7 8 9 10 11
𝑚×𝑚 − 1 3 8 15 24 35 48 53 80 99 120

The discovered pattern dictates that 𝑚 cannot be divisible by 3.

Since it has already been proven that any 2w ×2w square board with one
arbitrary square removed is completely tillable, the first case to consider is when
𝑚 = 5.

Figure 14

By inspection, a 5×5 board can be tiled by bent trominoes if the corner


squares are removed, but not tiled if the adjacent squares next to the corners are
removed. Review fig.14, black squares are that which, if removed, allow the
board to be tiled by bent trominoes. Gray squares are that which, if removed, do
not allow the board to be tiled.

Figure 15

Further, an 𝑚×𝑛 board can be completely tiled by bent trominoes without


removing a square if 𝑚 = 2, 𝑛 = 3, and 𝑚, 𝑛𝜖ℕ > 1, as fig. 15 shows. Clearly,
this is true whether the 𝑚×𝑛 board is rotated or not.

11
Fermaint

2𝑥3

5𝑥5 − 1

3𝑥2

Figure 16_a

Using the knowledge of 2×3 boards and 5×5 boards, it can be proven
upon inspection that a 7×7 board can be completely tiled by bent trominos if an
arbitrary square is removed. Because of symmetry, only cell index locations 0 ≤
𝑖 ≤ 𝑗 ≤ 3 are required for inspection. Let 𝑆 be a 7×7 parallelogram board with an
arbitrary square removed. Upon inspection, it’s shown in fig. 16_𝑎 that if the
square is removed at 𝑆(v,v) it is tillable since a 5×5 board with one square
removed at the corner is tillable.

5𝑥5 − 1

3𝑥2

2𝑥3

Figure 16_b

If a square at 𝑆(_,_) , 𝑆(_,x) , 𝑆(x,_) or 𝑆(x,x) is removed, then it is tillable. Fig.


16_b shows this inspection, where all four locations can be proven by rotating the
bent tromino in the upper left of the board.

12
Fermaint


3 4𝑥3

4𝑥3
3𝑥4

Figure 16_c

Using the same logic, fig. 16_𝑐 proves that if a square is removed from
𝑆(_,v) , 𝑆(_,|) , 𝑆(x,v) , or 𝑆(x,|) then 𝑆 is tillable, by rotating the uppermost bent
tromino. Notice that 3×4 and 4×3 rectangle pieces are multiples of the 3×2 and
2×3 rectangle pieces and hence are also completely tillable by bent trominoes.

3𝑥4 3𝑥2 3
4𝑥3

4𝑥3
3𝑥4
2𝑥3

Figure 16_d Figure 16_e

Figure 16_𝑑 and figure 16_𝑒 both prove by inspection that 𝑆 is tillable if a
square is removed at 𝑆(|,|) or 𝑆(|,v) .

It has been proven that a 7×7 board with one square removed at any
location can be completely tiled by bent trominoes. Enough has been proven that
two conjectures by parity can be formulated.

13
Fermaint

𝑪𝒐𝒏𝒋𝒆𝒄𝒕𝒖𝒓𝒆𝟏 :

There exists at least one arbitrary square such that if removed from any
𝑚×𝑚 parallelogram board, with 𝑚 ∈ ℕ is not divisible by 3, 𝑚 > 5, 𝑚 is odd,
allows the 𝑚×𝑚 board to be tiled by bent trominoes.

𝑪𝒐𝒏𝒋𝒆𝒄𝒕𝒖𝒓𝒆𝟐 :

There exists at least one arbitrary square such that if removed from any
𝑚×𝑚 parallelogram board, with 𝑚 ∈ ℕ is not divisible by 3, 𝑚 > 1, 𝑚 is odd,
allows the 𝑚×𝑚 board to be tiled by bent trominoes.

𝑰𝒏𝒅𝒖𝒄𝒕𝒊𝒐𝒏 𝒑𝒓𝒐𝒐𝒇 𝒇𝒐𝒓 𝒕𝒊𝒍𝒊𝒏𝒈 𝒎×𝒎 𝒃𝒐𝒂𝒓𝒅𝒔 𝒘𝒊𝒕𝒉 𝒃𝒆𝒏𝒕 𝒕𝒓𝒐𝒎𝒊𝒏𝒐𝒆𝒔:

Let 𝑚 ∈ ℕ be arbitrary but fixed. Let 𝑆 be a parallelogram board


constructed with 1𝑥1 squares where one square is removed such that the size of 𝑆
is 𝑚×𝑚 − 1. Let 𝑆([,\) be the location of an arbitrary cell in 𝑆
where 𝑖, 𝑗 ∈ 𝕫: 0 ≤ 𝑖 ≤ 𝑗 ≤ 𝑚 − 1.

𝑩𝒂𝒔𝒆 𝑪𝒂𝒔𝒆𝟏 :

Let 𝑚 = 11, then 𝑆UVWU − 1 = 120 which is divisible by 3 since 𝑐 = 40


when 3𝑐 = 120. So 𝑆 should be tillable by the polyomino divisibility theorem.
Consider fig. 17 below:

14
Fermaint

6𝑥4
7𝑥7

5𝑥5
4𝑥6

Figure 17

The square is removed at location 𝑆(Z,Z) . The 4𝑥6 and 6𝑥4 rectangle
pieces are multiples of the 2𝑥3 and 3𝑥2 rectangle pieces and hence proven to be
completely tillable by bent trominoes. The 5𝑥5 and 7𝑥7 square pieces have been
proven to be tillable by bent trominoes given a square is removed. Thus, an
11𝑥11 square board is tillable by bent trominoes.

𝑰𝒏𝒅𝒖𝒄𝒕𝒊𝒐𝒏 𝑯𝒚𝒑𝒐𝒕𝒉𝒆𝒔𝒊𝒔:

Assume 𝑚 > 11 is odd and not divisible by 3. Suppose also that all
boards smaller than 𝑚 which are odd and not divisible by 3 are tillable.

𝑰𝒏𝒅𝒖𝒄𝒕𝒊𝒐𝒏 𝑺𝒕𝒆𝒑:

15
Fermaint

(𝑚 − 7) 𝑥 6
(𝑚 − 6) 𝑥 (𝑚 − 6)

7 𝑥 7
6 𝑥 (𝑚 − 7)

Figure 17

Consider figure 17 above, it is assumed that the (𝑚 − 6) 𝑥 (𝑚 − 6) piece


contains the removed square. If 𝑚 is odd, then 𝑚 − 7 is even, so both 6𝑥(𝑚 − 7)
and (𝑚 − 7)𝑥6 pieces are multiples of the 2𝑥3 pieces proven to be tillable on pg.
10. The 7𝑥7 square piece with an arbitrary square removed is clearly tillable by
bent trominoes as previously proven. If 𝑚 is odd and not divisible by 3, then so is
𝑚 − 6. If that is the case, then the (𝑚 − 6) 𝑥 (𝑚 − 6) piece is tillable by bent
trominoes from the induction hypothesis.

𝑩𝒂𝒔𝒆 𝑪𝒂𝒔𝒆𝟐 :

Let 𝑚 = 2, then 𝑆UVWU − 1 = 3 which is divisible by 3 since 𝑐 = 1 when


3𝑐 = 3. Further, 𝑆 is tillable by bent trominoes by definition of a tromino.

𝑰𝒏𝒅𝒖𝒄𝒕𝒊𝒐𝒏 𝑯𝒚𝒑𝒐𝒕𝒉𝒆𝒔𝒊𝒔:

Assume 𝑚 > 8 is even and not divisible by 3. Suppose also that all
boards smaller than 𝑚 which are even and not divisible by 3 are tillable.

16
Fermaint

𝑰𝒏𝒅𝒖𝒄𝒕𝒊𝒐𝒏 𝑺𝒕𝒆𝒑:

(𝑚 − 3) 𝑥 (𝑚 − 3) (𝑚 − 4) 𝑥 6

4 𝑥 4
6 𝑥 (𝑚 − 4)

Figure 18

Consider fig. 18, it is assumed the square was removed from the (𝑚 −
3)𝑥(𝑚 − 3) piece. If 𝑚 is even and not divisible by 3, then (𝑚 − 3) is odd and
not divisible by 3. Since 𝑚 > 8 and 𝑚 − 3 > 5, the previous induction
assumption applies. The (𝑚 − 4)×3 are multiples of the 2𝑥3 pieces proven to be
tillable on pg. 10. Considering the previous work performed with Golomb’s
theorem5, it is known that any 2w 𝑥2w square board is tillable by bent trominoes,
which applies to the 4𝑥4 square board. So, the entire board can be tiled.

VII. Summary
If 𝑚 ≠ 5, then there exists an arbitrary square removed from any 𝑚×𝑚
parallelogram board such that the board can be tiled with bent trominoes if
and only if 𝑚 is not a multiple of 3.


5
Refer to V on the bottom of page 6.

17
Fermaint

VIII. Cited Aid


The mathematics and the paper was written by Derek Fermaint. However,
the author got extensive help from the following:

• “Chapter Zero: Fundamental Notions of Abstract Mathematics”,


Carol Schumacher
• “How to Prove It”, Daniel J Velleman
• MC2 Math Tutoring Center, Arizona State University
• Wendy Caldwell, former Math TA.
• Wolfram Alpha, (www.wolframalpha.com)
• Khan Academy, (www.khanacademy.org)
• Cut The Knot, (http://www.cut-the-knot.org)

18

Anda mungkin juga menyukai