Anda di halaman 1dari 4

Math 109 - Homework #7 Selected Solutions

Professor: Bejenaru
TA: Robert Won
10.

Write the gcd as an integral linear combination of (i) 165 and 252 and (ii) 4284 and 3480.

Proof. We performed the Euclidean algorithm in HW 5.6. So we can backsolve. For (i) we had
252 = 165 1 + 87
165 = 87 1 + 78
87 = 78 1 + 9
78 = 9 8 + 6
9=61+3
6=32+0
And hence
3 = 9 6 = 9 (78 9 8) = 9 9 78 = (87 78) 9 78 = 87 9 78 10
= 87 9 (165 87) 10 = 87 19 165 10 = (252 165) 19 165 10 = 252 19 + 165 (29)
For (ii) the Euclidean algorithm was
4284 = 3480 1 + 804
3480 = 804 4 + 264
804 = 264 3 + 12
264 = 12 22 + 0
so
12 = 804 262 3 = 804 (3480 804 4) 3 = 804 13 3480 3
= (4284 3480) 13 3480 3 = 4284 13 + 3480 (16)

11.
Given positive integers a and b, suppose that applying the Euclidean algorithm leads to a
sequence of positive integers a0 , a1 , . . . , aN . Suppose that the procedure of the proof of Theorem
17.1.1 leads to expressions of the form
ak = amk + bnk
for integers mk and nk . Prove that mk nk1 mk1 nk = (1)k for 1 k N . Deduce that the
numbers mk and nk are coprime.

I will always provide full solutions to graded homework exercises. If I think I left something too unclear during
section, or if many students are having trouble with the same thing, I will post solutions to those problems as well.

Proof. Proceed by induction on k. For the base case, k = 1, since a0 = a1+b0 and a1 = a0+b1,
we have
m1 n0 m0 n1 = 0 0 1 1 = 1 = (1)1 .
Now assume the result is true for some 1 j < N . If we can prove the result is true for the
j + 1 case, then we will be done by induction. By the induction hypothesis, we have that
aj1 = amj1 + bnj1
aj = amj + bnj
and mj nj1 mj1 nj = (1)j . Performing one more step of the Euclidean algorithm yields
aj1 = q aj + aj+1 for some q Z
and hence
aj+1 = aj1 q aj = amj1 + bnj1 q(amj + bnj ) = a(mj1 qmj ) + b(nj1 qnj ).
So we see mj+1 = mj1 qmj and nj+1 = nj1 qnj . Now observe that
mj+1 nj mj nj+1 = (mj1 qmj )nj mj (nj1 qnj ) = mj1 nj mj nj1 .
But by the induction hypothesis,
mj1 nj mj nj1 = (mj nj1 mj1 nj ) = (1)j = (1)j+1 .
So we have finished our proof by induction.
Now since for all k, we can write (1)k as an integral linear combination of mk and nk , by Prop
17.3.1, mk and nk are coprime. (If k is odd, then just negate the coefficients mk1 and nk1 ). 
13.
(i) Give a proof by contradiction that if a positive integer n is a common multiple of a and
b then [a, b] divides n.
(ii) This means ab/[a, b] is an integer. Prove that this integer is a common divisor of a and b.
Deduce that ab/[a, b] (a, b).
(iii) Prove that ab/(a, b) is a common multiple of a and b. Deduce that (a, b)[a, b] = ab if a and
b are positive. Find lcm[612, 696].
Proof. (i) As suggested, we give a proof by contradiction. Suppose n is a common multiple of n
such that [a, b] does not divide n. Using the division theorem, since [a, b] does not divide n, we have
n = [a, b]q + r for 0 < r < [a, b]. Since both n and [a, b] are multiples of a, we can write n = ak1
and [a, b] = ak2 . Hence, ak1 = ak2 q + r so r = a(k1 k2 q). Hence, r is a multiple of a. By the
same reasoning, r is a multiple of b. So r is a common multiple of a and b, and 0 < r < [a, b] but
this contradicts that [a, b] was the least common multiple.
(ii) Since ab is a common multiple of a and b, we thus have that ab/[a, b] is an integer. But
since [a, b] is a multiple of b, we have that [a, b]/b is an integer as well. So ab/[a, b] [a, b]/b = a.
Hence, ab/[a, b] is a divisor a. Similarly, ab/[a, b] is a divisor of b. Therefore, ab/[a, b] is a common
divisor, and so ab/[a, b] (a, b) by the definition of greatest common divisor.
2

(iii) Since (a, b) is a divisor of b, this means b/(a, b) is an integer. Hence, a b/(a, b) = ab/(a, b)
is a multiple of a. Similarly, ab/(a, b) is a multiple of b. Therefore ab/(a, b) is a common multiple
of a and b. If both a and b are positive, then ab/(a, b) [a, b] by the definition of least common
multiple. So in the case that a, b > 0, taking (ii) and (iii) together, we have ab (a, b)[a, b] from
(ii) and ab (a, b)[a, b] from (iii). Hence, ab = (a, b)[a, b].
We use this fact to compute lcm[612, 696]. To do this, use the Euclidean algorithm to find

gcd(612, 696) = 12. So lcm[612, 696] = 612696/12


= 35496.

14.
Let a and b be positive integers. By the well-ordering principle the non-empty set of positive
integers
{am + bn | m, n Z and am + bn > 0}
has a minimum element c. Prove by contradiction that c is a common divisor of a and b and hence
give an alternative proof of Theorem 17.1.1.
Proof. Let c = min{am + bn | m, n Z and am + bn > 0}. Suppose for contradiction that c is not
a common divisor of a and b. Then either c does not divide a or c does not divide b. Suppose c
does not divide a. Then by the division theorem, there are integers q, r Z with 0 < r < c such
that a = qc + r. Write c = am0 + bn0 . Then
r = a qc = a q(am0 + bn0 ) = a(1 qm0 ) + b(qn0 ).
So r {am + bn | m, n Z and am + bn > 0} and 0 < r < c, but this contradicts the minimality
of c.
Since the statement is symmetric in a and b, the same proof works in the case that c does not
divide b.

15.

Solve the linear diophantine equations

(i) 165m + 252n = 15,


(ii) 165m + 252n = 20,
(iii) 4284m + 3480n = 60,
(iv) 4284m 3480n = 36.
Proof. First, use the Euclidean algorithm to find gcd(165, 252) = 3 and 3 = 252(19) + 165(29) (as
in 10). This shows that (ii) has no solution by Theorem 18.2.1. For (i), we have that 15 = 252(95)+
165(145). So to find all solutions, by Proposition 18.4.1, we need to solve the homogeneous case
165(m + 145) + 252(n 95) = 0
which is true whenever (m + 145, n 95) = (84q, 55q) for some q Z. Hence, (m, n) = (145 +
84q, 95 55q) for q Z.
Next use Euclidean algorithm to find 12 = 4284(13) + 3480(16). So for (ii) we have 60 =
4284(65) + 3480(80). The set of all solutions is given by (m, n) = (65 + 290q, 80 357q) for
q Z.
For (iv) we have 36 = 4284(39) + 3480(48) so the set of all solutions is (m, n) = (39 +
290q, 48 357q) for q Z.

3

16.

Solve the linear diophantine equation


336m + 238n = 5558.

Prove that there is a unique pair of positive integers m and n satisfying this equation and find this
solution.
Proof. Use Euclidean algorithm to find gcd(336, 238) = 14 and 14 = 336(5) + 238(7). Then
5558 = 336(1985) + 238(2779)
and the set of all solutions is (m, n) = (1985 + 17q, 2779 24q) for q Z. Take q = 116 to
get the solution (13, 5). This is a solution where m and n are both positive integers. All other
solutions are of the form (13 + 17q, 5 24q) for q 6= 0. If q > 0, then 5 24q < 0. If q < 0, then
13 + 17q < 0. Hence, no other solutions have both coefficients positive, so m 13, n = 5 is the
unique pair of positive integers satisfying the equation.

18.

Solve the linear diophantine equation


6m + 10n + 15p = 1.

Proof. Rewrite the equation so that we must solve 6m + 5(2n + 3p) = 1. One particular solution
is given by m = 1, 2n + 3p = 1, so the set of all possible solutions is given by m = 1 + 5q,
2n + 3p = 1 6q for q Z. So now we need to solve the second diophantine equation. Fix q Z.
Notice that gcd(2, 3) = 1 with 1 = 2(1) + 3(1). So a particular solution to this second diophantine
equation is n = 1 + 6q and p = 1 6q. For fixed q, the set of all solutions are thus given by
n = 1 + 6q + 3r and p = 1 6q 2r for some r Z.
Hence, the full set of all possible solutions is given by m = 1+5q, n = 1+6q+3r, p = 16q2r
for any q, r Z.


Anda mungkin juga menyukai