2

Suppose $a,b \in \mathbb{N}$. Prove that if there are integers $m$ and $n$ such that $am +bn =1$ then $a$ and $b$ are coprime.

I came up with the following proof, but I am sure a shorter argument is possible.

To prove: $\forall a,b \in \mathbb{N}$ , $\exists m,n \in \mathbb{Z}$ | $am + bn = 1\rightarrow $ $gcd(a,b) = 1$

In order to prove this by contradiciton, suppose then that $\exists m,n \in \mathbb{Z}$ | $am + bn = 1$ and that $gcd(a,b) \neq 1$.

Take $ k = gcd(a,b) \neq 1$. Now, $k = ra+sb$ and $s,b \in \mathbb{Z}$, assuming that k can be written as a linear combination of a and b. This is an established theorem.

So we have:

(1) $am + bn = 1$

(2) $ra + sb = k$

Adding $(1) + (2)$ , we get :

$(r+m)a + (s+n)b = k+1$

Since $ k= gcd(a,b)$, then $k|(r+m)a$ and $k|(s+n)b$. Then $k|(r+m)a + (s+n)b = k+1$.

So $k|k+1$. But this is impossible, since dividing $k \neq1$ into $k$ gives a remainder of 1.

Having derived this contradiction, it cannot be the case that if $am + bn = 1$, then $gcd(a,b) \neq 1$. So it must be the case that:

($\exists m,n \in \mathbb{Z}$ | $am + bn = 1$) $\rightarrow $ ($gcd(a,b) = 1$)

JKM
  • 449
  • Notably, this is the converse of Bezout's identity, which gives us that, for $a,b$ coprime, then there exist $x,y$ integers such that $ax+by=1$. The converse is discussed on MSE here -- https://math.stackexchange.com/questions/1279900/converse-of-b%C3%A9zouts-identity – PrincessEev Mar 26 '19 at 03:30
  • Now you need to prove that $k|k+1$ is impossible for $k>1$. If you had done the argument at the end with $ma+nb=1$ to say that $k$ would divide $1$, then would have only had to prove that $k|1$ is impossible for $k>1$. – user647486 Mar 26 '19 at 03:32
  • 1
    Shorter argument: if $k = \operatorname{gcd}(a, b)$, then $a = xk$ and $b = yk$ for integers $x$ and $y$. Thus $$1 = am + bn = k(xm + yn),$$ i.e. $k$ divides $1$. – Theo Bendit Mar 26 '19 at 03:38

2 Answers2

3

Suppose $am+bn=1$. If $k$ divides both $a$ and $b$ then there exist $p$ and $q$ such that $a=kp$ and $b=kq$.

Substituting that into our first equation gives $kpm+kqn=1\implies k$ divides $1$

Therefore, $k=1$ and $a$ and $b$ are coprime.

John Douma
  • 11,426
  • 2
  • 23
  • 24
0

This is the shortest proof that I could think of: Suppose $a, b \in \mathbb{N}$. Let $am + bn = 1$ for $m, n \in \mathbb{Z}$.

Now, if $am + bn = 1$ then this implies there's a solution to the congruence $am \equiv 1 \mod b$. Now, let $d= \gcd(a, b)$ where $d>0$.

Then, we have; $$b \vert (am-1)$$ and $$d \vert b$$ and so $$ d \vert(am-1)$$ Since $d \vert a$ also, it must be that $d \vert -1$. Thus $d=1$, and, hence, $a$ and $b$ are co-prime as $gcd(a, b) = 1$.

YFP
  • 157