0
$\begingroup$

I need to (1) identify the error in this proof, and (2) give a counterexample to show that the conditional is false.

Proposition: Let $u$, $m$, $n$ be three integers. If $u\mid mn$ and $\gcd(u,m) = 1$, then $m = \pm 1$.

Proof: If $\gcd(u,m) = 1$, then $1 = us + mt$ for some integers $s$ and $t$. If $u\mid mn$, then $us = mn$ for some integer $s$. Hence, $1 = mn + mt = m(s + t)$, which implies that $m\mid 1$, and therefore $m = \pm 1$.

For the counterexample, I have:

Let $d = \gcd(u,m)$, then $a = dn$ and $b = dm$ for some integers $m$ and $n$. By Theorem 1.34 (from study guide), for positive integers $a$, $b$ there exists a unique positive integer $c$ such that $c = \gcd(a,b)$ and $c = ax + by$ for some integers $x$ and $y$. Let $c = 1$ and if $1 = ax + by$, then $1 = dnx + dmy = d(nx + my)$. Thus $d \mid 1$, and since $d$ is a positive integer, we conclude $d = 1$. So $d = ax + by$ for some integers $x$ and $y$. Thus $d = dnx + dmy$ and by the cancellation law, $1 = nx + my$, and $1 = \gcd(m,n)$.

Do I have the right idea with this or am I completely off track? If this is not correct for a counterexample any help pointing me in the right direction would be appreciated.


As for the error in the conditional, if $u\mid mn$ and $\gcd(u,m) = 1$, then $m = \pm 1$, I believe it is in "$m = \pm 1$". I believe that $m$ should be positive and be equal to $1$.

$\endgroup$

3 Answers 3

0
$\begingroup$

For the counterexample, you want to use specific numbers. For example: $u=2, m=3, n=4$. If you check those, you'll see that they satisfy the premises, but not the conclusion. See if you can make more examples yourself, and then note what they have in common.

$\endgroup$
3
  • $\begingroup$ Thank you for your help. I keep forgetting that I can use numbers in counterexamples. That made it much easier. Also, if you don't mind, am I correct is saying that s is being 'defined' in two different if-then statements. I also believe that s is being used incorrectly here, ' mn + mt = m(s + t)'. Should this read mn + mt = m(n + t) instead? $\endgroup$
    – user92986
    Sep 6, 2013 at 2:47
  • $\begingroup$ Something like that. It's like they tried to substitute $us$ for $mn$, but then it turned into $ms$ instead. You're right about that step being wrong, anyway. $\endgroup$ Sep 6, 2013 at 2:52
  • $\begingroup$ It turned out that there was a typo on the exercise and us should have been mn. $\endgroup$
    – user92986
    Sep 13, 2013 at 22:20
0
$\begingroup$

Hint: how is $s$ defined? Is it used consistently in the proof?

$\endgroup$
5
  • $\begingroup$ The s is being 'defined' in two different if-then statements. I also believe it is being used incorrectly here, ' mn + mt = m(s + t)'. Should this read mn + mt = m(n + t) instead? $\endgroup$
    – user92986
    Sep 6, 2013 at 2:27
  • $\begingroup$ You're right that it's a typo in that final equality. But I was referring to the two different definitions. Are they interchangeable? $\endgroup$
    – Glen O
    Sep 6, 2013 at 2:48
  • $\begingroup$ Sorry, I just can't see where the error is. Also, I'm not sure how to tell if the two definitions are interchangeable or not. $\endgroup$
    – user92986
    Sep 6, 2013 at 3:32
  • 1
    $\begingroup$ Try it for a moment with actual numbers. See G Tony Jacobs' answer for numbers to test. Now you have $us=mn$, which becomes $2s=3\times4$, which gives $s=6$. Now look at the other equation, $1=us+mt = 2s+3t$. If $s=6$, this becomes $1=12+3t$; you can't satisfy this with an integer $t$. Clearly, the values of $s$ are not the same, and one should instead use a different variable for one of the two equations - say, $uq=mn$, rather than $us=mn$. What impact does this have on the proof? $\endgroup$
    – Glen O
    Sep 6, 2013 at 11:59
  • $\begingroup$ Hi, thanks for you help. I think I have everything that I need in order to complete the exercise. By using s in both equations this makes the proof false. Also, the t will not be an integer which it should be. $\endgroup$
    – user92986
    Sep 7, 2013 at 16:58
0
$\begingroup$

I think the error in that proof is the use of the variable $s$, since the existential instantiation states that

If the existence of a certain kind of object is assumed or has been deduced then it can be given a name, as long as that name is not currently being used to denote something else

Therefore, if $\gcd(u,m)=1$ then $1\mid m: m=1p$ and $1\mid u: u=1q$. Hence, $\gcd(u,m)=1=us+mt=(1p)s+(1q)t=1(ps+qt)$ which implies that $m\mid 1$, therefore $m=\pm 1$

$\endgroup$

You must log in to answer this question.