Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

If $1/a + 1/b = 1 /c$ where $a, b, c$ are positive integers with no common factor, $(a + b)$ is the square of an integer.

I found this question in RMO 1992 paper !

Can anyone help me to prove this ?

May be it's too easy and simple . But , I'm just novice in number theory !

share|cite|improve this question
I wonder, whether such numbers exist? – pushpen.paul Oct 17 '14 at 7:31
OK. Found some: $(2,2,1), (3, 6, 2), (4, 12, 3), (5, 20, 4), (6, 30, 5), (7, 42, 6), (8, 56, 7), (9, 72, 8) \ldots$. – pushpen.paul Oct 17 '14 at 7:46
up vote 10 down vote accepted

Let $\gcd(a,b) = g$, and $a = a'g$ and $b = b'g$ (so that $\gcd(a', b') = 1$). The equation $\frac1a + \frac1b = \frac1c$ is the same as $c(a+b) = ab$, which, dividing throughout by $g$, is $$c(a' + b') = a'b'g.$$

Now, as $(a' + b')$ divides $a'b'g$ but is relatively prime to both $a'$ and $b'$, it must divide $g$. Similarly, as $g$ divides $c(a'+b')$ but is relatively prime to $c$ (note that $\gcd(g, c) = \gcd(\gcd(a,b), c) = \gcd(a, b, c) = 1$) it must divide $(a' + b')$. Thus as both $g$ and $(a' + b')$ divide each other, we have $$(a' + b') = g,$$ and therefore $$(a + b) = g(a' + b') = g^2.$$

share|cite|improve this answer
Your answer accepted :) – arnab Oct 20 '13 at 13:14
actually your argument is quite similar to mine . in this mean time , due to my INTERNET server failed to connect , i had to try it on my own .. i augmented in quite a similar way ,except i love your final step argumentation to stand a'+b' =g !! :) – arnab Oct 20 '13 at 13:20

The hypothesis may be written as $(a+b)c = ab$, which is equivalent to $(a-c)(b-c)=c^2$.

Let $p$ be a prime factor of $c$. Then $p$ divides $a-c$ or $b-c$, but it cannot divide both because $\gcd(a,b,c)=1$. Hence $p^2$ divides $a-c$, say. This means that both $a-c$ and $b-c$ are squares since their product is a square: $a-c=u^2$, $b-c=v^2$. This implies that $c=uv$. But then $a+b-2c=u^2+v^2$ and so $a+b=(u+v)^2$.

The argument above means that all examples of $$ \frac 1 a + \frac 1 b = \frac 1 c $$ with $\gcd(a,b,c)=1$ are given by $$\frac1{u(u+v)}+\frac1{v(u+v)}=\frac1{uv}$$ with $\gcd(u,v)=1$.

share|cite|improve this answer
But then $gcd(a,b)=gcd(u(u+v),v(u+v))=u+v$ – chubakueno Oct 18 '13 at 22:24
@chubakueno: Yes, that is necessarily true. Note that the question says $\gcd(a, b, c) = 1$, not that $\gcd(a, b) = 1$. The latter is not possible because as these answers show, $\gcd(a,b) = u + v$ for positive integers $u$ and $v$. – ShreevatsaR Oct 20 '13 at 1:47
excellent ......! – arnab Oct 27 '13 at 16:04

The hypothesis may be rewritten as $$c(a+b) = ab.$$ Let $p$ be a prime factor of $a+b$.

Assume that $p\mid c$. If $p\mid a$ then $p\mid (a+b)-a = b$ which implies that $p$ is a common factor for $a, b, c$. Thus $p$ and $a$ are coprime. The same argument shows that $p$ and $b$ are coprime. Then $p$ and $ab$ are coprime. This contradicts the hypothesis $p \mid c$. Thus $p$ is coprime with $c$.

Now let denote by $\nu(x)$ the $p$-adic order of an integer $x$. From the first equality we get, thanks to $\gcd(p, c) = 1$, $$\nu(a+b) = \nu(ab).$$ This can be rewritten as $$\nu(a+b) = \nu(a) + \nu(b).$$ Thus $\nu(a) \leq \nu(a+b)$ i.e. $p^{\nu(a)} \mid (a+b) - a = b$, thus $\nu(a) \leq \nu(b)$. The same argument shows that $\nu(b) \leq \nu(a)$, hence $$\nu(a) = \nu(b).$$ Thus $$\nu(a+b) \in 2\mathbb{N}.$$

As this holds for any of its prime factors, $a+b$ is a square.

share|cite|improve this answer
FWIW: I'm pretty sure there exists an intended solution that does not require knowledge of p-adic orders. :-) – ShreevatsaR Oct 20 '13 at 1:33

We have, $ab=(a+b)c$. Then $$a+b=(a/m)(b/n)=a^\prime b^\prime$$ where $a=ma^\prime ,b=nb^\prime$ and $c=mn$.

Now, let a prime $p\mid a^\prime$. Then, $p\mid a$ and hence $p\mid b$. If $p\mid n$, then $p\mid c$ but this is a contradiction. Hence $p\mid b^\prime$. Similarly, if a prime $p\mid b^\prime$ then $p\mid a^\prime$.

Hence, $a^\prime$ and $b^\prime$ has same prime factors. Now suppose for some prime $p$, let $r,s$ (with $r<s$) be the highest exponents such that $p^r\mid a^\prime$ and $p^s\mid b^\prime$. Then $p^s\mid a^\prime b^\prime, p^s\mid b$. Now if $p^s \mid a$ then $p\mid m$, but then $p\mid c$ but this contradicts $\gcd (a,b,c)=1$. Hence prime factorizations of $a^\prime$ and $b^\prime$ are same and hence $a+b$ is square.

share|cite|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.