Take the 2-minute tour ×
Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

How to prove following statement :


An odd number $n$ , $(n>1)$ can be uniquely expressed as : $n= x^2-y^2$ ; $x,y \in \mathbb{Z}^{*}$

if and only if $n$ is a prime number .

If $x-y=m$ , where $m>1$ then $m \mid n$

Proof :

$n=x^2-y^2=(y+m)^2-y^2=y^2+2\cdot y\cdot m +m^2-y^2 \Rightarrow$

$\Rightarrow n=m\cdot (2\cdot y+m) \Rightarrow m \mid n$

Therefore , if $m \neq 1$ it follows that $n$ is a composite number , but how to prove that every odd

composite number ,other than $1$ , has representation : $x^2-y^2$ , where $x-y>1$ ?

share|improve this question
If $n=ab$ with $a\ge b>1$ (all odd), can you find an integer solution to the pair $x+y=a$, $x-y=b$? –  Jyrki Lahtonen Jan 10 '12 at 15:47
@JyrkiLahtonen,Quite simple...thanks !! –  pedja Jan 10 '12 at 16:01
What exactly is $\mathbb{Z}^{*}$? Does your conjecture hold for 9 (or the square of any other odd prime)? –  Mike Jan 10 '12 at 16:36
@Mike,Set of positive integers + $0$... –  pedja Jan 10 '12 at 16:39
@Mike, I was wondering about 9 myself, but only looked at the condition $x-y>1$. Didn't notice $\mathbf{Z}^*$! –  Jyrki Lahtonen Jan 10 '12 at 16:41

4 Answers 4

up vote 3 down vote accepted

Write $n=ab$ with $b>1$ and $x^2-y^2=(x+y)(x-y)$. Solve $x+y=a$, $x-y=b$. This system has integer solutions $x=(a+b)/2$ and $y=(a-b)/2$ because $a$ and $b$ are odd and hence $a\pm b$ is even. Finally, $x-y=b>1$, as required.

share|improve this answer
Uniqueness is an important part of the conjecture. Also, the conjecture is a biconditional (if and only if). It is not clear where your answer addresses either of these issues. –  robjohn Aug 9 '12 at 14:45
@robjohn, my answer adresses the actual question at the end of the OP text: "but how to prove that every odd composite number...". –  lhf Aug 9 '12 at 16:57
fair enough :-) –  robjohn Aug 9 '12 at 17:57

I tried to solve & got somewhere & I am presenting the same. This may be very long way, but it may be someway ...

We want prime numbers only by performing $x^2 - y^2$. Since all primes (except $2$) are odd, this suggests that both $x$ and $y$ are neither even nor odd. One is odd and other is even

So, Let us take $x = 2k_2 + 1$ and $y = 2k_1$, with $k_2 \geq k_1$, $k_1,k_2 \in \mathbb{N}$ (This is necessary since in $x^2 - y^2, y^2$ must not be $0$, so $y$ must not be $0$, so $k_1$ must not be $0$) (And this assumption does not take all odd numbers and so not all primes, but later we can reverse the role of $x$ and $y$)

Now, our final result $N = x^2 - y^2$ is a prime number. And we need to prove that for every prime number, we only have a unique $x$ and $y$.

Let $k_2 = k_1 + m, m \in \{0,1,2,\cdots\}$
Replacing $k_2$,
we get $N = 4m^2 + 8k_1m + 4k_1 + 4m + 1$ which is a prime number.

It is observed that the unique solution for all primes should have $x$ and $y$ as consecutive numbers. So if we prove that is $N$ has to prime, then $m$ has to $0$ for any $k_1$ that makes $N$ prime then we are done.

Taking $k_1$ as $k$,

$N = 4m(m + 2k + 1) + 4k + 1$

If we take $m = 0, N = 4k + 1$ which is superset of all primes (except $2$).

But then we need to prove that above $N$ can never be prime for any other value of $m$.

share|improve this answer

HINT $\ $ (Non) uniqueness follows from the composition law for differences of squares

$\rm\qquad\qquad\ (a^2-b^2)\ (A^2-B^2)\ =\ (a\:A+b\:B)^2-(a\:B+A\:b)^2$

$\rm\qquad\qquad\ \phantom{(a^2-b^2)\ (A^2-B^2)}\ =\ (a\:A-b\:B)^2-(a\:B-A\:b)^2$

E.g. composing $\rm\ 7 = 4^2 - 3^2\ $ with $\ 11 = 6^2 - 5^2\ $ yields for $\rm\: 7\cdot 11\:$ the following $\,2\,$ rep's

$\rm\qquad\qquad\ (4^2-3^2)\ (6^2-5^2)\ =\ (4\cdot 6+3\cdot 5)^2-(4\cdot 5+6\cdot 3)^2\ =\ 39^2 - 38^2$

$\rm\qquad\qquad\ \phantom{(4^2-3^2)\ (6^2-5^2)}\ =\ (4\cdot 6-3\cdot 5)^2-(4\cdot 5-6\cdot 3)^2\ =\ 9^2 - 2^2$

share|improve this answer
An elegant solution –  Hiral Jhaveri Jan 12 '12 at 4:54
See also this question. –  Bill Dubuque Aug 9 '12 at 14:38

If $n$ is odd and factors nontrivially as $ab$ with $a\le b$, then $n$ is the sum of $a$ consecutive odd numbers of which $b$ is the center one. Since the odd numbers are the first differences of the sequence of perfect squares, this means that $n$ is $(x+a)^2-x^2$ for some integer $x$.

share|improve this answer
Since $a-1$ is even it follows that $n$ is even also....,so I don't understand how $n$ can be $(x+a-1)^2-x^2$ –  pedja Jan 10 '12 at 16:13
@pedja -- oops, off-by-one error. I was counting numbers instead of steps-between-numbers. Fixed. –  Henning Makholm Jan 11 '12 at 19:34

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.