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

Prove that $x^2 + y^2 = z^4$ has infinitely many solutions with $(x,y,z)=1$.

Do I use the terms $x= r^2 - s^2$, $y = 2rs$, and $z = r^2 + s^2$ to prove this problem?

Thanks for any help.

share|cite|improve this question
While it is true that any solution can be expressed with $x=r^2-s^2$, $y=2rs$, and $z^2=r^2+s^2$ (note the square in $z$) because $x$, $y$, and $z^2$ are a primitive pythagorean triple, not every pair of values of $r$ and $s$ (with $r\gt s$, $\gcd(r,s)=0$, and $r$ and $s$ of opposite parity) will yield a solution, since you need $r^2+s^2$ to be a square. But you can then implement the formula again to find solutions to $z^2=r^2+s^2$, and then... – Arturo Magidin Jul 18 '12 at 19:14
Let $x+yi = (a+bi)^4$ – Thomas Andrews Jul 18 '12 at 19:24
What do you mean with "$(x,y,z)=1$"? – celtschk Jul 18 '12 at 20:23
@celtschk: It means $x,y,$ and $z$ are relatively prime. – Brandon Carter Jul 18 '12 at 20:27

Hint Let $w=z^2$. Then


Solve this Pytagorean equation, and prove that infinitely many $w$ are perfect squares.

With your notation $w=r^2+s^2$ so you need $r,s$ to satisfy $r^2+s^2=z^2$...

share|cite|improve this answer
N.S I want to adapt your answer to prove $x^2 + y^2 = z^4$ has infinitely many "primitive" solutions with x,y,z natural numbers. I'm just unsure what you mean by "so you need r,s to satisfy $r^2 +s^2=z^2...$ " – jakey Nov 16 '13 at 21:26
@user109323 The general primitive solution to the Pytagorean equation $x^2+y^2=w^2$ is $x=2rs, y=r^2-s^2$ and $w=r^2+s^2$. Now if you want $w=z^2$ you get $x=2rs, y=r^2-s^2$ and $z^2=r^2+s^2$. Since you need $z$ to be an integer, the equation $z^2=r^2+s^2$ must have integral solution. This is again the Pytagorean equation. Solve it, and then replace the $s$ and $r$ you get in $x=2rs, y=r^2-s^2$.... – N. S. Nov 16 '13 at 23:06
I can't add comments so have to put this in an answer each time, rather than awkwardly try and discuss it here, I would be extremely thankful if you could check out my response here :… thanks! -user109323 with a more human username, Jakey – jakey Nov 17 '13 at 13:16

Actually, a very strong statement is true, and quite easy: given any positive integer $n,$ such that all prime factors $p$ of $n$ satisfy $p \equiv 1 \pmod 4,$ then there is a solution to $$ x^2 + y^2 = n $$ with $$ \gcd(x,y) = 1.$$

For $n$ a prime (so itself $1 \pmod 4$) the existence of a solution was finally proved by Euler. After that, there is a simple mathematical induction, using the two choices in someone's formula (Brahmagupta with $N=-1$), anyway $$ (a^2 + b^2)(c^2 + d^2) = (ac+bd)^2 + (ad-bc)^2, $$ or $$ (a^2 + b^2)(c^2 + d^2) = (ac-bd)^2 + (ad+bc)^2 $$ with the induction hypothesis $$ \gcd(a,b) = \gcd(c,d) = 1. $$

Indeed, given such an $n,$ with $r$ distinct prime factors and each $p_i \equiv 1 \pmod 4,$ but we otherwise ignore the exponents, the number of representations $$ x^2 + y^2 = n $$ with $$ 0 < x < y, \; \; \gcd(x,y) = 1 $$ is exactly $$ 2^{r-1}. $$ For example, about ignoring exponents, we get $$ 1 +4 = 5, $$ $$ 9+16 = 25 = 5^2, $$ $$ 4 + 121 = 125 = 5^3, $$ $$ 49 + 576 = 625 = 5^4. $$

Alright, nobody is paying attention to this one, but I thought I ought to make sure I knew how the induction proof went. It is easier if we demand the second factor prime, so we are taking $$ a^2 + b^2 = n, \; \; \gcd(a,b) = 1,$$ $$ c^2 + d^2 = p, \; \; p \equiv 1 \pmod 4, \; \; \mbox{of course} \; \; \gcd(c,d) = 1. $$ We are going to represent $np.$ One way is $$ np = (ac+bd)^2 + (ad-bc)^2. $$ Suppose these are not relatively prime, there is some positive prime $q$ such that $$ q | ac+bd \; \; \mbox{and} \; \; q | ad-bc. $$ Multiply the first by $d$ and the second by $c$ and subtract, we get $$ q | b (c^2 + d^2), \; \; \mbox{or} \; \; q | b p. $$ Back to our pair, multiply the first by $c$ and the second by $d$ and add, we get $$ q | a (c^2 + d^2), \; \; \mbox{or} \; \; q | a p. $$ Since $ \gcd(a,b) = 1, $ it follows that $q |p$ and then that $q=p.$ So the only possible GCD is a power of $p$ itself. If, in addition, $$ p | ac-bd \; \; \mbox{and} \; \; p | ad+bc, $$ we get $ p | 2ac $ and $p | 2bc,$ or $p | a$ and $p|b.$ This is false, as $\gcd(a,b) = 1.$

To put it briefly, the only possible obstacle to a proper representation of $np$ is a common factor of $p$ itself, but if $p$ divides all four of $ac+bd, \; ad-bc, \; ac-bd, \; ad+bc, $ then it divides both $a,b,$ a contradiction. So, at least one of the four given expressions is not divisible by $p,$ and the pair using that expression gives a proper/primitive representation of $np$ as the sum of two squares. Naturally, the other expression in the successful pair is also prime to $p.$

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.