Given a natural number $n$, can we completely characterize when $n$ and $2n$ are each a sum of two squares?

For example: $446,382,709=(13010)^{2}+(16647)^{2}$ and $892,765,418=2(446,382,709)=(3637)^{2}+(29657)^{2}$

Do these imply, perhaps, that 446,782,709 is the hypotenuse of a primitive pythagorean triple?

(I found this question on a slip of paper while cleaning my office. It turns out that a trivial algebraic identity resolves the question...beyond Euler's characterization of when a natural number is the sum of two squares. Is the question nontrivial if we replace $2n$ by $3n$ above?)

EDIT: The modification is trivial, too. This is my fault, as I didn't think about my question before posting it. (This was a problem found on a slip of paper in my office. I posted it because it looked like a fun problem for students. This is certainly not the best forum for such things!!!)

  • $\begingroup$ Oddly, though, people like this problem more than many of the ones I HAVE thought about and posted on MO. Perhaps I should take a hint! :) $\endgroup$ – Jon Bannon Mar 9 '11 at 1:03

A natural is the sum of two squares iff every prime of the form $\rm\:4\:k+3$ occurs to even power in its prime factorization (iff the same is true for $\rm\:2\:n\:$). Note $\rm\ n = x^2 + y^2\ \Rightarrow\ 2\:n = (x+y)^2 + (x-y)^2$ which arises by compositions of forms, or, in linear form, with the norm $\rm\ N(a+b\ i)\ =\ a^2 + b^2 $

$$\rm 2\ (x^2+y^2)\ =\ N(1+i)\ N(x-y\ i)\ =\ N((1+i)\ (x-y\ i))\ =\ N(x+y + (x-y)\ i) $$

  • $\begingroup$ Silly me! Simple algebraic identity gets them all... $\endgroup$ – Jon Bannon Jan 3 '11 at 14:46
  • $\begingroup$ Hi Bill! Is there anything interesting to be said if we replace 2n by 3n above? $\endgroup$ – Jon Bannon Jan 3 '11 at 14:54
  • 1
    $\begingroup$ @Jon: at most one of n or 3n can be written as a sum of squares, but not both. This follows from the criterion Bill gives. It sounds like you are not aware of the classical theory of factorization in the Gaussian integers: en.wikipedia.org/wiki/Gaussian_integer $\endgroup$ – Qiaochu Yuan Jan 3 '11 at 14:56
  • 1
    $\begingroup$ @Jon: HINT $\ $ Apply the mentioned criterion noting that the exponents of $\:3\:$ in the prime factorizations of $\rm n$ and $\rm 3n$ have opposite parity. $\endgroup$ – Bill Dubuque Jan 3 '11 at 15:00
  • $\begingroup$ Hi Qiaochu. Thanks for the link! I posted the question because it looked like fun for youngsters. I must confess I hadn't thought at all about it before posting it. $\endgroup$ – Jon Bannon Jan 3 '11 at 15:00

$2(a^2 + b^2) = (a + b)^2 + (a - b)^2$. And, of course, the classification of numbers which are the sums of two squares is well-known (it is necessary and sufficient that if $p \equiv 3 \mod 4$, then the largest $k$ such that $p^k | n$ is even).

  • 1
    $\begingroup$ Thanks, Qiaochu! Bill got there first, though. $\endgroup$ – Jon Bannon Jan 3 '11 at 14:48

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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