6
$\begingroup$

For any $r\in\mathbb{N}$, let $A_r$ denote the set of all natural numbers that are potentially a side of a Pythagorean triple with hypotenuse $r$.

Given any $N\in\mathbb{N}$, does there exist $r,s$ such that $A_r\cap A_s$ contains at least $N$ elements?

I've been looking at the parameterization of Pythagorean triples as $k(m^2-n^2,2mn,m^2+n^2)$, and considering numbers $r$ that can be written as the sum of two squares in a large number of ways. But the components of this sum seem rather unpredictable, and I'm unsure how to proceed in this fashion.

$\endgroup$

2 Answers 2

6
$\begingroup$

Yes, given any $N \in \mathbb{N}$ there do exist $r$ and $s$ so that $A_{r} \cap A_{s}$ contains at least $N$ elements. In fact, I will show that one can take $s = 6r$. (The approach I suggest below is definitely not the most efficient, nor is it along the lines you were pursuing.)

We want to find values of $r$ so that the equations $$ C : a^2 = r^2 - b^2, \quad a^2 = 36r^2 - c^2 $$ have at least $N$ integer solutions for $a$. This system of two quadratic equations is isomorphic to the elliptic curve $$ E : y^2 = x^3 - x^2 - 420x - 3168, $$ which has infinitely many rational solutions. A generator for the group of rational points is $P = (69,540)$, and this generator maps to the solution $145^2 = 144^2 + 17^2$, $(6 \cdot 145)^2 = 144^2 + 858^2$, showing that $A_{145}$ and $A_{870}$ have (at least) one element in common. The point $2P$ is $(14113/576,352495/13824)$ and this shows that $A_{430106689}$ and $A_{2580640134}$ also have an element in common (namely $101518560$). Taking the least common multiple of $145$ and $430106689$ shows that for $r = 145 \cdot 430106689$ there are at least two elements of $A_{r} \cap A_{6r}$.

Using the points $P$, $2P$, $\ldots$, $NP$, finding their images on $C$, and taking the least common multiple of the corresponding values of $n$ will give a value of $r$ so that $A_{r} \cap A_{6r}$ will have size at least $2N$.

$\endgroup$
1
  • $\begingroup$ I had a suspicions that Eliptic curves would be involved. Unfortunately, I havent studied them since My early undergraduate years, so your comment is a bit out of my area. But it seems entirely reasonable and correct, so I'll do some reading on this subject. Thank you very much! (I will vote once I am able, when my account is officially confirmed) $\endgroup$
    – G. Flowers
    Mar 13, 2014 at 4:14
2
$\begingroup$

Take integers that are sums of two squares; this set is multiplicatively closed because of the identity $(a^2+b^2)(c^2+d^2) = (ad-bc)^2 + (ac+bd)^2$

Now interchanging $a$ and $b$, the new rhs gives another expression for the same LHS as sum of two squares. Now this explodes when a number is a product of many terms which are individually sum of two squares. And primes 1 mod 4 are by Fermat sum of two squares, and a number that is a product of many such primes will provide examples you need.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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