5
$\begingroup$

Let p>3 be a prime. For every a=1,2,...,p-1 let x(a) be the number in {1,...,p-1} such that ax(a) is congruent to 1 mod p (its inverse mod p). Let S be the sum of ax(a)^2 and let T be the sum of x(a) (in both cases sum over all a=1,2,...,p-1). Is it true that S-2T is always divisible by p^2? I noted this by computing for small values of p and I wonder if this is a known theorem or if there is a simple proof.

$\endgroup$

1 Answer 1

11
$\begingroup$

This follows from well-known results (although perhaps not in a totally straightforward way).

Let us write $x(a) = a^{-1} + p y(a)$, where $a^{-1}$ denotes the inverse of $a$ modulo $p^2$, and $y(a) \in \{0, \dots, p-1\}$. Then we find that $$ a x(a)^2 = a^{-1} + 2p y(a) \pmod{p^2} $$ so $$ S = \sum_{a = 1}^{p-1} a^{-1} + 2p \sum_{a = 1}^{p-1} y(a) \pmod{p^2} $$ while $$ T = \sum_{a = 1}^{p-1} a^{-1} + p \sum_{a = 1}^{p-1} y(a) \pmod{p^2}. $$

So $p^2 \mid S - 2T$ if and only if $\sum_{a = 1}^{p-1} a^{-1} = 0 \pmod {p^2}$, and this latter statement is a consequence of Wolstenholme's theorem.

$\endgroup$

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.