A generalization on the difference between an integer and its inverse modulo \(q\). II (Q2577609): Difference between revisions

Let \(q>2\) and \(c\) be two integers with \((q,c)=1\). For each integer \(0<a<q\) with \((a,q)=1\) define \(b\) by the congruence \(ab\equiv c\pmod q\) and \(0<b<q\). The authors study the asymptotic behaviour of the function \[ M(q,k,c)=\sum_{\substack{ 1\leq a,b\leq q\\(a,q)=1,ab\equiv c\pmod q}}(a-b)^{2k}. \] They prove for any positive integer \(k\) \[ M(q,k, c)=\frac{\varphi(q)q^{2k}} {2k+1}+O\bigl(4^kq^{2k+ \tfrac12}d^2 (q) \ln q\bigr), \] where \(\varphi(q)\) is the Euler function and \(d(n)\) the divisor function. They replace \(a\) by \(a_1,a_2,\dots,a_n\) and sum over each \(a_i\), \(1\leq a_i\leq q\), \((a_i,q)=1\), to get a more general formula.