The Gauss-Wilson theorem for quarter-intervals (Q2439822): Difference between revisions

Let \(N_n!=\prod_{_{\substack{ 1\leq j\leq N\\ \text{gcd}(j,n)=1}}} j\) be the Gauss-factorial. In the paper under review, the authors study the positive integers \(n\) with \(n\equiv 1\pmod 4\), having the property that the multiplicative order of \((\frac{n-1}{4})_n!\) modulo \(n\) is a power of \(2\). If \(p\) is a prime with the above property, then \(p\) is called a Gauss prime. The authors show that if \(p>5\) is a prime which is congruent to \(1\) modulo \(4\), then \((\frac{p-1}{4})_p!\) cannot have multiplicative order \(1,~2,~4\) or \(8\) modulo \(p\). The above order is \(16\) exactly when \(p=a^2+b^2\) and \(p-1=4ab\). Such primes are exactly the ones appearing in the sequence \(\{u_k\}_{k\geq 0}\) whose initial terms are \(1,~17,~241\) and which satisfies the three term linear recurrence whose characteristic polynomial is monic and has roots at \((2\pm {\sqrt{3}})^2\) and \(1\). The authors also show that if the \(n\)th Fermat number \(F_n=2^{2^n}+1\) is prime, then it is a Gauss prime and the multiplicative order of \((\frac{F_n-1}{4})_{F_n}!\) modulo \(F_n\) is \(2^{n+2}\). The main result of the paper is a characterization of all such \(n\) of the form \(n=p^{\alpha} w\), where \(p\equiv 1\pmod 4\) is prime, \(\alpha\geq 1\), and all prime factors of \(w\) are congruent to \(3\) modulo \(4\). The paper contains several numerical examples including a list of all Gauss primes \(p<10^{14}\).