Intersections of binary quadratic forms in primes and the paucity phenomenon (Q2668924)

In this present work the author investigates positive solutions to \(a^{2}+b^{2}=c^{2}+d^{2}\) when \textit{some} of \(a, b, c, d\) are primes, particularly, when only \(c\) is a prime; when both \(c, d\) are primes and, finally, \(b, c, d\) are primes. The results obtained confirm that there is no paucity of the non-diagonal solutions in the first two cases. Denote by \(\mathbb{P}\) the set of all prime numbers, where the letters \(p, q, r\) represent elements of \(\mathbb{P}\). For natural number \(n\in \mathbb{N}\) define the following functions \(r_{0}(n)=\#\{(a,b)\in \mathbb{N}^{2}:n=a^{2}+b^{2}\},\) \(r_{1}(n)=\#\{(a,p)\in \mathbb{N}\times \mathbb{P}:n=a^{2}+p^{2}\},\) \(r_{2}(n)=\#\{(p,q)\in \mathbb{P}^{2}:n=p^{2}+q^{2}\},\) so that for \(i, j\in\{0,1,2\}\) the mean values can be obtained via the function \(S_{i,j}(x)=\displaystyle \sum_{n\leq x}r_{i}(n)r_{j}(n)\). Here the asymptotic formulas for \(S_{i,j}(x)\) can be interpreted in terms of the number of solutions to \(a^{2}+b^{2}=c^{2}+d^{2}\), where some of \(a, b, c, d\in \mathbb{N}\) are restricted to primes, depending on the values \(i, j.\) In the present work the author considers \(S_{i,j}(x)\) with \(i\neq j;i, j\in\{0,1,2\}\), with the main results stated below. Theorem 1. As \( x\rightarrow\infty\) we have that \(S_{0,1}(x)=\displaystyle \sum_{n\leq x}r_{0}(n)r_{1}(n)=\frac{x}{2}+O\left (\frac{x\log\log x}{\log x}\right)\) . Corollary 1. For any \(c>0\) the dispersion satisfies \(\displaystyle \sum_{n\leq x}\left(r_{1}(n)-\frac{cr_{0}(n)}{\log n}\right)^{2}\asymp\frac{x}{\log x}.\) Theorem 2. As \( x\rightarrow\infty\) we have \(S_{0,2}(x)=\displaystyle \sum_{n\leq x}r_{0}(n)r_{2}(n)=\frac{12G}{\pi^{2}}\frac{x}{\log x}+O\left(\frac{x\log\log x}{\log^{2}x}\right)\) , where \(G\) is the Catalan constant given by \( G=L(2,\displaystyle \chi_{4})=\sum_{k=0}^{\infty}\frac{(-1)^{k}}{(2k+1)^{2}}=0.915\ldots\) Theorem 3. As \( x\rightarrow\infty\) we have that \(\displaystyle \frac{2\pi x}{\log^{2}x}\) << \(S_{1,2}(x)=\displaystyle \sum_{n\leq x}r_{1}(n)r_{2}(n)\) << \(\displaystyle \frac{x}{\log^{2}x}(\log\log x)^{2}\).