Sum of one prime and two squares of primes in short intervals (Q746927)

It is believed that every sufficiently large odd integer \(n\) such that \(n\not\equiv2\pmod{3}\) can be represented as a sum of a prime and two squares of primes. [\textit{L. Zhao}, Mich. Math. J. 63, No. 4, 763--779 (2014; Zbl 1360.11092)] proved that the cardinality \(E(N)\) of the set of positive odd integers \(n\leq N\) not congruent to \(2\pmod{3}\) that are not representable as a sum of a prime and two prime squares satisfies \(E(N)\ll N^{1/3+\varepsilon}\) with arbitrarily small \(\varepsilon>0\). In fact, he also proved that if \(r(n)=\sum_{p_1+p_2^2+p_3^2=n}(\log p_1)(\log p_2)(\log p_3)\) then a suitable asymptotic formula for \(r(n)\) holds for every odd positive integer \(n\) not congruent to \(2\pmod{3}\) except for \(N^{1/3+\varepsilon}\) exceptions. In the paper under review, the authors study the average behavior of \(r(n)\) over short intervals \([N, N +H]\) where \(H = o(N)\). They show that assuming the Riemann hypothesis a suitable formula for such an average of \(r(n)\) is true in short intervals without exceptions. Namely, if RH is true then \(\sum_{n=N+1}^{N+H} r(n)=(\pi/4)HN+O\left(H^{1/2}N(\log N)^2+HN^{3/4}(\log N)^3+H^2(\log N)^{3/2}\right)\) as \(N\to\infty\) uniformly for \(\infty((\log N)^4)\leq H\leq o(N(\log N)^{-3/2})\), where \(f=\infty(g)\) stands for \(g=o(f)\). Consequently, assuming the Riemann hypothesis, every interval \([N,N+H]\) contains an integer which is sum of a prime and two prime squares provided \(CL^4\leq H=o(NL^{-3/2})\) with \(C>0\) a suitable large constant and \(L=\log N\). The proof of the main results employs the original setting of Hardy-Littlewood circle method working with infinite weighted exponential sums \(\sum_{n=1}^\infty \Lambda(n)e^{-n^\ell/N}e(n^\ell\alpha)\).