Consecutive primes modulo 4 (Q1866442)

Let \(\pi_{ij}(x)\) denote the number of pairs of primes with \(p_n \equiv i \bmod 4\) and \(p_{n+1} \equiv j \bmod 4\). \textit{S. Knapowski} and \textit{P. Turán} [Number theory and algebra, Academic Press, New York, 157-165 (1977; Zbl 0374.10026)] gave a lower bound of the form \((\log x)^c\) for \(\pi_{11}(x)\). \textit{D. Shiu} [J. Lond. Math. Soc. (2) 61, 359-373 (2000; Zbl 0973.11083)] proved a very general result on consecutive primes which implies in particular that \(\pi_{11}(x) \gg x^{1-\varepsilon(x)}\), where \(\varepsilon (x)= c( \frac{\log \log \log x}{\log \log x})^{1/2}\). The present paper improves this to the order of \(\frac{x \log \log x}{(\log x)^2}\). More generally the author proves: Let \(q \geq 3\) be an integer, and let \(A\) be a set of reduced residue classes modulo \(q\) satisfying \(|A |= \varphi(q)/2\). For \(x> x_0(q)\) one has \[ |\{ n: p_{n+1}\leq x,\;p_n \;\text{ and } \;p_{n+1} \in A \bmod q \}|\geq c_0 \frac{\varphi(q)}{q^3} \frac{x\log \log x}{(\log x)^2}, \] with an absolute constant \(c_0>0\). The paper ends with a discussion of two contradicting conjectures. Knapowski and Turán conjectured that \(\pi_{11}(x)=o(x/\log x)\). \textit{P. Erdős} [Collected papers of Paul Turán. Volume 1. Akadémiai Kiadó (1990; Zbl 0703.01019)] conjectured that \(\pi_{11}(x) \sim x/(4\log x)\). The paper under review presents some results that suggest that the Knapowski and Turán conjecture might be wrong.