Primes in tuples and Romanoff's theorem
Artyom Radomskii
TL;DR
This work proves a quantitative lower bound for the number of primes in tuples of linear forms and derives Romanoff-type representation consequences. It builds on the Maynard sieve framework, augmented by well-distributed set hypotheses and Bombieri–Vinogradov-type distribution controls, to show that for $x\ge10$ and modest growth of $k$, there exists a prime $p_0$ such that, for a family of admissible linear forms $L_i(n)=a_i n+b_i$ with $(a_i,b_i)=1$, $a_i\le \exp(c_0\sqrt{\ln x})$, $(a_i,p_0)=1$ and $b_i\le x\exp(c_0\sqrt{\ln x})$, one has a positive lower bound on the count of $n\in[x,2x)$ for which at least $C^{-1}\ln k$ of the $L_i(n)$ are prime, namely at least $x/(\ln x)^k\exp(Ck)$. The authors extend the result to a related setting (Theorem T2) via a linear transformation and obtain Romanoff-type density estimates for representation functions $f_{\\mathcal{A}}(n)$. Central to the argument are three preparatory lemmas that produce an admissible subfamily, establish a BV-type distribution bound around a carefully chosen prime $B$, and verify the required hypotheses for the sieve framework. Together, these results advance quantitative understanding of primes in linear patterns and have implications for Romanoff-type representations, connecting sieve methods with density questions for primes in arithmetic structures.
Abstract
We obtain a lower bound for a number of primes in tuples. As applications, we obtain a lower bound for the Romanoff type representation functions.
