On exceptional sets of biquadratic Waring-Goldbach problem (Q2308831)

The Waring-Goldbach problem is concerned with the solubility of the equation \[ n=p_1^k+\cdots +p_s^k \] in primes \(p_1,\dots,p_s\). It is conjectured that for given \(k,s\in \mathbb{N}\), there exists a fixed modulus \(q_{k,s}\) and a (non-empty) collection \(\mathcal{N}_{k,s}\) of residue classes modulo \(q_{k,s}\) such that every sufficiently large integer \(n\in \mathcal{N}_{k,s}\) can be represented in the form above, provided that \(s\ge k+1\). This conjecture is out of reach of current methods but significant progress has been made over the time. The paper under review is about the sizes of the exceptional sets of \(n\)'s in \(\mathcal{N}_{k,s}\) for which there is no representation as above. Let \(E_{k,s}(N)\) be the number of such exceptional \(n\)'s not exceeding \(N\). \textit{A. V. Kumchev} [Can. J. Math. 57, No. 2, 298--327 (2005; Zbl 1080.11070)] obtained estimates of the form \[ E_{k,s}(N)\ll N^{e(k,s)+\varepsilon} \] with concrete \(e(k,s)<1\) for the pairs \((k,s)=(4,s)\) with \(7\le s\le 12\). The authors improve these estimates by combining methods of \textit{L. Zhao} [Proc. Lond. Math. Soc. (3) 108, No. 6, 1593--1622 (2014; Zbl 1370.11116)] and \textit{K. Kawada} and \textit{T. D. Wooley} [J. Lond. Math. Soc., II. Ser. 82, No. 2, 437--458 (2010; Zbl 1279.11097)]. Along similar lines, new estimates for certain pairs \((5,s)\) are obtained as well. As usual in this area, the basic approach is the circle method, and the technical details become rather complicated.