On some results of Hua in short intervals (Q619351): Difference between revisions

As the authors summarize, they prove results of Goldbach-Waring type (representation of integers as a sum of powers of primes) in ``short intervals'', namely, with prime variables restricted in intervals with their lengths which go to infinity in a slower fashion w.r.t. the endpoints. It starts with a short history of the subject, quoting the famous 3-primes theorem of Vinogradov (say, the solution of ternary Goldbach problem, compare it with the very recent breakthrough by \textit{H. A. Helfgott} (see, e.g., [``Minor arcs for Goldbach's problem'', Preprint, \url{arxiv:1205.5252}; ``Major arcs for Goldbach's theorem'', Preprint, \url{arxiv:1305.2897}; with D. J. Platt, ``Numerical verification of the ternary Goldbach conjecture up to 8.875e30'', Preprint, \url{arxiv:1305.3062}) and the variations on the theme, namely with square (and prime-powers) of primes as summands, introduced by \textit{L. K. Hua} (relying on his very famous non-trivial bounds for exponential sums over squares of prime numbers and prime-powers), see for example his beautiful book [Additive Theory of Prime Numbers. Providence, RI: AMS (1965; Zbl 0192.39304)]. They solve the particular problem to represent the large integer \(N\) (constrained by the natural congruence condition of the problem at hand) as a sum of \(4\) squares of primes plus a \(k\)th power of a prime; this, with the additional requirement that all of these primes are within a short interval. In order to be more precise (however, see the paper for the exact statement) \[ N=p_1^2+p_2^2+p_3^2+p_4^2+p^k, \quad |p_j-\sqrt{N/5}|\leq U, \forall j\leq 4, \enspace |p-(N/5)^{1/k}|\leq UN^{-1/2+1/k} \tag{*} \] \noindent holds (if \(N\) satisfies quoted conditions) for \(U:=N^{1/2-\eta_1+\varepsilon}\), where \(\eta_1=\eta_1(k)>0\), \(\forall k\geq 3\). The method that proves \((\ast)\) above (namely, their Thm. 1) is, in the authors' words (see the paper, also for the References [\textit{S. K. K. Choi} and \textit{A. V. Kumchev}, Acta Arith. 123, 125--142 (2006; Zbl 1182.11048)]; [\textit{J. Y. Liu}, Q. J. Math. 54, No. 4, 453--462 (2003; Zbl 1080.11071)]; [\textit{J. Y. Liu, G. S. Lü} and \textit{T. Zhan}, Sci. China, Ser. A 49, No. 5, 611--619 (2006; Zbl 1106.11033)] here): ``Our Theorem 1 depends on the following three ingredients: (1) a technique to get an asymptotic formula on the enlarged major arcs developed by Liu [Zbl 1080.11071]; (2) estimates for exponential sums over primes in short intervals established by Liu et al. [Zbl 1106.11033]; (3) a new estimate for Dirichlet polynomials introduced by Choi and Kumchev [Zbl 1182.11048].'' The same method enables them to prove (see Thm. 2) \[ \begin{multlined} N=p_1+p_2^2+p_3^2+p^k, \\ |p_1-N/4|\leq U\sqrt{N}, \quad |p_j-\sqrt{N/4}|\leq U (j=2,3), \enspace |p-(N/4)^{1/k}|\leq UN^{-1/2+1/k}\end{multlined} \tag{**} \] where \(N\) has now different congruence constraints and \(U:=N^{1/2-\eta_2+\varepsilon}\), with another \(\eta_2=\eta_2(k)>0\). Finally, combining a method introduced by \textit{T. D. Wooley} in 2002 [Proc. Lond. Math. Soc. (3) 85, No. 1, 1--21 (2002; Zbl 1039.11066)] they prove Theorem 3 that represents almost all (congruence constrained large enough) integers as, resp., a sum of two, resp., three squares of primes in short intervals and a \(k\)-th power \((k\geq 3)\), again with all prime variables in short intervals.