The Hilbert-Kamke problem in analytic number theory (Q1089375)

The Hilbert-Kamke problem is concerned with the solvability and the number of solutions of the system \(x^ s_ 1+... +x^ s_ k=N_ s,\) \(s=1,...,n\), in nonnegative integers \(x_ i\). It is easily seen that the numbers \(N_ s\) must satisfy certain necessary conditions. So the actual question is whether these conditions are sufficient for the solvability for suitable \(k=k(n).\) The author surveys the history of this problem (starting from Waring's problem) with a bibliography of 66 references. The existence of k(n) was proved by E. Kamke in 1921 by Hilbert's method. Subsequent developments were connected with the exponential sum method of I. M. Vinogradov, invented by him in the thirties and developed in recent years notably by the author and his school. Denote by r(n) the least number k such that the Hilber-Kamke problem has a solution for sufficiently large ''admissible'' values of the \(N_ s\), and let \(r_ 1(n)\) be analogously defined if the \(x_ i\) are restricted to primes. It was proved by \textit{G. I. Arkhipov} in 1980 that \(2^ n-1\leq r(n)\leq 3n^ 3 2^ n-n,\) and by \textit{V. N. Chubarikov} in 1984 that \(2^ n-1\leq r_ 1(n)\leq 2^{20} n^ 3 4^ n.\)