Simultaneous additive congruences to a large prime modulus (Q4024701)

Let \(r\), \(k\), \(n\) be positive integers with \(k>1\) and \(n>2rk\). \textit{E. Dörner} [J. Number Theory 36, 1--11 (1990; Zbl 0709.11027)] proved that there exists a bound \(p_0=p_0(r,k)\) such that every system of equations \[ a_{i1} x_1^k+\cdots+ a_{in} x_n^k =0, \qquad i=1,\ldots,r, \] with coefficients \(a_{ij}\in\mathbb{Z}\) has a non-trivial \(p\)-adic solution for all \(p>p_0(r,k)\). It is known that the bound \(n>2rk\) is best possible so it remains to obtain a bound for \(p_0(r,k)\). In this paper, with the help of some results in \textit{H. Davenport} and \textit{D. J. Lewis} [Philos. Trans. R. Soc. Lond., Ser. A 264, 557--595 (1969; Zbl 0207.35304)] the authors show that \(p_0(r,k)\leq k^{2r+2}\). It is interesting to notice that \textit{T. D. Wooley} [Mathematika 37, 85--96 (1990; Zbl 0707.11021)] has conjectured that \(p_0(r,k)=k^{2r+2}\).