The lattice points of an \(n\)-dimensional tetrahedron (Q5896604)

Let \textbf{w}\(=(w_ 1,...,w_ n)\) and \textbf{x}\(=(x_ 1,...,x_ m)\) be vectors of positive real numbers, and let \(k\geq 2\) be an integer. Consider m by n matrices \textbf{X} with non-negative integer entries \(X_{ij}\) such that each column contains at most k non-zero elements. The primary result of the paper is an asymptotic formula for the number of such matrices for \({\mathbf {Xw}}\leq {\mathbf{x}}.\) Such inequalities, with \textbf{w} consisting of the logarithms of the first few primes, occur in considerations of the first case of Fermat's Last Theorem, when one uses generalizations of the Wieferich and Mirimanoff congruences, as in Gunderson's thesis. The most recent such work, by \textit{D. Coppersmith} [Math. Comput. 54, 895-902 (1990; Zbl 0701.11008)] relates to the inequality \(m^ 2+n^ 2\leq x\) in coprime integers m,n with prime factors at most \(y\ll (\log x)^{1/2}\). The present paper gives, in particular, an asymptotic estimate for the number of such m,n.