On a number theoretic conjecture on positive integral points in a 5-dimensional tetrahedron and a sharp estimate of the Dickman-de Bruijn function (Q479516): Difference between revisions

Let \(\Delta (a_1,\dots,a_n)\) be an \(n\)-dimensional simplex \[ \frac{x_1}{a_1}+\dots +\frac{x_n}{a_n}\leq 1 ,\quad x_1,\dots ,x_n\geq 0 , \] where \(a_1\geq a_2\geq \dots \geq a_n\geq 1\) are real numbers and let \(P_n=P(a_1,\dots ,a_n)\) and \(Q_n=Q(a_1,\dots ,a_n)\) be the numbers of positive and nonnegative integral solutions of the above inequality. The number \(Q_n\) of integer lattice points inside the \(n\)-dimensional tetrahedron \(\Delta (a_1,\dots ,a_n)\) with \(a_i=\frac{\log x}{\log p_i}\), \(1\leq i\leq n\), where \(p_1<\dots <p_n\) denote all the primes up to \(y\) is precisely \(\psi (x, y)\); here \(\psi (x, y)\) is the number of positive integers \(\leq x\) composed only of prime factors \(\leq y\). In the article the authors (more precisely, Stephen) formulate the following conjecture: if \(a_1\geq \dots \geq a_n>1, n\geq 3, P_n>0\) then \[ n!P_n\geq (a_1-1)\dots (a_n-1)-(a_n-1)^n+a_n\big(a_n-1\big)\dots \big(a_n-(n-1)\big) \] and equality holds if and only if \(a_1= \dots = a_n=\) integer. In the case \(n=5\) the above conjecture is proved. Also a sharp estimate of the function \(\psi (x, y)\) for \(5\leq y<13\) is obtained.