The cross-entropy method for combinatorial and continuous optimization (Q1961837)

A new and fast method, called the cross-entropy method, for finding an optimal solution of problems of the form: \[ \min_x L(x),\quad x\in X\in\mathbb{R}^n, \] where \(X\) is a convex bounded domain, \(L(x)\) is a bounded nonconvex function defined on \(X\). It is also allowed that either \(L(x)\) or \(X\) or both have a discrete (combinatorial) structure. The author presents a fast randomized adaptive algorithm for solving this problem in the sense that the algorithm converges with very high probability to a very small neighborhood of an optimal solution. Numerical tests for both continuous and combinatorial optimization problems are given.