Title:Computing the Expected Values of some Properties of Randomly Weighted Graphs

Abstract: Consider the setting of \emph{randomly weighted graphs}, namely, graphs whose edge weights are independent discrete random variables with finite support over the non-negative reals. Given a randomly weighted graph $G$, we are interested in computing the expected values of various graph properties of $G$. In particular, we focus on the problem of computing the expected diameter of $G$. It is easy to show that this problem is \SharpP-hard even in the restricted case in which all edge weights are identically distributed. In this paper we prove that this problem admits a \emph{fully polynomial time randomized approximation scheme (FPRAS)}. Our technique can also be used to derive an FPRAS for the problem of computing the expected weight of an MST of $G$.