COMPUTATIONAL COMPARISONS OF SPATIAL PRICE EQUILIBRIUM METHODS*

Abstract

ABSTRACT. In this paper we propose new algorithms for the solution of both general and standard spatial price equilibrium problems, and test their performance with existing algorithms on randomly generated problems. For the standard problem, we propose decomposition schemes based on the concept of “equilibration operator” and compare their performance with the Frank‐Wolfe method. For the general problem, we present alternative variational inequality formulations defined over Cartesian products of sets and then exploit these formulations to construct Gauss‐Seidel‐type serial decomposition methods. We then compare their performance with the projection method. Our computational tests suggest that the new schemes are substantially more efficient than earlier ones.