An Alternating Algorithm for Finding Linear Arrow-Debreu Market Equilibria

Motivated by the convergence result of mirror-descent algorithms to market equilibria in linear Fisher markets, it is natural for one consider designing dynamics (specifically, iterative algorithms) agents arrive at Arrow-Debreu equilibria. Jain (SIAM J. Comput. 37(1), 303–318, 2007) reduced equilibrium computation markets bijective where everyone a seller only good and buyer bundle goods. In this paper, we design an algorithm computing equilibrium, based on solving rational convex program formulated Devanur et al. The repeatedly alternates between step gradient-descent-like updates distributed optimization exploiting property such program. Convergence can be ensured new analysis different from