Application of decomposition method to linear market equilibrium model

Abstract

In this thesis, we study the modular-based framework of combining econometric submodels characterizing the final demands, and linear programming process analysis models and present the algorithm for computing market equilibria of the combined system. The suggested algorithm stems from the incorporation of the $\underline{decentralized}$ inte-gration of submodels with the $\underline{hybrid}$ decomposition method, where the decentralized integration was deviced to be processed without explicit master problem with contrast to the traditional decomposition methods, and the hybrid docomposition was done to use both of price- and resource- directive decomposition techniques for the rapid convergence. So this algorithm is done without explicit master problem and with both price -and resource- directive decomposition methods. This algorithm has the shape of Gauss-Seidal iteration algorithm. The convergence properties have been investigated.