skip to main content
OSTI.GOV title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: Parallel and serial variational inequality decomposition algorithms for multicommodity market equilibrium problems

Abstract

The authors have applied parallel and serial variational inequality (VI) diagonal decomposition algorithms to large-scale multicommodity market equilibrium problems. These decomposition algorithms resolve the VI problems into single commodity problems, which are then solved as quadratic programming problems. The algorithms are implemented on an IBM 3090-600E, and randomly generated linear and nonlinear problems with as many as 100 markets and 12 commodities are solved. The computational results demonstrate that the parallel diagonal decomposition scheme is amenable to parallelization. This is the first time that multicommodity equilibrium problems of this scale and level of generality have been solved. Furthermore, this is the first study to compare the efficiencies of parallel and serial VI decomposition algorithms. Although the authors have selected as a prototype an equilibrium problem in economics, virtually any equilibrium problem can be formulated and studied as a variational inequality problem. Hence, their results are not limited to applications in economics and operations research.

Authors:
;
Publication Date:
Research Org.:
Univ. of Massachusetts, Amherst, MA (US)
OSTI Identifier:
6089821
Resource Type:
Journal Article
Resource Relation:
Journal Name: Int. J. Supercomput. Appl.; (United States); Journal Volume: 3:1
Country of Publication:
United States
Language:
English
Subject:
99 GENERAL AND MISCELLANEOUS//MATHEMATICS, COMPUTING, AND INFORMATION SCIENCE; PARALLEL PROCESSING; ALGORITHMS; ECONOMICS; MARKET; NONLINEAR PROBLEMS; PROGRAMMING; RESEARCH PROGRAMS; USES; MATHEMATICAL LOGIC; 990210* - Supercomputers- (1987-1989)

Citation Formats

Nagurney, A., and Kim, D.S. Parallel and serial variational inequality decomposition algorithms for multicommodity market equilibrium problems. United States: N. p., 1989. Web. doi:10.1177/109434208900300104.
Nagurney, A., & Kim, D.S. Parallel and serial variational inequality decomposition algorithms for multicommodity market equilibrium problems. United States. doi:10.1177/109434208900300104.
Nagurney, A., and Kim, D.S. 1989. "Parallel and serial variational inequality decomposition algorithms for multicommodity market equilibrium problems". United States. doi:10.1177/109434208900300104.
@article{osti_6089821,
title = {Parallel and serial variational inequality decomposition algorithms for multicommodity market equilibrium problems},
author = {Nagurney, A. and Kim, D.S.},
abstractNote = {The authors have applied parallel and serial variational inequality (VI) diagonal decomposition algorithms to large-scale multicommodity market equilibrium problems. These decomposition algorithms resolve the VI problems into single commodity problems, which are then solved as quadratic programming problems. The algorithms are implemented on an IBM 3090-600E, and randomly generated linear and nonlinear problems with as many as 100 markets and 12 commodities are solved. The computational results demonstrate that the parallel diagonal decomposition scheme is amenable to parallelization. This is the first time that multicommodity equilibrium problems of this scale and level of generality have been solved. Furthermore, this is the first study to compare the efficiencies of parallel and serial VI decomposition algorithms. Although the authors have selected as a prototype an equilibrium problem in economics, virtually any equilibrium problem can be formulated and studied as a variational inequality problem. Hence, their results are not limited to applications in economics and operations research.},
doi = {10.1177/109434208900300104},
journal = {Int. J. Supercomput. Appl.; (United States)},
number = ,
volume = 3:1,
place = {United States},
year = 1989,
month = 1
}
  • In this paper we present a new class of sequential and parallel algorithms for multicommodity transportation problems with linear and convex costs. First, we consider a capacitated multicommodity transportation problem with an orthogonal quadratic objective function. We develop two new solution methods. Both exploit the fact that a projection on the conservation of flow constraints has an explicit form which was proved in an early paper by I. Chabini and M. Florian. The two algorithms deal differently with the remaining constraints namely the non negativity and capacity constraints. We prove the convergence of both algorithms using a basic general theorymore » (developed by the author) which generalizes Bregman`s theory. The above algorithms can be extended for differentiable convex cost multicommodity transportation problems as follows. For strictly convex costs we use a variant of the projected gradient method. The quadratic proximal minimization algorithm is applied to the linear cost multicommodity transportation problems. For both cases we solve an orthogonal projection multicommodity transportation problem at each iteration. The algorithms developed are well-suited for a coarse grained parallelization. The different steps may be decomposed by nodes, by arcs and/or by commodities. We investigate different strategies depending on the structure of the problem, the number of commodities and the architecture of the parallel machine. We present computational results for these different approaches on parallel and serial platforms such as a network of Transputers or Sun workstations. Very large problems are solved. The parallel implementations are analyzed using especially a new measure of performance developed previously by I. Chabini and M. Florian.« less
  • The alternating direction method is an attractive method for solving large-scale variational inequality problems whenever the subproblems can be solved efficiently. However, the subproblems are still variational inequality problems, which are as structurally difficult to solve as the original one. To overcome this disadvantage, in this paper we propose a new alternating direction method for solving a class of nonlinear monotone variational inequality problems. In each iteration the method just makes an orthogonal projection to a simple set and some function evaluations. We report some preliminary computational results to illustrate the efficiency of the method.
  • In this paper we develop the convergence theory of a general class of projection and contraction algorithms (PC method), where an extended stepsize rule is used, for solving variational inequality (VI) problems. It is shown that, by defining a scaled projection residue, the PC method forces the sequence of the residues to zero. It is also shown that, by defining a projected function, the PC method forces the sequence of projected functions to zero. A consequence of this result is that if the PC method converges to a nondegenerate solution of the VI problem, then after a finite number ofmore » iterations, the optimal face is identified. Finally, we study local convergence behavior of the extragradient algorithm for solving the KKT system of the inequality constrained VI problem.« less
  • No abstract prepared.
  • High-fidelity simulation of nuclear reactor accidents such as the rupture of a main steam line in a pressurized water reactor (PWR) requires three-dimensional core hydrodynamics modeling because of the strong effect channel cross flow has on reactor kinetics. A parallel nested Krylov linear solver was developed and implemented in the RETRAN-03 reactor systems analysis code to make such high-fidelity core modeling practical on engineering workstations. Domain decomposition techniques were also applied to the RETRAN-03 solution algorithm and demonstrated using a distributed memory parallel computer. Applications were performed for a four-loop Westinghouse PWR steam-line-break accident, and performance improvements of over amore » factor of 30 were achieved for models with 25 flow channels in the core. Larger models (e.g., 104-core channels), previously inaccessible because of memory limitations, were also solved with practical execution times.« less