Parallel successive overrelaxation methods for symmetric linear complementarity problems and linear programs

A parallel successive overrelaxation (SOR) method is proposed for the solution of the fundamental symmetric linear complementarity problem. Convergence is established under a relaxation factor which approaches the classical value of 2 for a loosely coupled problem. The parallel SOR approach is then applied to solve the symmetric linear complementarity problem associated with the least norm solution of a linear program.This work was sponsored by the United States Army under Contract No. DAAG29-80-C-0041. This material is based on research sponsored by National Science Foundation Grant DCR-84-20963 and Air Force Office of Scientific Research Grants AFOSR-ISSA-85-00080 and AFOSR-86-0172.on leave from CRAI, Rende, Cosenza, Italy.     

Multi-sweep asynchronous parallel successive overrelaxation for the nonsymmetric linear complementarity problem

Convergence is established for themulti-sweep asynchronous parallel successive overrelaxation (SOR) algorithm for thenonsymmetric linear complementarity problem. The algorithm was originally introduced in [4] for the symmetric linear complementarity problem. Computational tests show the superiority of the multi-sweep asynchronous SOR algorithm over its single-sweep counterpart on both symmetric and nonsymmetric linear complementarity problems.This material is based on research supported by National Science Foundation Grants CCR-8723091 and DCR-8521228, and Air Force Office of Scientific Research Grants AFOSR-86-0172 and AFOSR-86-0124.     

Asynchronous parallel successive overrelaxation for the symmetric linear complementarity problem

Convergence is established for asynchronous parallel successive overrelaxation (SOR) algorithms for the symmetric linear complementarity problem. For the case of a strictly diagonally dominant matrix convergence is achieved for a relaxation factor interval of (0, 2] with line search, and (0, 1] without line search. Computational tests on the Sequent Symmetry S81 multiprocessor give speedup efficiency in the 43%–91% range for the cases for which convergence is established. The tests also show superiority of the asynchronous SOR algorithms over their synchronous counterparts.This material is based on research supported by National Science Foundation Grants DCR-8420963 and DCR-8521228 and Air Force Office of Scientific Research Grant AFOSR-86-0172.     

Iterative linear programming solution of convex programs

An iterative linear programming algorithm for the solution of the convex programming problem is proposed. The algorithm partially solves a sequence of linear programming subproblems whose solution is shown to converge quadratically, superlinearly, or linearly to the solution of the convex program, depending on the accuracy to which the subproblems are solved. The given algorithm is related to inexact Newton methods for the nonlinear complementarity problem. Preliminary results for an implementation of the algorithm are given.This material is based on research supported by the National Science Foundation, Grants DCR-8521228 and CCR-8723091, and by the Air Force Office of Scientific Research, Grant AFOSR-86-0172. The author would like to thank Professor O. L. Mangasarian for stimulating discussions during the preparation of this paper.     

Interior proximal point algorithm for linear programs

An interior proximal point algorithm for finding a solution of a linear program is presented. The distinguishing feature of this algorithm is the addition of a quadratic proximal term to the linear objective function. This perturbation has allowed us to obtain solutions with better feasibility. Implementation of this algorithm shows that the algorithms. We also establish global convergence and local linear convergence of the algorithm.This research was supported by National Science Foundation Grants DCR-85-21228 and CCR-87-23091 and by Air Force Office of Scientific Research Grants AFOSR-86-0172 and AFOSR-89-0410. It was conducted while the author was a Graduate Student at the Computer Sciences Department, University of Wisconsin, Madison, Wisconsin.     

Parallel pivotal algorithm for solving the linear complementarity problem

We propose a parallel implementation of the classical Lemke's algorithm for solving the linear complementarity problem. The algorithm is designed for a loosely coupled network of computers which is characterized by relatively high communication costs. We provide an accurate prediction of speedup based on a simple operation count. The algorithm produces speedup nearp, wherep is the number of processors, when tested on large problems as demonstrated by computational results on the CRYSTAL token-ring multicomputer and the Sequent Balance 21000 multiprocessor.This material is based on research supported by National Science Foundation Grants DCR-84-20963 and DCR-850-21228 and by Air Force Office of Scientific Research Grants AFSOR-86-0172 and AFSOR-86-0255 while the author was at the University of Wisconsin, Madison, Wisconsin.     

A two-stage successive overrelaxation algorithm for solving the symmetric linear complementarity problem

We propose a two-stage successive overrelaxation method (TSOR) algorithm for solving the symmetric linear complementarity problem. After the first SOR preprocessing stage this algorithm concentrates on updating a certain prescribed subset of variables which is determined by exploiting the complementarity property. We demonstrate that this algorithm successfully solves problems with up to ten thousand variables.This material is based on research supported by National Science Foundation Grants DCR-8420963 and DCR-8521228 and Air Force Office of Scientific Research Grants AFSOR-86-0172 and AFSOR-86-0255 while the author was at the Computer Sciences Department at the University of Wisconsin-Madison, USA.     

Finite perturbation of convex programs

This paper concerns a characterization of the finite-perturbation property of a convex program. When this property holds, finite perturbation of the objective function of a convex program leads to a solution of the original problem which minimizes the perturbation function over the set of solutions of the original problem. This generalizes a finite-termination property of the proximal point algorithm for linear programs and characterizes finite Tikhonov regularization of convex programs.This material is based on research supported by National Science Foundation Grants DCR-8521228 and CCR-8723091 and Air Force Office of Scientific Research Grants AFOSR-86-0172 and AFOSR-89-0410.     

Finite termination of the proximal point algorithm

This paper concerns the notion of a sharp minimum on a set and its relationship to the proximal point algorithm. We give several equivalent definitions of the property and use the notion to prove finite termination of the proximal point algorithm.This material is based on research supported by National Science Foundation Grants DCR-8521228 and CCR-8723091, and Air Force Office of Scientific Research Grant AFOSR-86-0172.     

Error bounds for nondegenerate monotone linear complementarity problems

Error bounds and upper Lipschitz continuity results are given for monotone linear complementarity problems with a nondegenerate solution. The existence of a nondegenerate solution considerably simplifies the error bounds compared with problems for which all solutions are degenerate. Thus when a point satisfies the linear inequalities of a nondegenerate complementarity problem, the residual that bounds the distance from a solution point consists of the complementarity condition alone, whereas for degenerate problems this residual cannot bound the distance to a solution without adding the square root of the complementarity condition to it. This and other simplified results are a consequence of the polyhedral characterization of the solution set as the intersection of the feasible region {zMz + q 0, z 0} with a single linear affine inequality constraint.This material is based on research supported by National Science Foundation Grants CCR-8723091 and DCR-8521228 and Air Force Office of Scientific Research Grant AFOSR-86-0172.     

Partially and totally asynchronous algorithms for linear complementarity problems

A unified treatment is given for partially and totally asynchronous parallel successive overrelaxation (SOR) algorithms for the linear complementarity problem. Convergence conditions are established and compared to previous results. Convergence of the partially asynchronous method for the symmetric linear complementarity problem can be guaranteed if the relaxation factor is sufficiently small. Unlike previous results, this relaxation factor interval does not depend explicitly on problem size.This material is based on research supported by the Air Force Office of Scientific Research Grant No. AFOSR-89-0410.The author wishes to thank the referee for pointing out how to improve the bound (12). The same technique can be used to reduce the factorn in Ref. 5, p. 553, to .     

A dual exact penalty formulation for the linear complementarity problem

We consider a dual exact penalty formulation for the monotone linear complementarity problem. Tihonov regularization is then used to reduce the solution of the problem to the solution of a sequence of positive-definite, symmetric quadratic programs. A modified form of an SOR method due to Mangasarian is proposed to solve these quadratic programs. We also indicate how to obtain approximate solutions to predefined tolerance by solving a single quadratic program, in special cases.This research was sponsored by US Army Contract DAAG29-80-C-0041, by National Science Foundation Grants DCR-8420963 and MCS-8102684, and AFSOR Grant AFSOR-ISSA-85-0880.     

A parallel algorithm for generalized networks

This paper presents an application of parallel computing techniques to the solution of an important class of planning problems known as generalized networks. Three parallel primal simplex variants for solving generalized network problems are presented. Data structures used in a sequential generalized network code are briefly discussed and their extension to a parallel implementation of one of the primal simplex variants is given. Computational testing of the sequential and parallel codes, both written in Fortran, was done on the CRYSTAL multicomputer at the University of Wisconsin, and the computational results are presented. Maximum efficiency occurred for multiperiod generalized network problems where a speedup approximately linear in the number of processors was achieved.This research was supported in part by NSF grants DCR-8503148 and CCR-8709952 and by AFOSR grant AFOSR-86-0194.     

Deflated Krylov subspace methods for nearly singular linear systems

This paper concerns the use of Krylov subspace methods for the solution of nearly singular nonsymmetric linear systems. We show that the incomplete orthogonalization methods (IOM) in conjunction with certain deflation techniques of Stewart, Chan, and Saad can be used to solve large nonsymmetric linear systems which are nearly singular.This work was supported by the National Science Foundation, Grants DMS-8403148 and DCR-81-16779, and by the Office of Naval Research, Contract N00014-85-K-0725.     

Minimum principle sufficiency

We characterize the property of obtaining a solution to a convex program by minimizing over the feasible region a linearization of the objective function at any of its solution points (Minimum Principle Sufficiency). For the case of a monotone linear complementarity problem this MPS property is completely equivalent to the existence of a nondegenerate solution to the problem. For the case of a convex quadratic program, the MPS property is equivalent to the span of the Hessian of the objective function being contained in the normal cone to the feasible region at any solution point, plus the cone generated by the gradient of the objective function at any solution point. This in turn is equivalent to the quadratic program having a weak sharp minimum. An important application of the MPS property is that minimizing on the feasible region a linearization of the objective function at a point in a neighborhood of a solution point gives an exact solution of the convex program. This leads to finite termination of convergent algorithms that periodically minimize such a linearization.This material is based on research supported by National Science Foundation Grants DCR-8521228 and CCR-8723091, and Air Force Office of Scientific Research Grants AFOSR 86-0172 and AFOSR and AFOSR 89-0410.     

A parallel balance scheme for banded linear systems

A parallel algorithm is proposed for the solution of narrow banded non‐symmetric linear systems. The linear system is partitioned into blocks of rows with a small number of unknowns common to multiple blocks. Our technique yields a reduced system defined only on these common unknowns which can then be solved by a direct or iterative method. A projection based extension to this approach is also proposed for computing the reduced system implicitly, which gives rise to an inner–outer iteration method. In addition, the product of a vector with the reduced system matrix can be computed efficiently on a multiprocessor by concurrent projections onto subspaces of block rows. Scalable implementations of the algorithm can be devized for hierarchical parallel architectures by exploiting the two‐level parallelism inherent in the method. Our experiments indicate that the proposed algorithm is a robust and competitive alternative to existing methods, particularly for difficult problems with strong indefinite symmetric part. Copyright © 2001 John Wiley & Sons, Ltd.     

Error bounds and strong upper semicontinuity for monotone affine variational inequalities

Global error bounds for possibly degenerate or nondegenerate monotone affine variational inequality problems are given. The error bounds are on an arbitrary point and are in terms of the distance between the given point and a solution to a convex quadratic program. For the monotone linear complementarity problem the convex program is that of minimizing a quadratic function on the nonnegative orthant. These bounds may form the basis of an iterative quadratic programming procedure for solving affine variational inequality problems. A strong upper semicontinuity result is also obtained which may be useful for finitely terminating any convergent algorithm by periodically solving a linear program.This material is based on research supported by Air Force Office of Scientific Research Grant AFOSR-89-0410 and National Science Foundation Grants CCR-9101801 and CCR-9157632.     

Analysis for the end of block wasted space

The problem examined in this report is the calculation of the average wasted space at the end of the block when variable length records are inserted in the file. Previous efforts are based in approximations. Here, a detailed analysis based on Markov chains gives the exact solution. A framework is presented which shows the relations between the previous approaches. The proposed model includes the previous models as special limiting cases. Simulation results close to the analytic results are also presented.This research was sponsored partially by the National Science Foundation under the grants DCR-86-16833, IRI-8719458 and IRI-8958546 and by the Air Force Office of Scientific Research under grant AFOSR-89-0303.     

A unified view of interior point methods for linear programming

The paper shows how various interior point methods for linear programming may all be derived from logarithmic barrier methods. These methods include primal and dual projective methods, affine methods, and methods based on the method of centers. In particular, the paper demonstrates that Karmarkar's algorithm is equivalent to a classical logarithmic barrier method applied to a problem in standard form.Invited paper presented at the Workshop on Supercomputers in Optimization, Minneapolis, Minn., May 1988.The work of this author was supported by the Air Force Office of Scientific Research, Air Force Systems Command, USA, under Grants AFOSR-87-0215 and AFOSR-85-0271. The US Government is authorized to reproduce and distribute reprints for Governmental purposes not withstanding any copyright notation thereon.     

Conjugate residual methods for almost symmetric linear systems

This paper concerns the use of conjugate residual methods for the solution of nonsymmetric linear systems arising in applications to differential equations. We focus on an application derived from a seismic inverse problem. The linear system is a small perturbation to a symmetric positive-definite system, the nonsymmetries arising from discretization errors in the solution of certain boundary-value problems. We state and prove a new error bound for a class of generalized conjugate residual methods; we show that, in some cases, the perturbed symmetric problem can be solved with an error bound similar to the one for the conjugate residual method applied to the symmetric problem. We also discuss several applications for special distributions of eigenvalues.This work was supported in part by the National Science Foundation, Grants DMS-84-03148 and DCR-81-16779, and by the Office of Naval Research, Contract N00014-85-K-0725.