A Massively Parallel Algorithm for Nonlinear Stochastic Network Problems

We develop an algorithm for solving nonlinear, two-stage stochastic problems with network recourse. The algorithm is based on the framework of row-action methods. The problem is formulated by replicating the first-stage variables and then adding nonanticipativity side constraints. A series of independent deterministic network problems are solved at each step of the algorithm, followed by an iterative step over the nonanticipativity constraints. The solution point of the iterates over the nonanticipativity constraints is obtained analytically. The row-action nature of the algorithm makes it suitable for parallel implementations. A data representation of the problem is developed that permits the massively parallel solution of all the scenario subproblems concurrently. The algorithm is implemented on a Connection Machine CM-2 with up to 32K processing elements and achieves computing rates of 276 MFLOPS. Very large problems-8,192 scenarios with a deterministic equivalent nonlinear program with 868,367 constraints and 2,474,017 variables-are solved within a few minutes. We report extensive numerical results regarding the effects of stochasticity on the efficiency of the algorithm.

[1]  Stavros A. Zenios,et al.  Proximal minimizations with D-functions and the massively parallel solution of linear network programs , 1993, Comput. Optim. Appl..

[2]  Guy E. Blelloch,et al.  Vector Models for Data-Parallel Computing , 1990 .

[3]  Roger J.-B. Wets.  On Parallel Processors Design for Solving Stochastics Programs , 1985 .

[4]  Andrzej Ruszczynski,et al.  A regularized decomposition method for minimizing a sum of polyhedral functions , 1986, Math. Program..

[5]  R. J-B. Wets,et al.  Large Scale Linear Programming Techniques , 1988 .

[6]  R. Wets.  Stochastic Programs with Fixed Recourse: The Equivalent Deterministic Program , 1974 .

[7]  George B. Dantzig,et al.  Planning under uncertainty using parallel computing , 1988 .

[8]  John R. Birge,et al.  The value of the stochastic solution in stochastic linear programs with fixed recourse , 1982, Math. Program..

[9]  R. Wets,et al.  Programming under Uncertainty and Stochastic Optimal Control , 1966 .

[10]  E. Beale.  ON MINIMIZING A CONVEX FUNCTION SUBJECT TO LINEAR INEQUALITIES , 1955 .

[11]  Stavros A. Zenios,et al.  Massively Parallel Algorithms for Singly Constrained Convex Programs , 1992, INFORMS J. Comput..

[12]  Y. Censor.  Row-Action Methods for Huge and Sparse Systems and Their Applications , 1981 .

[13]  R. Wets,et al.  L-SHAPED LINEAR PROGRAMS WITH APPLICATIONS TO OPTIMAL CONTROL AND STOCHASTIC PROGRAMMING. , 1969 .

[14]  Stavros A. Zenios,et al.  Nonlinear network optimization on a massively parallel connection machine , 1988 .

[15]  A. Ruszczynski.  Regularized Decomposition and Augmented Lagrangian Decomposition for Angular Linear Programming Problems , 1989 .

[16]  John M. Mulvey,et al.  A distributed algorithm for convex network optimization problems , 1988, Parallel Comput..

[17]  Y. Censor,et al.  An iterative row-action method for interval convex programming , 1981 .

[18]  L. Qi.  Forest iteration method for stochastic transportation problem , 1985 .

[19]  Stavros A. Zenios,et al.  On the Fine-Grain Decomposition of Multicommodity Transportation Problems , 1991, SIAM J. Optim..

[20]  Y. Censor,et al.  Proximal minimization algorithm withD-functions , 1992 .

[21]  G. Dantzig,et al.  Large-scale linear programming : proceedings of a IIASA workshop, 2-6 June 1980 , 1981 .

[22]  R. Tyrrell Rockafellar,et al.  Augmented Lagrangians and Applications of the Proximal Point Algorithm in Convex Programming , 1976, Math. Oper. Res..

[23]  W. Daniel Hillis,et al.  The connection machine , 1985 .

[24]  Yair Censor,et al.  Massively Parallel Row-Action Algorithms for Some Nonlinear Transportation Problems , 1991, SIAM J. Optim..

[25]  Stein W. Wallace.  A Two-Stage Stochastic Facility-Location Problem with Time-Dependent Supply , 1988 .

[26]  Stein W. Wallace,et al.  Solving stochastic programs with network recourse , 1986, Networks.

[27]  Andrzej Ruszczynski,et al.  Parallel decomposition of multistage stochastic programming problems , 1993, Math. Program..

[28]  John M. Mulvey,et al.  Solving multistage stochastic networks: An application of scenario aggregation , 1991, Networks.

[29]  R. Rockafellar.  Monotone Operators and the Proximal Point Algorithm , 1976 .

[30]  Roger J.-B. Wets,et al.  The aggregation principle in scenario analysis stochastic optimization , 1989 .

[31]  George B. Dantzig,et al.  Parallel processors for planning under uncertainty , 1990 .

[32]  R. Tyrrell Rockafellar,et al.  Scenarios and Policy Aggregation in Optimization Under Uncertainty , 1991, Math. Oper. Res..

[33]  Gerd Infanger,et al.  Monte Carlo (importance) sampling within a benders decomposition algorithm for stochastic linear programs , 1991, Ann. Oper. Res..

[34]  John N. Tsitsiklis,et al.  Parallel and distributed computation , 1989 .

[35]  O. Mangasarian,et al.  NONLINEAR PERTURBATION OF LINEAR PROGRAMS , 1979 .

[36]  K. A. Ariyawansa,et al.  Performance of a benchmark parallel implementation of the Van Slyke and Wets algorithm for two-stage stochastic programs on the Sequent/Balance , 1991, Concurr. Pract. Exp..

[37]  Y. Censor,et al.  On the proximal minimization algorithm with D-Functions , 1992 .