An investigation of the relationship between problem characteristics and algorithm performance: a case study of the GAP

Abstract We compare synthetic Generalized Assignment Problems (GAP) generated under two correlation-induction strategics: Implicit Correlation Induction (ICI) and Explicit Correlation Induction (ECI). We present computational results for two commercially-available solvers and four heuristics on 590 test problems. We conclude that the solvers’ performances degrade as the population correlation between the objective function and capacity constraint coefficients decreases. However, the heuristics’ performances improved as the absolute value of the same population correlation increases. We find that problems generated under ECI arc more challenging than problems generated under ICI and may lead to a better understanding of the capabilities and limitations of the solution methods being evaluated. We recommend that one consider the purpose(s) of an experiment and types of solution procedures to be evaluated when determining what types of test problems to generate.

[1]  E. Balas,et al.  Pivot and Complement–A Heuristic for 0-1 Programming , 1980 .

[2]  John N. Hooker,et al.  Needed: An Empirical Science of Algorithms , 1994, Oper. Res..

[3]  Charles H. Reilly.  Optimization test problems with uniformly distributed coefficients , 1991, 1991 Winter Simulation Conference Proceedings..

[4]  Charles H. Reilly,et al.  The Effects of Coefficient Correlation Structure in Two-Dimensional Knapsack Problems on Solution Procedure Performance , 2000 .

[5]  Thuruthickara C. John.  Tradeoff solutions in single machine production scheduling for minimizing flow time and maximum penalty , 1989, Comput. Oper. Res..

[6]  Charles H. Reilly,et al.  Multivariate composite distributions for coefficients in synthetic optimization problems , 2000, Eur. J. Oper. Res..

[7]  George L. Nemhauser,et al.  Experiments with parallel branch-and-bound algorithms for the set covering problem , 1993, Oper. Res. Lett..

[8]  Charles H. Reilly.  A comparison of alternative input models for synthetic optimization problems , 1993, WSC '93.

[9]  M. Trick.  A Linear Relaxation Heuristic for the Generalized Assignment Problem , 1992 .

[10]  Joseph B. Mazzola,et al.  An algorithm for the bottleneck generalized assignment problem , 1993, Comput. Oper. Res..

[11]  Paolo Toth,et al.  The bottleneck generalized assignment problem , 1995 .

[12]  L. V. Wassenhove,et al.  Algorithms for scheduling a single machine to minimize the weighted number of late jobs , 1988 .

[13]  John N. Hooker,et al.  Testing heuristics: We have it all wrong , 1995, J. Heuristics.

[14]  David Connolly.  Knapsack Problems: Algorithms and Computer Implementations , 1991 .

[15]  Mohammad M. Amini,et al.  A rigorous computational comparison of alternative solution methods for the generalized assignment problem , 1994 .

[16]  Monique Guignard-Spielberg,et al.  Technical Note - An Improved Dual Based Algorithm for the Generalized Assignment Problem , 1989, Oper. Res..

[17]  Charles H. Reilly,et al.  Composition for multivariate random variables , 1994, Proceedings of Winter Simulation Conference.

[18]  Egon Balas,et al.  An Algorithm for Large Zero-One Knapsack Problems , 1980, Oper. Res..

[19]  Benjamin W. Lin,et al.  Controlled Experimental Design for Statistical Comparison of Integer Programming Algorithms , 1979 .

[20]  C. H. Reilly.  Generating coefficients for optimization test problems with implicit correlation induction , 1997, 1997 IEEE International Conference on Systems, Man, and Cybernetics. Computational Cybernetics and Simulation.

[21]  L. V. Wassenhove,et al.  A survey of algorithms for the generalized assignment problem , 1992 .

[22]  Mauricio G. C. Resende,et al.  Designing and reporting on computational experiments with heuristic methods , 1995, J. Heuristics.