© J.C. Baltzer AG, Science Publishers A simulated annealing code for general integer linear programs ★ David Abramson a and Marcus Randall b a School of Computer Science and Software Engineering, Department of Digital Systems, Monash University, Clayton, VIC 3168, Australia E-mail: davida@dgs.monash.edu.au b School of Environmental and Applied Science, Griffith University, QLD 4217, Australia E-mail: m.randall@eas.gu.edu.au This paper explores the use of simulated annealing (SA) for solving arbitrary com- binatorial optimisation problems. It reviews an existing code called GPSIMAN for solving 0–1 problems, and evaluates it against a commercial branch-and-bound code, OSL. The problems tested include travelling salesman, graph colouring, bin packing, quadratic assign- ment and generalised assignment. The paper then describes a technique for representing these problems using arbitrary integer variables, and shows how a general simulated anneal- ing algorithm can also be applied. This new code, INTSA, outperforms GPSIMAN and OSL on almost all of the problems tested. Keywords: simulated annealing, combinatorial optimisation, integer linear programming AMS subject classification: 90C05, 90C10, 90C27 1. Introduction For many years, the holy grail of operations research has been an efficient integer solver which can be applied to a wide range of problems with little modification. In this model, a user specifies a problem using some mathematical notation which is processed directly by the solver. One of the most general specification techniques uses a linear cost function and constraints, together with variables which are restricted to the values 0 and 1. There are a number of algorithms which attempt to find solutions for these 0–1 problems; however, their performance is highly variable. ★ The authors would like to acknowledge the assistance of Mohan Krishnamoorthy in this work. Marcus Randall is funded by a Griffith University Post Graduate Scholarship. The project is funded by the Australian Research Council. Annals of Operations Research 86(1999)3–21 3