This example demonstrates several approaches to get multiple solutions when solving a MIP problem. These approaches are:
- Using the incumbent (solver) callback
- Using the solution pool (only supported by CPLEX)
- Repeatedly solve the problem forbidding the previous solutions
In the dice problem a set of three dice has to be designed by assigning an integer number to each face such that on average dice 1 beats dice 2, dice 2 beats dice 3 and dice 3 beats dice 1. The goal is to maximize the number of total wins on average. The dice problem has many solutions.
Multiple solutions, Incumbent callback, Solution pool, GMP.
Bosch, R.A., Mindsharpener - Monochromatic Squares, OPTIMA Newsletter 71 (2004), Mathematical Optimization Society, pp. 6-7.
A zip file with this example can be downloaded here.