User Tools

Site Tools


solver:a_newer_version_of_the_solver_returns_a_different_solution

A newer version of the Solver returns a different Solution

Q: I noticed that different versions of GAMS give different results for the same input file. In my case I am running a quadratic model with MINOS

Contributed by Arne Drud: Do you know if your model can have more than one optimal solution (is it STRICTLY convex or has other nice properties)? Are the differences large or small (round-off errors)? Different versions of an NLP algorithms can give different results if the model has more than one solution. And if the optimum is very flat, then you may get very small differences in objective but fairly large differences in primal values. A very small change caused by new ordering methods or small round-off errors early in the computation can change the whole iteration path and lead to different solutions if there are multiple optima. Example: variable X1 and X2 have mathematically the same reduced cost in the initial point. In one version of the algorithm X1 will be selected to enter the basis because it is first. In another version the variables have been sorted according to some criteria and X2 appears to be first and it will be selected. From there on, everything is different.

solver/a_newer_version_of_the_solver_returns_a_different_solution.txt · Last modified: 2007/10/21 04:44 by franz