On Apr 4, 2011 7:36pm, Marcelo Pinto <address@hidden> wrote: > I applied the patch and could see that it really has much more > features. However, when I ran a script that I know that the best > solution is 0.1% away from the optimal solution, the optimization > doesn't stop, even configuring 'param.mipgap = 0.1/100'.
One additional remark. Don't panic when you'll see that the patch you posted cannot produce solutions that make sense when the gap is reached. This is due to the fact that you didn't update the errnum check to include a reached gap. In other words, in your patch if the gap is reached, xmin is never retrieved and therefore it is always [NA,...,NA]. Fix that and go through 2) and 3) again. Of course make sure you don't have random matrices/vectors in your script, otherwise you won't get much insight by using CPLEX or Gurobi.