Model/scale_it.gms

GAMS is sometimes sensitive to scaling. Typically not a problem in LP but critical in NLP. Here we scale the objective function parameters to get it closer to the scale of other variables.