Try 'fr[r]..1 by -1' The by makes it count in reverse order.
(page 16 of manual, Arithmetic set). 2011/11/2 Kasper Tordrup <[email protected]> > Hi Guys > I came across a weird bug or maybe it was just me that was doing something > stupid. > I had this constraint: > s.t. overlap1{r in R, j in fr[r]..1}: sum{s in S, i in > max(1,j-hs[s]+1)..j} x[s,r,i] <= 1; > but GLPK kept violating it. I tried making the sum in the end of my .mod > file (after solve;) and I could see that there was a clear violation. > I could see in my shell that it was generating the constraint, but when I > checked the output file there was no values for overlap1 like it has just > ignored the constraint? > I then played a little with the constraint and changed it to: > s.t. overlap1{r in R, j in 1..fr[r]}: sum{s in S, i in > max(1,j-hs[s]+1)..j} x[s,r,i] <= 1; > only changed "fr[r]..1" to "1..fr[r]" and now it works perfectly. Is there > some logical explanation that I'm missing? > > Best regards, > Kasper > > _______________________________________________ > Help-glpk mailing list > [email protected] > https://lists.gnu.org/mailman/listinfo/help-glpk > > -- *Nilo Cesar Teixeira* [email protected] (55) (11) 8571-5314
_______________________________________________ Help-glpk mailing list [email protected] https://lists.gnu.org/mailman/listinfo/help-glpk
