Re: [Help-glpk] I got a solution that violates the constraints

[xiaomi]

The value range is the problem I don't know. So far, the benchmark gives 
the result that value ranges from 0 to 2000. However, the value may 
increase in larger benchmark.

By the way, is it the only way to formulate the total number of non-zero 
value among all Xmax***_*** by following constrants:

0<= Xmax***_*** <= M B,
where M is a large number( I used 999,999 in my case), and B is a binary 
variable.

By minimizing sum of B, I can get the minimum total number of non-zero 
value. Is there another way to do so?


> Hello
>
> GLPK does not use exact math for solving MIP problems. Hence rounding 
> errors may occur.
>
> What is the value range for Xmax67_134, and Xmax67_136? Can you 
> replace 999,999 by a smaller value?
>
> Best regards
>
> Xypron
>
>
> xiaomi wrote:
> > version:
> > GLPSOL: GLPK LP/MIP Solver, v4.43
> >
> > In my lp file, there are 2 constraint voilated by solving in GLPK:
> >
> > Xmax67_134 - 999999 B67_134BB<= 0
> > Xmax67_136 - 999999 B67_136BB<= 0
> >
> > B67_134BB and B67_136BB are both binary variables. The solution given by
> > GLPK is following:
> >
> > Row name Activity Lower bound Upper bound
> > Xmax67_134 6.8975 0
> > B67_134BB * 0 0 1
> > Xmax67_136 0.7975 0
> > B67_136BB * 0 0 1
> >
> > How could GLKP give a solution that is not feasible?
> >
> > Thanks.
> >
> > _______________________________________________
> > Help-glpk mailing list
> > address@hidden
> > http://lists.gnu.org/mailman/listinfo/help-glpk
> >
> >