Re: [Help-glpk] suboptimal solution from glp_intopt

[Andrew Makhorin]

> thank you very much for having a look at my problem. 
> Attached you can find two problem files and two result files. 

Thank you.

> 
> glpk_mip_problem.glpk is my original problem and glpk_mip_result.txt 
> is the solution found by glpk of which I think it is suboptimal. 
> 
> the problem file with _cplex in its name has additional constraints that fix 
> all variables to the cplex solution. The result you can find in file number 
> 4. 
> 
> I am using glpk version 4.52. 
> 

Some double-bounded variables in your instances have very large upper 
bound (1e+20) that may cause numerical difficulties. I guess that your
intention was simply to indicate that such variables have no upper
bound. The correct way to do this is to specify the column type GLP_LO
(rather than GLP_DB) in corresponding calls to glp_set_col_bnds. For
more details please see the glpk reference manual included in the
distribution.