Re: [Help-glpk] Re: Feasibility Pump

[Andrew Makhorin]

> Successful case:
> [...]
> 67.740683s      Solution found by heuristic: 11620                          

> Unsuccessful case:
> [...]
> 419.944408s     +135375: mip =     not found yet <=              +inf        
> (1; 0)
> 430.816306s     |137000: obj =   6.281588968e+05  infeas =  1.279e-09 (0)
> 434.008313s     |137500: obj =   6.281588968e+05  infeas =  4.095e-10 (0)
> 437.104311s     |138000: obj =   6.281588968e+05  infeas =  5.192e-10 (0)
> 439.824312s     |138500: obj =   6.281588968e+05  infeas =  4.817e-10 (0)
> 442.912308s     |139000: obj =   6.281588968e+05  infeas =  4.381e-10 (0)
> 445.824312s     |139500: obj =   6.281588968e+05  infeas =  5.841e-10 (0)
> 448.744309s     |140000: obj =   6.281588968e+05  infeas =  3.702e-10 (0)
> 451.93631s      |140500: obj =   6.281588968e+05  infeas =  4.055e-10 (0)
> 455.31632s      |141000: obj =   6.281588968e+05  infeas =  3.728e-10 (0)
> 458.424312s     |141500: obj =   6.281588968e+05  infeas =  3.749e-10 (0)
> 461.800309s     |142000: obj =   6.281588968e+05  infeas =  3.718e-10 (0)
> 465.060322s     |142500: obj =   6.281588968e+05  infeas =  4.039e-10 (0)
> 468.652321s     |143000: obj =   6.281588968e+05  infeas =  3.858e-10 (0)

> After this point, it goes an hour without making progress.  In
> particular, it never outputs "Applying FPUMP heuristic" like the
> successful case.

> My best guess is that there aren #39;t many integer solutions at all...

Thank you for information.

The output indicates that the fpump heuristic has not been applied
yet, because the dual simplex is stalled on solving lp relaxation of
the root problem. Sorry, it is one of glpk defects.

To avoid stalling you can try using options '--noscale' and/or
'--norelax' and/or '--bib' and/or '--std'.