mathematical optimization - Constraint violation in GLPK (Pulp wrapper) -


i'm using pulp glpk solver , have follow constraint:

-1000000*f1 + v1 + 1000000*f2 - v2 >= -10.999 

solver:

problem.solve(glpk(options=['--mipgap', '0.000001', '--cuts'])) 

but in result solved problem (status 'optimal') constraint violation:

f1 = 0 v1 = 11 f2 = 1 v2 = 1000000 

so, can in case? can increase accuracy somehow or use solution?

i think, run in problems numerical accuracy. coefficients in order of 1.e7 (and want solution 1.e-6). solution needs @ least correct in 1.e11 hold constraint , argument require maximum 1.e-14 relative error.

simply pure numercal solver. (expect maximum 1.e-6 1.e-7 run reliable lp-solver)there exact solvers qsopt , think, soplex got features now.

here link qsopt http://www.math.uwaterloo.ca/~bico/qsopt/downloads/downloads.htm

here link soplex stuff. http://soplex.zib.de/doc/html/exact.php

qsopt available binary (i think) , soplex free research (as understand).

further u can use scip soplex. have or had exact ip project. so, if license ok you, go scip in case.

maybe, u give bit more insight application, 1 can reformulate ur restrictions.


Comments

Popular posts from this blog

PySide and Qt Properties: Connecting signals from Python to QML -

c# - DevExpress.Wpf.Grid.InfiniteGridSizeException was unhandled -

scala - 'wrong top statement declaration' when using slick in IntelliJ -