unscaled primal/dual violation

Solver related questions
yanzhiping
User
User
Posts: 85
Joined: 9 months ago

unscaled primal/dual violation

Postby yanzhiping » 8 months ago

I tried Gurobi&Cplex in GAMS for a MILP problem.
I got infeasible message from Cplex that says
model has been proven infeasible.
But in Gurobi, it says
optimal objective XXXX
warning: unscaled primal violation = xxxxe-06 and residual = yyye-06
warning: unscaled dual violation=xxxxe-05 and residual yyye-05
So I am confused whether the optimal value found by Gurobi is the actual optimal objective or not. when either Cplex or Gurobi throws warning like this, can we accept the optimal value still?

Thanks,
Ping

User avatar
bussieck
Moderator
Moderator
Posts: 54
Joined: 1 year ago

Re: unscaled primal/dual violation

Postby bussieck » 8 months ago

Ping,

All solvers work with tolerances and suffer from rounding errors. There are cases where your model is on the brink of feasibility. Hence with some solver settings (or different solvers) you get an "optimal" solution with others the model is reported as infeasible. The Cplex guys have done a nice video about analyzing infeasible models. Perhaps this helps: https://www.youtube.com/watch?v=tUALbhJ2xWY

-Michael


Who is online

Users browsing this forum: No registered users and 1 guest