Hi All,

I'm having some issues solving a LP problem.

The orginal problem is infeasible and I'm trying to relax the problem using repairInfeas(iscode, 'c', 'o', ' ', 20, 20, 0, 0, 0.001);

The manual reads

If a feasible solution is identified for the relaxed problem, with a sum of violations p, then the sum of violations is restricted to be no greater than (1+delta)p, and the problem is optimized with respect to the original objective function.

It seems that a feasible solution is identified in my case. But after solving for a long time, Xpress reports the relaxed problem is also infeasible, which doesn't make sense to me. BTW, I'm using Newton-Barrier. Shall I switch to Primal Simplex in this case? Any suggestion is appreciated.

Hi,

You are right in your expectations, as all constraints have been allowed to be relaxed in your call, so the problem should have been found feasible. This sounds like a numerical failure, especially if barrier is involved. Trying the simplex methods would indeed be a good attempt, as well as trying to increase the delta parameter. Are the numbers involved large?