Originally posted by: kc78

Hi

I have a LP problem that I solved using the CPLEX 12.4 academic version in MATLAB. When i checked the A*x = b, where b is a zero vector. it gave a significant value of ~ 1E-7. I was wondering if this is usual or is there somewhere parameters that i can make it less significant?

The LP file is attached.

Thank you.
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: SystemAdmin

I ran your problem in the interactive optimizer. (Note for future reference: SAV files reproduce the model with greater fidelity than LP files, which introduce a little rounding error by translating coefficients to/from ASCII.) Initially, the interactive optimizer gave me a solution with constraint violations bounded in magnitude by about 3e-14, but with violations in variable bounds reaching about 2.3e-7. I set the feasibility tolerance to 1e-9, ran again, and got constraint errors about the same size but bound violations no worse than around 1e-12.

Paul

Mathematicians are like Frenchmen: whenever you say something to them, they translate it into their own language, and at once it is something entirely different. (Goethe)
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: kc78

Hi Paul

Indeed it was the feasibility tolerance that was set at 1e-6 in my case. Thank you very much for your help!

best rgds
#CPLEXOptimizers
#DecisionOptimization