Originally posted by: rocarvaj

Dear all,
I'm using the CPXstrongbranch() function in CPLEX 12.5 and I see in the documentation that for a particular variable...

"If either of the fixings results in a problem which is dual unbounded (primal infeasible), the corresponding objective value is set to a large positive value for a minimization problem and a large negative value for a maximization problem. This value is system dependent, but it is usually of magnitude 1.0e+75."

When I check the results I obtain for my particular minimization problem (using an iteration limit of 10), I get a down objective value and a up objective value of -1e+20 (yes, negative for both) for certain variables. So I have the following question:

I've checked using the interactive optimizer and both LP's (down and up fixings) are feasible with objectives of around 1000. How do I interpret the output numbers then?

Thanks in advance for any help,

Rodolfo

Originally posted by: RWunderling

Could it be that the variables in questions are nonbasic and/or already fixed to some value?

Roland

Originally posted by: rocarvaj

I don't think so. Actually the way I solved this was by coding my own strong branching (of course using CPLEX for solving LPs). I used CPXdualopt, CPXgetbasednorms and CPXcopybasednorms. I get numbers that are consistent with results returned by CPXstrongbranch (when it does no return -1e+20).

Thanks!