Originally posted by: NSopt

Hello.

I am using CPXsetbranchcallbackfunc to solve binary problem.
Suppose we are at certain node in Branch and Bound tree and we need to solve LP to get lower bound.
Do you have any experience that following two cases make different LP relaxation result?

case 1) change the variable bound
e.g.) x_1 <= 0, x_2 <= 0 using

CPXbranchcallbackbranchbds

case 2) add a constraint
e.g.) x_1 + x_2 = 0 using

CPXbranchcallbackbranchconstraints

I think it's supposed to be same but it is not.
Same issue happens even I was changing the algorithms option(i.e., CPX_PARAM_SUBALG) for both cases.
Do you have any idea?
Thank you.
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: SystemAdmin

Can you please elaborate on what difference between the two strategies you observe and what exactly is unexpected about that difference?
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: NSopt

I'm sorry it was my mistake. I think both of strategy make same LP optimal value.
Thank you for your time.
#CPLEXOptimizers
#DecisionOptimization