Originally posted by: Heungjo

CPLEX manual says about pseudo costs as follows:

VarSel Parameter Values for Branching Variable Choice:
2
CPX_VARSEL_PSEUDO
Use pseudo costs, which derives an estimate about the effect of each proposed branch from duality information.

Is there any more detail material about pseudo costs in CPLEX?
How does CPLEX use duality information to derive an estimate about the effect of each proposed branch?

Another thing is that a slide presented by IBM in 2008 at SIAM says that pseudo costs strategy initializes with strong branching.
Is this correct still in CPLEX 12?

Thank you in advance for your help.
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: SystemAdmin

Since version 11, the "pseudo cost" branching rule is essentially identical to the "reliability branching" rule as explained in

{code]
@ARTICLE{Achterberg04branchingrules,
author = {Tobias Achterberg and Thorsten Koch and Alexander Martin},
title = {Branching rules revisited},
journal = {OPERATIONS RESEARCH LETTERS},
year = {2004},
volume = {33},
pages = {42--54}
}
{code}

which you can get, for example, here:

http://miplib.zib.de/paper/achterbergkochmartin2005_pp.pdf
http://opus.kobv.de/zib/volltexte/2004/789/pdf/ZR-04-13.pdf
Duality information is not used in pseudo cost branching. Pseudo costs are initialized with strong branching.
Tobias
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: Heungjo

Thank you so much for your quick and kind answer.
I really appreciate that.
#CPLEXOptimizers
#DecisionOptimization