[SCIP] Dual value about SCIP

[Benjamin Müller]

Dear Kai,

can you check if the solution value of the restricted master problem 
(RMP) is equal for SCIP and CPLEX or is there already something strange? 
In general the dual multipliers you obtain after solving the RMP are not 
unique so it is okay if SCIP reports other values than CPLEX does. If 
you solve the pricing problem with the dual multipliers obtained by SCIP 
then you should find at least some improving column since there is 
obviously one left. When this is not the case then I think there is a 
problem with or how you solve your pricing problems.

Best,

Benny

On 11/09/2015 10:04 AM, weikaimas at 163.com wrote:
> Hi,
>
> I have a branch and price algorithm.In some round, the dual value 
> returned be SCIP is  different from another LP solver (CPLEX).
> (At the begining of a round, I printed the model and solve it by CPLEX 
> to obtain the dual value.)
>
> Using the dual value provided by CPLEX, I can find some eatra 
> variables, which reduced the lower bound.  These varibales cannot
> find by dual variables provided by SCIP. So, in my code, I cannot get 
> the real low bound by SCIP.
> My computer is 64-bit and the SCIP I used is 32-bit. Is this the 
> reason about my problem ?
>
> best regard,
>
> Kai Wei.
>
>
>
>
> _______________________________________________
> Scip mailing list
> Scip at zib.de
> http://listserv.zib.de/mailman/listinfo/scip

-- 
______________________________
Benjamin Müller
Zuse Institute Berlin
Takustr. 7, 14195 Berlin
benjamin.mueller at zib.de
+49 30 841 85-195

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.zib.de/pipermail/scip/attachments/20151118/5e858c92/attachment.html>