[Clp] OsiClp resolve primal or dual simplex?

[John J Forrest]

Matt,

I have exposed lastAlgorithm.

John


                                                                                                                             
  From:       Matthew Galati <magh at lehigh.edu>                                                                               
                                                                                                                             
  To:         Matthew Galati <magh at lehigh.edu>                                                                               
                                                                                                                             
  Cc:         tkr2 at lehigh.edu, clp at list.coin-or.org, rob.pratt at sas.com                                                       
                                                                                                                             
  Date:       05/30/2008 05:37 PM                                                                                            
                                                                                                                             
  Subject:    Re: [Clp] OsiClp resolve primal or dual simplex?                                                               
                                                                                                                             





Matthew Galati wrote:
>>> Does OsiClp's "resolve()" use primal or dual simplex? I was always
under
>>> the impression that it used dual simplex, but I guess that only makes
>>> more sense if you are adding rows (like, cut generation). What if you
>>> are adding columns? Does it use primal simplex? Or is there some
>>> heuristic that determines on the fly?
>>>
>>>
>>>
>> You can configure this using OsiHintsParams (I think the righ option is
>> OsiDoDualInResoilve).
>>
>> Note that clp may chose the wrong option if you don't give the hint.
>>
>>
> Thanks Paulo.
>
> Is there some status flag to tell what CLP decided to use primal or dual
> simplex that I can check afterwards? The "resolve" function has lots of
> conditions so it is hard to tell without tracing through line by line.
>
It looks like there is a data member called lastAlgorithm_ that might
contain this information. However, the data is protected. John - can you
provide an access method to this information?

Thanks,
Matt

_______________________________________________
Clp mailing list
Clp at list.coin-or.org
http://list.coin-or.org/mailman/listinfo/clp