[Ipopt] MKL pardiso clarification

Panos Lambrianides panos at soe.ucsc.edu
Tue Mar 1 10:59:24 EST 2016


Thank you for the responses.  I can verify this works very well indeed.
The only thing perhaps to keep in mind for people trying to get this to
work in the future is that there is a new linking model for the MKL library
that is not documented in the Ipopt manuals.

The information is here for those who may be interested:
https://software.intel.com/en-us/articles/a-new-linking-model-single-dynamic-library-mkl_rt-since-intel-mkl-103

Thanks for all the help.

Best,
--Panos

On Sat, Feb 27, 2016 at 4:31 AM, Stefan Vigerske <stefan at math.hu-berlin.de>
wrote:

> Hi,
>
> On 02/26/2016 09:23 PM, Panos Lambrianides wrote:
>
>> Hi everyone,
>>
>> Please excuse me if this has been asked before.  I have found conflicting
>> documentation on this issue.
>> My question is the following
>> 1.  Does MKL Pardiso work as a solver with Ipopt?
>>
>
> It seems so.
>
> 2.  According to the instructions below it is just a matter of linking to
>> the mkl libraries-is this correct?
>>
>
> Yes.
>
> If you want to compile IPOPT with the Pardiso library that is included in
>> Intel MKL, it should be sufficient to ensure that MKL is used for the
>> linear algebra routines (Blas/Lapack). On some systems, configure is able
>> to find MKL automatically when looking for Blas. On other systems, one has
>> to specify the MKL libraries with the --with-blas option.
>>
>
> 3.  If 2 is true, how does one invoke the mkl-pardiso solver?  Specifically
>> what is the option for the linear solver?
>> app->Options()->SetStringValue("linear_solver", "ma57");
>>
>
> app->Options()->SetStringValue("linear_solver", "pardiso");
>
>
> The libraries from the Pardiso-Project and the MKL Pardiso share the same
> interface code. One cannot use both with the same Ipopt library.
>
> Stefan
>
>
>> Many thanks,
>>
>>
>>
>> _______________________________________________
>> Ipopt mailing list
>> Ipopt at list.coin-or.org
>> http://list.coin-or.org/mailman/listinfo/ipopt
>>
>>
>
> --
> http://www.gams.com/~stefan
>



-- 
Panos Lambrianides
panos at soe.ucsc.edu
(415) 713-6718 (Mobile)
https://sites.google.com/a/ucsc.edu/uncertainty-quantification/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://list.coin-or.org/pipermail/ipopt/attachments/20160301/44bcbf12/attachment.html>


More information about the Ipopt mailing list