[Ipopt] Using ipopt with PARDISO

[Renata Leitao Dias da Silva]

Hi Olaf,
 
> If possible use the optimized LAPACK and BLAS routines (e.g. AMD ACML, Intel's
> MKL, or Goto's BLAS) in connection with MUMPS or PARDISO, but do not use the
> BLAS and LAPACK routines that you can get with the ./get.Lapack script.
> 
> In general, you will see an IPOPT acceleration of a factor 2 to 10 by
> switching to optimized BLAS.
 
Ok, I will do what you suggest. I think Intel's MKL is not good for ipopt 
(as I read somewhere) but I will get one of the others soon.

Anyway, for now I just wanted to compare ipopt with MA27 vs. ipopt 
with PARDISO in terms of robustness. I suppose the optimized BLAS/LAPACK 
don't change the results, just the CPU time, am I right?

Thanks again,
Renata

--
Renata Silva
Department of Mathematics
University of Coimbra