Closed svigerske closed 5 years ago
Attachment install.out by Kemmler created at 2008-01-13 15:05:27
Binary attachment Kemmler.zip by Kemmler created at 2008-01-13 15:08:26 lost by Trac to GitHub conversion.
Comment by @andrea5w created at 2008-01-15 02:59:27
Hi Roman,
It seems that you are trying to use Pardiso from the Intel MKL. The version of Pardiso in the MKL does not (yet?) include the features necessary for the use of Pardiso in Ipopt (it doesn't compute the inertia of the matrix), and as a consequence it might actually fail the way you describe. If you want to use Pardiso, you need to download it directly from the Pardiso home page.
There was recently an exchange on the Ipopt mailing list regarding this issue:
http://list.coin-or.org/pipermail/coin-ipopt/2008-January/000991.html
I hope this helps.
Andreas
Comment by @andrea5w created at 2008-01-15 02:59:27
Changing assignee from ipopt-team to @andrea5w.
Comment by @andrea5w created at 2008-01-15 02:59:27
Changing status from new to assigned.
Comment by @andrea5w created at 2008-03-01 21:45:11
No information added to ticket in a while, closing it.
Comment by @andrea5w created at 2008-03-01 21:45:11
Resolution: worksforme
Issue created by migration from Trac.
Original creator: Kemmler
Original creation time: 2008-01-13 15:04:42
Assignee: ipopt-team
Version: 3.3
Hi, I have installed Ipopt-3.3.4 and compiled using specific flags. A summary of the “./configure” output is added in the appendix (configure.out). For C I use the gnu compilers and for FORTRAN the Intel one. The output of “make install” is also given in the appendix (install.out). The solver Pardiso, blas and lapack is provided by Intel’s MKL. All work fine (compilation output: hs071_f.out) until I try to run the test examples. In pardisoc occurs a SIGSEGV-problem.
In an own written program Pardiso work fine. Can you detect the problem in combination with your IPOPT?
Thanks
Roman