@Lawrence: I will go through your tests. At the moment I try to use my own petsc version (and not the pip installed one) for petsc4py. This crashed because I did not have "--download-ctetgen --download-triangle --download-chaco". I'm doing a petsc ./configure atm. I will see how that goes. I will then kill this version install petsc with pip again and do your tests. Will take a while due to all the PETSc configuring. Thank you! Henrik
-----Ursprüngliche Nachricht----- Von: firedrake-bounces@imperial.ac.uk [mailto:firedrake- bounces@imperial.ac.uk] Im Auftrag von Lawrence Mitchell Gesendet: 17 September 2015 10:40 An: firedrake@imperial.ac.uk Betreff: Re: [firedrake] undefined symbol: kspfgmresmodifypcksp_
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Dear Henrik,
On 17/09/15 09:07, Buesing, Henrik wrote:
I tried to rebuild everything from scratch, but the error persists. It already occurs in the tests for pyop2 (see attachement).
Just for sanity checking, I presume you used the provided "firedrake- install" script to install firedrake?
I think the problem has already started when installing PETSc (and/or petsc4py). Let's try and figure out what's gone wrong. Can you first try building and running the attached PETSc C program. You'll need to do:
PETSC_DIR=/home/user/.local/lib/python2.7/site-packages/petsc/ make testksp
Having saved the provided testksp.c and Makefile.
Then, if that builds, try running. I get something like:
$ ./testksp KSP Object: 1 MPI processes type: fgmres GMRES: restart=30, using Classical (unmodified) Gram-Schmidt Orthogonalization with no iterative refinement GMRES: happy breakdown tolerance 1e-30 maximum iterations=10000, initial guess is zero tolerances: relative=1e-05, absolute=1e-50, divergence=10000 right preconditioning using UNPRECONDITIONED norm type for convergence test PC Object: 1 MPI processes type: ilu ILU: out-of-place factorization 0 levels of fill tolerance for zero pivot 2.22045e-14 matrix ordering: natural factor fill ratio given 1, needed 1 Factored matrix follows: Mat Object: 1 MPI processes type: seqaij rows=1, cols=1 package used to perform factorization: petsc total: nonzeros=1, allocated nonzeros=1 total number of mallocs used during MatSetValues calls =0 not using I-node routines linear system matrix = precond matrix: Mat Object: 1 MPI processes type: seqaij rows=1, cols=1 total: nonzeros=1, allocated nonzeros=5 total number of mallocs used during MatSetValues calls =0 not using I-node routines
If that doesn't work, we know the problem has already appeared and need to figure out what went wrong when installing PETSc.
If it does work let's then see if petsc4py works:
$ python -c "from petsc4py import PETSc; print PETSc.__file__"
I suspect you'll already run into trouble at one of those two points.
Cheers,
Lawrence
-----BEGIN PGP SIGNATURE----- Version: GnuPG v1
iQEcBAEBAgAGBQJV+nw/AAoJECOc1kQ8PEYvqRUH/1YLbWyPnwwDx5kOGQwKCQUz x2OzQ+uXEScIR8IUeZ2MGwvQ4/uHpLFxXeT+ZPqZR1wYON1F/oGNkPx9rJm3dDsn 5mmESToh+dznE8S/bmRDlI/4j3QURS9+52rStl8GMG2S/Vb7figFiBA6/EuL5sBg kf4Crbj6+JpQW8UNo9Z3xHVD3SAMyhssdtYHXmR5baRgY/b5vkxdbttT65BkGt5c Ll7y2Gkauu9EpfWFob6mjZFJFcbeKGmRGgrwJ5Wdt6PLVs3VF1zPzOpoxYWWj+k1 xh8YG7yu8qcAW5z1fRbe+dd5bKFkC//Zycn2+kcqWtg9ysn57JjSN9f+2/M56u4= =xjd7 -----END PGP SIGNATURE-----