Hi John,
I modified the initialization of y to UniquePtr< NumericVector< Number > > x = es.get_system(system_name).solution->zero_clone(); std::vector< dof_id_type > some_dofs( some_dof_set.begin(), some_dof_set.end() ); PetscVector< Number > y( mesh.comm(), x->size(), some_dofs.size(), PARALLEL ); x->create_subvector( y, some_dofs ); In this case, I get no segfault and the ouput is PROC = 1: local size of some_dofs = 62 PROC = 1: global size of y = 121 PROC = 1: local size of y = 62 PROC = 1: (first,last) local index of y = (10, 72) PROC = 0: local size of some_dofs = 10 PROC = 0: global size of y = 121 PROC = 0: local size of y = 10 PROC = 0: (first,last) local index of y = (0, 10) which looks okay. Jonas ________________________________ Von: John Peterson <jwpeter...@gmail.com> Gesendet: Donnerstag, 14. Juli 2016 17:25 An: Jonas Ballani; libmesh-users Betreff: Re: [Libmesh-users] Bug in PetscVector::create_subvector On Thu, Jul 14, 2016 at 9:08 AM, Jonas Ballani <jonas.ball...@epfl.ch<mailto:jonas.ball...@epfl.ch>> wrote: Hi John, Thanks for your quick reply. If you don't initialize y, say by just calling PetscVector< Number > y( mesh.comm() ); you get a memory leak. [1]PETSC ERROR: --------------------- Error Message -------------------------------------------------------------- [1]PETSC ERROR: Petsc has generated inconsistent data [1]PETSC ERROR: ith 28 block entry 36 not owned by any process, upper bound 36 [1]PETSC ERROR: See http://www.mcs.anl.gov/petsc/documentation/faq.html for trouble shooting. [1]PETSC ERROR: Petsc Release Version 3.7.1, May, 15, 2016 [1]PETSC ERROR: ./example-opt on a arch-linux2-c-opt named iacspc13 by jballani Thu Jul 14 17:06:33 2016 [1]PETSC ERROR: Configure options --with-shared-libraries=1 --with-debugging=0 --download-suitesparse --download-blacs --download-ptscotch=yes --download-scalapack --download-mumps --download-metis --prefix=/home/jballani/software/libmesh_install/opt_real/petsc --download-hypre --download-ml [1]PETSC ERROR: #1 VecScatterCreate_PtoS() line 2348 in /home/jballani/software/petsc-3.7.1/src/vec/vec/utils/vpscat.c [1]PETSC ERROR: #2 VecScatterCreate_StoP() line 2804 in /home/jballani/software/petsc-3.7.1/src/vec/vec/utils/vpscat.c [1]PETSC ERROR: #3 VecScatterCreate_PtoP() line 2993 in /home/jballani/software/petsc-3.7.1/src/vec/vec/utils/vpscat.c [1]PETSC ERROR: #4 VecScatterCreate() line 1656 in /home/jballani/software/petsc-3.7.1/src/vec/vec/utils/vscat.c Looks more like a segfault, but yeah, that's not good. Can you go back to your original line of code and tack on PARALLEL for the type? PetscVector< Number > y( mesh.comm(), x->size(), PARALLEL ); I'm not sure what will happen in this case (creating a PARALLEL vector with n==n_local, seems like we should use PETSC_DECIDE there) but it might do the right thing. -- John ------------------------------------------------------------------------------ What NetFlow Analyzer can do for you? Monitors network bandwidth and traffic patterns at an interface-level. Reveals which users, apps, and protocols are consuming the most bandwidth. Provides multi-vendor support for NetFlow, J-Flow, sFlow and other flows. Make informed decisions using capacity planning reports.http://sdm.link/zohodev2dev _______________________________________________ Libmesh-users mailing list Libmesh-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/libmesh-users
