February 2016 - petsc-dev - mailman3.cels.anl.gov

Re: [petsc-dev] MatNest in petsc4py with empty submatrices
by Lawrence Mitchell 17 Feb '16

17 Feb '16

On 17/02/16 14:47, Chris Eldred wrote: > Hey All, > > I was wondering if anyone had used MatNest with petsc4py. I am having > issues trying to create a MatNest where some of the submatrices are > empty. In PETSc, I can just pass NULL in the array that I give to > MatCreateNest and it works fine. However, the following code We use it, but haven't used it with empty slots. > nest_list = [[None,None],[None,None]] > nest_list[0][0] = some matrix > nest_list[0][1] = some matrix > nest_list[1][0] = some matrix > A = PETSc.Mat().createNest(nest_list,comm=PETSc.COMM_WORLD) > > fails with: > File "Mat.pyx", line 445, in petsc4py.PETSc.Mat.createNest > (src/petsc4py.PETSc.c:103202) > TypeError: Cannot convert NoneType to petsc4py.PETSc.Mat > > Is there an equivalent to NULL in petsc4py (there doesn't appear to be one)? > > If all of the entries in nest_list are filled, everything works fine. > > Any help would be appreciated! Maybe a patch like this to petsc4py? diff --git a/src/PETSc/Mat.pyx b/src/PETSc/Mat.pyx index f281ba0..490c58e 100644 --- a/src/PETSc/Mat.pyx +++ b/src/PETSc/Mat.pyx @@ -460,7 +460,10 @@ cdef class Mat(Object): tmp1 = oarray_p(empty_p(nr*nc), NULL, <void**>&cmats) for i from 0 <= i < mr: for j from 0 <= j < mc: - cmats[i*mc+j] = (<Mat?>mats[i][j]).mat + if mats[i][j] is None: + cmats[i*mc+j] = NULL + else: + cmats[i*mc+j] = (<Mat?>mats[i][j]).mat if isrows is not None: tmp2 = oarray_p(empty_p(nr), NULL, <void**>&cisrows) for i from 0 <= i < mr: cisrows[i] = (<IS?>isrows[i]).iset Untested. If you want to pass null isrows and iscols as well you'll need to do a similar thing there. Cheers, Lawrence

1 0

MatNest in petsc4py with empty submatrices
by Chris Eldred 17 Feb '16

17 Feb '16

Hey All, I was wondering if anyone had used MatNest with petsc4py. I am having issues trying to create a MatNest where some of the submatrices are empty. In PETSc, I can just pass NULL in the array that I give to MatCreateNest and it works fine. However, the following code nest_list = [[None,None],[None,None]] nest_list[0][0] = some matrix nest_list[0][1] = some matrix nest_list[1][0] = some matrix A = PETSc.Mat().createNest(nest_list,comm=PETSc.COMM_WORLD) fails with: File "Mat.pyx", line 445, in petsc4py.PETSc.Mat.createNest (src/petsc4py.PETSc.c:103202) TypeError: Cannot convert NoneType to petsc4py.PETSc.Mat Is there an equivalent to NULL in petsc4py (there doesn't appear to be one)? If all of the entries in nest_list are filled, everything works fine. Any help would be appreciated! Thanks, Chris -- Chris Eldred Postdoctoral Fellow, LAGA, University of Paris 13 PhD, Atmospheric Science, Colorado State University, 2015 DOE Computational Science Graduate Fellow (Alumni) B.S. Applied Computational Physics, Carnegie Mellon University, 2009 chris.eldred(a)gmail.com

1 0

Re: [petsc-dev] reading vector from HDF5 with DMPlex
by Matthew Knepley 16 Feb '16

16 Feb '16

On Mon, Feb 15, 2016 at 4:22 PM, Adrian Croucher <a.croucher(a)auckland.ac.nz> wrote: > > > On 15/02/16 20:36, Matthew Knepley wrote: > > Sorry I took too long. It has been hectic this semester. The logic is > somewhat complicated with > regard to global and local here, but it is to ensure that boundary values > are put in on output. I > am not sure how to make it more transparent, but I am willing to take > suggestions. > > > Thanks Matt. > > I see now the documentation for VecLoad() does say "The input file must > contain the full global vector...". > > Maybe it'd be useful to add a note to the VecView() documentation saying > that the vectors to view should be global rather than local? > The problem is that VecView knows nothing about global/local. Its at the algebraic level. The DM is using extra information to organize the HDF5 file. It should be documented in the manual at least. Thanks, Matt > > - Adrian > > -- > Dr Adrian Croucher > Senior Research Fellow > Department of Engineering Science > University of Auckland, New Zealand > email: a.croucher(a)auckland.ac.nz > tel: +64 (0)9 923 84611 > > -- What most experimenters take for granted before they begin their experiments is infinitely more interesting than any results to which their experiments lead. -- Norbert Wiener

1 0

Re: [petsc-dev] reading vector from HDF5 with DMPlex
by Adrian Croucher 15 Feb '16

15 Feb '16

On 15/02/16 20:36, Matthew Knepley wrote: > Sorry I took too long. It has been hectic this semester. The logic is > somewhat complicated with > regard to global and local here, but it is to ensure that boundary > values are put in on output. I > am not sure how to make it more transparent, but I am willing to take > suggestions. Thanks Matt. I see now the documentation for VecLoad() does say "The input file must contain the full global vector...". Maybe it'd be useful to add a note to the VecView() documentation saying that the vectors to view should be global rather than local? - Adrian -- Dr Adrian Croucher Senior Research Fellow Department of Engineering Science University of Auckland, New Zealand email: a.croucher(a)auckland.ac.nz tel: +64 (0)9 923 84611

1 0

Re: [petsc-dev] Regression in Vec assembly
by Lisandro Dalcin 15 Feb '16

15 Feb '16

Any news about this regression? My PetIGA tests are still failing with petsc/master. On 21 October 2015 at 06:23, Jed Brown <jed(a)jedbrown.org> wrote: > Barry Smith <bsmith(a)mcs.anl.gov> writes: > >> Jed, >> >> Will you be working on this? I don't have a clue how your "new stuff" works. > > Yeah, I'll debug it. -- Lisandro Dalcin ============ Research Scientist Computer, Electrical and Mathematical Sciences & Engineering (CEMSE) Numerical Porous Media Center (NumPor) King Abdullah University of Science and Technology (KAUST) http://numpor.kaust.edu.sa/ 4700 King Abdullah University of Science and Technology al-Khawarizmi Bldg (Bldg 1), Office # 4332 Thuwal 23955-6900, Kingdom of Saudi Arabia http://www.kaust.edu.sa Office Phone: +966 12 808-0459

1 0

Re: [petsc-dev] reading vector from HDF5 with DMPlex
by Matthew Knepley 15 Feb '16

15 Feb '16

Sorry I took too long. It has been hectic this semester. The logic is somewhat complicated with regard to global and local here, but it is to ensure that boundary values are put in on output. I am not sure how to make it more transparent, but I am willing to take suggestions. Thanks, Matt On Sun, Feb 14, 2016 at 9:16 PM, Adrian Croucher <a.croucher(a)auckland.ac.nz> wrote: > OK, I think I've figured out what the problem was here. The > 'cell_geometry' vector I wrote to the HDF5 file was actually a local vector > (not global). If I first do DMLocalToGlobal on it and then VecView the > global vector, it turns up under the 'fields' group, and I can access it > for reading back in. > > Cheers, Adrian > > On 12/02/16 18:07, Adrian Croucher wrote: > >> hi >> >> I'm having trouble reading in a vector on a DMPlex from an HDF5 file. The >> attached minimal program (readdmvec2.F90) illustrates the problem. >> >> What it does is: >> - reads in a DMPlex from file (ghotcol.exo) and sets it up (distributes, >> constructs ghost cells and default section) >> - uses DMGetGlobalVector() to get the vector >> - sets its object name to 'cell_geometry' >> - opens the HDF5 file (hotcol_ss.h5) in a viewer and tries to read the >> vector using VecLoad() >> >> The VecLoad() dies, saying "object 'cell_geometry' doesn't exist" in the >> HDF5 file. (Full error message is below.) >> >> But it does exist- it shows up as a dataset in the '/' group in h5dump. I >> can also read it in from a Python script and h5py, e.g.: >> >> >>> import h5py >> >>> f = h5py.File('hotcol_ss.h5') >> >>> f['cell_geometry'][:] >> >> Any idea why VecLoad() can't find it? I get the same error running on any >> number of processors. >> >> The program that writes 'cell_geometry' to the HDF5 file also >> subsequently writes time-dependent simulation results to the same file >> (that's what the other datasets are in there) using >> DMSetOutputSequenceNumber() and VecView(). I am wondering if the problem is >> related to this, because if I just write a simple vector out to the file by >> itself it seems to read back in ok. >> >> Cheers, Adrian >> >> >> > -- > Dr Adrian Croucher > Senior Research Fellow > Department of Engineering Science > University of Auckland, New Zealand > email: a.croucher(a)auckland.ac.nz > tel: +64 (0)9 923 84611 > > -- What most experimenters take for granted before they begin their experiments is infinitely more interesting than any results to which their experiments lead. -- Norbert Wiener

1 0

Re: [petsc-dev] reading vector from HDF5 with DMPlex
by Adrian Croucher 14 Feb '16

14 Feb '16

OK, I think I've figured out what the problem was here. The 'cell_geometry' vector I wrote to the HDF5 file was actually a local vector (not global). If I first do DMLocalToGlobal on it and then VecView the global vector, it turns up under the 'fields' group, and I can access it for reading back in. Cheers, Adrian On 12/02/16 18:07, Adrian Croucher wrote: > hi > > I'm having trouble reading in a vector on a DMPlex from an HDF5 file. > The attached minimal program (readdmvec2.F90) illustrates the problem. > > What it does is: > - reads in a DMPlex from file (ghotcol.exo) and sets it up > (distributes, constructs ghost cells and default section) > - uses DMGetGlobalVector() to get the vector > - sets its object name to 'cell_geometry' > - opens the HDF5 file (hotcol_ss.h5) in a viewer and tries to read the > vector using VecLoad() > > The VecLoad() dies, saying "object 'cell_geometry' doesn't exist" in > the HDF5 file. (Full error message is below.) > > But it does exist- it shows up as a dataset in the '/' group in > h5dump. I can also read it in from a Python script and h5py, e.g.: > > >>> import h5py > >>> f = h5py.File('hotcol_ss.h5') > >>> f['cell_geometry'][:] > > Any idea why VecLoad() can't find it? I get the same error running on > any number of processors. > > The program that writes 'cell_geometry' to the HDF5 file also > subsequently writes time-dependent simulation results to the same file > (that's what the other datasets are in there) using > DMSetOutputSequenceNumber() and VecView(). I am wondering if the > problem is related to this, because if I just write a simple vector > out to the file by itself it seems to read back in ok. > > Cheers, Adrian > > -- Dr Adrian Croucher Senior Research Fellow Department of Engineering Science University of Auckland, New Zealand email: a.croucher(a)auckland.ac.nz tel: +64 (0)9 923 84611

1 0

reading vector from HDF5 with DMPlex
by Adrian Croucher 11 Feb '16

11 Feb '16

hi I'm having trouble reading in a vector on a DMPlex from an HDF5 file. The attached minimal program (readdmvec2.F90) illustrates the problem. What it does is: - reads in a DMPlex from file (ghotcol.exo) and sets it up (distributes, constructs ghost cells and default section) - uses DMGetGlobalVector() to get the vector - sets its object name to 'cell_geometry' - opens the HDF5 file (hotcol_ss.h5) in a viewer and tries to read the vector using VecLoad() The VecLoad() dies, saying "object 'cell_geometry' doesn't exist" in the HDF5 file. (Full error message is below.) But it does exist- it shows up as a dataset in the '/' group in h5dump. I can also read it in from a Python script and h5py, e.g.: >>> import h5py >>> f = h5py.File('hotcol_ss.h5') >>> f['cell_geometry'][:] Any idea why VecLoad() can't find it? I get the same error running on any number of processors. The program that writes 'cell_geometry' to the HDF5 file also subsequently writes time-dependent simulation results to the same file (that's what the other datasets are in there) using DMSetOutputSequenceNumber() and VecView(). I am wondering if the problem is related to this, because if I just write a simple vector out to the file by itself it seems to read back in ok. Cheers, Adrian -- HDF5-DIAG: Error detected in HDF5 (1.8.8) MPI-process 0: #000: ../../../src/H5D.c line 334 in H5Dopen2(): not found major: Dataset minor: Object not found #001: ../../../src/H5Gloc.c line 430 in H5G_loc_find(): can't find object major: Symbol table minor: Object not found #002: ../../../src/H5Gtraverse.c line 861 in H5G_traverse(): internal path traversal failed major: Symbol table minor: Object not found #003: ../../../src/H5Gtraverse.c line 641 in H5G_traverse_real(): traversal operator failed major: Symbol table minor: Callback failed #004: ../../../src/H5Gloc.c line 385 in H5G_loc_find_cb(): object 'cell_geometry' doesn't exist major: Symbol table minor: Object not found [0]PETSC ERROR: --------------------- Error Message -------------------------------------------------------------- [0]PETSC ERROR: Error in external library [0]PETSC ERROR: Error in HDF5 call H5Dopen2() Status -1 [0]PETSC ERROR: See http://www.mcs.anl.gov/petsc/documentation/faq.html for trouble shooting. [0]PETSC ERROR: Petsc Development GIT revision: v3.6.3-3170-g725a008 GIT Date: 2016-01-19 11:34:55 -0600 [0]PETSC ERROR: readdmvec2 `6` on a linux-gnu-c-opt named des108 by acro018 Fri Feb 12 17:36:00 2016 [0]PETSC ERROR: Configure options --download-netcdf --download-exodusii --with-hdf5-dir=/usr --download-triangle --download-ptscotch -- download-chaco [0]PETSC ERROR: #1 VecLoad_HDF5() line 269 in /home/acro018/software/PETSc/code/src/vec/vec/utils/vecio.c [0]PETSC ERROR: #2 VecLoad_Default() line 411 in /home/acro018/software/PETSc/code/src/vec/vec/utils/vecio.c [0]PETSC ERROR: #3 VecLoad_Plex_Local() line 214 in /home/acro018/software/PETSc/code/src/dm/impls/plex/plex.c [0]PETSC ERROR: #4 VecLoad_Plex_HDF5() line 211 in /home/acro018/software/PETSc/code/src/dm/impls/plex/plexhdf5.c [0]PETSC ERROR: #5 VecLoad_Plex() line 238 in /home/acro018/software/PETSc/code/src/dm/impls/plex/plex.c [0]PETSC ERROR: #6 VecLoad() line 975 in /home/acro018/software/PETSc/code/src/vec/vec/interface/vector.c -- Dr Adrian Croucher Senior Research Fellow Department of Engineering Science University of Auckland, New Zealand email: a.croucher(a)auckland.ac.nz tel: +64 (0)9 923 84611

1 0

PETSc User Meeting 2016
by Karl Rupp 05 Feb '16

05 Feb '16

Dear PETSc users, we would like to invite you to join us for a PETSc User Meeting held in Vienna, Austria, on June 28-30, 2016. The first day consists of tutorials on various aspects and features of PETSc. The second and third day are devoted to exchange, discussions, and a refinement of strategies for the future with our users. We encourage you to present work illustrating your own use of PETSc, for example in applications or in libraries built on top of PETSc. Registration for the PETSc User Meeting 2016 is free of charge. We can host a maximum of 120 participants, so register soon (but no later than May 1). We are still in the process of acquiring money for providing student travel grants; prospective sponsors, please contact us. Website with full information: http://www.mcs.anl.gov/petsc/meetings/2016/ We are looking forward to welcoming you in Vienna! Karl Rupp and the PETSc team

1 0

Re: [petsc-dev] Install step broken
by Garth N. Wells 04 Feb '16

04 Feb '16

Just to wrap this thread up, issue was fixed in https://bitbucket.org/petsc/petsc/commits/5a4feeed41cb1af9234d439bb06ea004d… . Garth On 1 February 2016 at 16:17, Garth N. Wells <gnw20(a)cam.ac.uk> wrote: > > On 28 January 2016 at 15:40, Matthew Knepley <knepley(a)gmail.com> wrote: > >> On Wed, Jan 27, 2016 at 9:10 AM, Garth N. Wells <gnw20(a)cam.ac.uk> wrote: >> >>> The 'install' step for master is broken for me since Matt's >>> change ddc4758 ( >>> https://bitbucket.org/petsc/petsc/commits/ddc47588dc986180af4263ab160cc4a55…) >>> I can build and install c502958 fine. >>> >> >> This change is only in an error branch, and should only be triggered by >> an unsuccessful configure. Can you send configure.log? >> >> > I had to unset my PYTHONPATH, which contained mpi4py and petsc4py. Could > something have been imported from petsc4py and messed things up? > > I can send configure.log to petsc-maint later, if it would be helpful. > > Garth > > > >> Thanks, >> >> Matt >> >> >>> Error at 'make install' is: >>> >>> No cached configure in RDict at >>> arch-linux2-c-opt/lib/petsc/conf/RDict.db >>> Traceback (most recent call last): >>> File "./config/install.py", line 380, in <module> >>> Installer(sys.argv[1:]).run() >>> File "./config/install.py", line 373, in run >>> self.runsetup() >>> File "./config/install.py", line 334, in runsetup >>> self.setup() >>> File "./config/install.py", line 60, in setup >>> self.setupModules() >>> File "./config/install.py", line 51, in setupModules >>> self.setCompilers = self.framework.require('config.setCompilers', >>> None) >>> AttributeError: 'NoneType' object has no attribute 'require' >>> makefile:261: recipe for target 'install' failed >>> >>> Garth >>> >> >> >> >> -- >> What most experimenters take for granted before they begin their >> experiments is infinitely more interesting than any results to which their >> experiments lead. >> -- Norbert Wiener >> > >

1 0