July 2012 - petsc-dev - mailman3.cels.anl.gov

Re: [petsc-dev] Fwd: [petsc-maint #124771] bug report
by Satish Balay 19 Jul '12

19 Jul '12

On Thu, 19 Jul 2012, Jed Brown wrote: > > > This is what I do for myself [with pine] > > > > Doesn't solve the problem because the user still has in their header > > the developers email address. > > For one - my replies [with pine] don't have that. Secondly - even if the developers email address is there - it shouldn't matter. We are primarily concerned with users using 'reply' command from their e-mail client. And it will honor 'reply-to:petsc-maint'. If the user is explicitly manipulating headers by other means - we can't [and shouldn't] be attempting to control that. Satish > This is something that a developer using certain hosts can do to make the > issue moot, not something that involves the user.

1 0

Re: [petsc-dev] Fwd: [petsc-maint #124771] bug report
by Jed Brown 19 Jul '12

19 Jul '12

On Thu, Jul 19, 2012 at 5:38 PM, Barry Smith <bsmith(a)mcs.anl.gov> wrote: > > This is what I do for myself [with pine] > > Doesn't solve the problem because the user still has in their header > the developers email address. > This is something that a developer using certain hosts can do to make the issue moot, not something that involves the user. > >> Alternative would be for petsc-maint to remove the user's email from > >> reply-to when it delivers to us, then match message headers so that our > >> replies which go only to petsc-maint would be forwarded to the original > >> reporter. Seems very fragile. > > That is what it should do. Nothing fragile about it. Not hard shit. > What happens when the user Cc's other people in a follow-up. We don't even see the Cc's, but petsc-maint keeps emailing them. Even if the user shortens the Cc list, does the server know to remove it? Suppose the initial report Cc's some people that might be interested, but later they remove those addresses and send something sensitive that the original Cc list should not be able to see. It would have to be written carefully to preserve email semantics while emulating all of us being a single address. I don't have a reason why it can't be done, but I think it's something that could easily be messed up. > > And then there will be no more private > > messages [between us] on petsc-maint :) > > Why? petsc-maint server will always bounce everything sent to it back > to the developers. How often does some not-to-be-named developer remove the user from the Cc list and make a snide remark? If you answered Never, you earn two points.

1 0

Re: [petsc-dev] Fwd: [petsc-maint #124771] bug report
by Barry Smith 19 Jul '12

19 Jul '12

On Jul 19, 2012, at 5:28 PM, Satish Balay wrote: > On Thu, 19 Jul 2012, Jed Brown wrote: > >> On Thu, Jul 19, 2012 at 5:06 PM, Barry Smith <bsmith(a)mcs.anl.gov> wrote: >> >>> >>> Satish, >>> >>> Is there anyway to change the configuration of petsc-maint so that >>> when developers respond it doesn't go directly to the user but instead goes >>> back to the petsc-maint server, it strips out the individual developers >>> email address and then forwards on to the user? >>> >> >> We would have to configure our mailers to set reply-to appropriately. > > This is what I do for myself [with pine] Doesn't solve the problem because the user still has in their header the developers email address. > >> Alternative would be for petsc-maint to remove the user's email from >> reply-to when it delivers to us, then match message headers so that our >> replies which go only to petsc-maint would be forwarded to the original >> reporter. Seems very fragile. That is what it should do. Nothing fragile about it. Not hard shit. > > I agree it can be fragile. > > For one - I'll have to check with systems if rt can be configured this > way. Also it would mean sender's info won't be in the header [so > where would it be?] Could add it to the message or subject > What about 'cc:' list from sender? [we would wan't > rt to strip this out aswell - and send e-mails to the cc:list > automaticaly from our replies]. Would strip those and handle in the same way as the users email. > And then there will be no more private > messages [between us] on petsc-maint :) Why? petsc-maint server will always bounce everything sent to it back to the developers. Barry > > Satish

1 0

Re: [petsc-dev] Fwd: [petsc-maint #124771] bug report
by Jed Brown 19 Jul '12

19 Jul '12

On Thu, Jul 19, 2012 at 5:28 PM, Satish Balay <balay(a)mcs.anl.gov> wrote: > This is what I do for myself [with pine] > Yeah, the problem is that some sending servers only allow you to set specific reply-to addresses. The biggest concern is gmail which only allows one person in all of gmail to have added petsc-maint(a)mcs.anl.gov as "one of their accounts". That person is Matt, so he sends mail as "Matt Knepley < petsc-maint(a)mcs.anl.gov>". > > > Alternative would be for petsc-maint to remove the user's email from > > reply-to when it delivers to us, then match message headers so that our > > replies which go only to petsc-maint would be forwarded to the original > > reporter. Seems very fragile. > > I agree it can be fragile. > > For one - I'll have to check with systems if rt can be configured this > way. Also it would mean sender's info won't be in the header [so > where would it be?] What about 'cc:' list from sender? [we would wan't > rt to strip this out aswell - and send e-mails to the cc:list > automaticaly from our replies]. And then there will be no more private > messages [between us] on petsc-maint :) > I think it's too dangerous at this point. We had discussions earlier about an email-integrated issue tracker, at which point this would be feasible.

1 0

Re: [petsc-dev] Fwd: [petsc-maint #124771] bug report
by Satish Balay 19 Jul '12

19 Jul '12

On Thu, 19 Jul 2012, Jed Brown wrote: > On Thu, Jul 19, 2012 at 5:06 PM, Barry Smith <bsmith(a)mcs.anl.gov> wrote: > > > > > Satish, > > > > Is there anyway to change the configuration of petsc-maint so that > > when developers respond it doesn't go directly to the user but instead goes > > back to the petsc-maint server, it strips out the individual developers > > email address and then forwards on to the user? > > > > We would have to configure our mailers to set reply-to appropriately. This is what I do for myself [with pine] > Alternative would be for petsc-maint to remove the user's email from > reply-to when it delivers to us, then match message headers so that our > replies which go only to petsc-maint would be forwarded to the original > reporter. Seems very fragile. I agree it can be fragile. For one - I'll have to check with systems if rt can be configured this way. Also it would mean sender's info won't be in the header [so where would it be?] What about 'cc:' list from sender? [we would wan't rt to strip this out aswell - and send e-mails to the cc:list automaticaly from our replies]. And then there will be no more private messages [between us] on petsc-maint :) Satish

1 0

Re: [petsc-dev] Fwd: [petsc-maint #124771] bug report
by Jed Brown 19 Jul '12

19 Jul '12

On Thu, Jul 19, 2012 at 5:06 PM, Barry Smith <bsmith(a)mcs.anl.gov> wrote: > > Satish, > > Is there anyway to change the configuration of petsc-maint so that > when developers respond it doesn't go directly to the user but instead goes > back to the petsc-maint server, it strips out the individual developers > email address and then forwards on to the user? > We would have to configure our mailers to set reply-to appropriately. Alternative would be for petsc-maint to remove the user's email from reply-to when it delivers to us, then match message headers so that our replies which go only to petsc-maint would be forwarded to the original reporter. Seems very fragile.

1 0

Fwd: [petsc-maint #124771] bug report
by Barry Smith 19 Jul '12

19 Jul '12

Satish, Is there anyway to change the configuration of petsc-maint so that when developers respond it doesn't go directly to the user but instead goes back to the petsc-maint server, it strips out the individual developers email address and then forwards on to the user? Barry Begin forwarded message: > From: Barry Smith <bsmith(a)mcs.anl.gov> > Date: July 19, 2012 5:04:14 PM CDT > To: Zahra Seifollahi <zahraseifollahi2009(a)gmail.com>, "petsc-maint(a)mcs.anl.gov Balay" <petsc-maint(a)mcs.anl.gov> > Subject: Re: [petsc-maint #124771] bug report > > > NEVER NEVER NEVER send petsc-maint email to individual petsc-developers; we have deadlines too. Always make sure it is sent to petsc-maint(a)mcs.anl.gov > > Barry > > On Jul 19, 2012, at 3:50 PM, Zahra Seifollahi wrote: > >> why do I get this? I'm terribly sorry for asking a lot, but my dead line is really soon and I don't have enough time for studying every thing. help me please. >> I also attached my program, I would be really grateful if you check it. >> >> $ /packages/petsc-3.3-p1/win32-gnu/bin/mpiexec -bootstrap fork -n 1 ./ex2 >> [0]PETSC ERROR: --------------------- Error Message ------------------------------------ >> [0]PETSC ERROR: Object is in wrong state! >> [0]PETSC ERROR: Not for unassembled matrix! >> [0]PETSC ERROR: ------------------------------------------------------------------------ >> [0]PETSC ERROR: Petsc Release Version 3.3.0, Patch 1, Fri Jun 15 09:30:49 CDT 2012 >> [0]PETSC ERROR: See docs/changes/index.html for recent updates. >> [0]PETSC ERROR: See docs/faq.html for hints about trouble shooting. >> [0]PETSC ERROR: See docs/index.html for manual pages. >> [0]PETSC ERROR: ------------------------------------------------------------------------ >> [0]PETSC ERROR: ./ex2 on a win32-gnu named ZAHRA-VAIO by Zahra Fri Jul 20 01:08:40 2012 >> [0]PETSC ERROR: Libraries linked from /packages/petsc-3.3-p1/win32-gnu/lib >> [0]PETSC ERROR: Configure run at Wed Jul 11 23:45:53 2012 >> [0]PETSC ERROR: Configure options PETSC_ARCH=win32-gnu --with-cc=gcc --with-cxx=g++ --with-fc=0 --with-clanguage=cxx --download-f2cblaslapack=1 --download-mpich=1 --download-parmetis=1 --downloa >> d-boost=1 --download-metis=1 --useThreads=0 >> [0]PETSC ERROR: ------------------------------------------------------------------------ >> [0]PETSC ERROR: MatGetOrdering() line 192 in /packages/petsc-3.3-p1/src/mat/order/sorder.c >> [0]PETSC ERROR: PCSetup_ICC() line 14 in /packages/petsc-3.3-p1/src/ksp/pc/impls/factor/icc/icc.c >> [0]PETSC ERROR: PCSetUp() line 832 in /packages/petsc-3.3-p1/src/ksp/pc/interface/precon.c >> [0]PETSC ERROR: KSPSetUp() line 278 in /packages/petsc-3.3-p1/src/ksp/ksp/interface/itfunc.c >> [0]PETSC ERROR: main() line 479 in src/ksp/ksp/examples/tutorials/ex2.c >> application called MPI_Abort(MPI_COMM_WORLD, 73) - process 0 >> [cli_0]: aborting job: >> application called MPI_Abort(MPI_COMM_WORLD, 73) - process 0 >> >> and also >> >> $ /packages/petsc-3.3-p1/win32-gnu/bin/mpiexec -bootstrap fork -n 2 ./ex2 -mat_view >> Matrix Object: 1 MPI processes >> type: mpiaij >> row 0: (0, 0) (1, 1) (2, 1) >> row 1: (0, 0) (1, 0) (2, 1) >> row 2: (0, 0) (1, 0) (2, 0) >> [0]PETSC ERROR: --------------------- Error Message ------------------------------------ >> [0]PETSC ERROR: No support for this operation for this object type! >> [0]PETSC ERROR: Only local values currently supported! >> [0]PETSC ERROR: ------------------------------------------------------------------------ >> [0]PETSC ERROR: Petsc Release Version 3.3.0, Patch 1, Fri Jun 15 09:30:49 CDT 2012 >> [0]PETSC ERROR: See docs/changes/index.html for recent updates. >> [0]PETSC ERROR: See docs/faq.html for hints about trouble shooting. >> [0]PETSC ERROR: See docs/index.html for manual pages. >> [0]PETSC ERROR: ------------------------------------------------------------------------ >> [0]PETSC ERROR: ./ex2 on a win32-gnu named ZAHRA-VAIO by Zahra Fri Jul 20 01:13:49 2012 >> [0]PETSC ERROR: Libraries linked from /packages/petsc-3.3-p1/win32-gnu/lib >> [0]PETSC ERROR: Configure run at Wed Jul 11 23:45:53 2012 >> [0]PETSC ERROR: Configure options PETSC_ARCH=win32-gnu --with-cc=gcc --with-cxx=g++ --with-fc=0 --with-clanguage=cxx --download-f2cblaslapack=1 --download-mpich=1 --download-parmetis=1 --downloa >> d-boost=1 --download-metis=1 --useThreads=0 >> [0]PETSC ERROR: ------------------------------------------------------------------------ >> [0]PETSC ERROR: MatGetValues_MPIAIJ() line 598 in /packages/petsc-3.3-p1/src/mat/impls/aij/mpi/mpiaij.c >> [0]PETSC ERROR: MatGetValues() line 1602 in /packages/petsc-3.3-p1/src/mat/interface/matrix.c >> [0]PETSC ERROR: --------------------- Error Message ------------------------------------ >> [0]PETSC ERROR: No support for this operation for this object type! >> [0]PETSC ERROR: Only local values currently supported! >> [0]PETSC ERROR: ------------------------------------------------------------------------ >> [0]PETSC ERROR: Petsc Release Version 3.3.0, Patch 1, Fri Jun 15 09:30:49 CDT 2012 >> [0]PETSC ERROR: See docs/changes/index.html for recent updates. >> .... >> >> [0]PETSC ERROR: --------------------- Error Message ------------------------------------ >> [0]PETSC ERROR: Object is in wrong state! >> [0]PETSC ERROR: Not for unassembled matrix! >> [0]PETSC ERROR: ------------------------------------------------------------------------ >> [0]PETSC ERROR: Petsc Release Version 3.3.0, Patch 1, Fri Jun 15 09:30:49 CDT 2012 >> [0]PETSC ERROR: See docs/changes/index.html for recent updates. >> [0]PETSC ERROR: See docs/faq.html for hints about trouble shooting. >> [0]PETSC ERROR: See docs/index.html for manual pages. >> [0]PETSC ERROR: ------------------------------------------------------------------------ >> [0]PETSC ERROR: ./ex2 on a win32-gnu named ZAHRA-VAIO by Zahra Fri Jul 20 01:13:49 2012 >> [0]PETSC ERROR: Libraries linked from /packages/petsc-3.3-p1/win32-gnu/lib >> [0]PETSC ERROR: Configure run at Wed Jul 11 23:45:53 2012 >> [0]PETSC ERROR: Configure options PETSC_ARCH=win32-gnu --with-cc=gcc --with-cxx=g++ --with-fc=0 --with-clanguage=cxx --download-f2cblaslapack=1 --download-mpich=1 --download-parmetis=1 --downloa >> d-boost=1 --download-metis=1 --useThreads=0 >> [0]PETSC ERROR: ------------------------------------------------------------------------ >> [0]PETSC ERROR: MatGetDiagonalBlock() line 91 in /packages/petsc-3.3-p1/src/mat/interface/matrix.c >> [0]PETSC ERROR: PCSetUp_BJacobi() line 130 in /packages/petsc-3.3-p1/src/ksp/pc/impls/bjacobi/bjacobi.c >> [0]PETSC ERROR: PCSetUp() line 832 in /packages/petsc-3.3-p1/src/ksp/pc/interface/precon.c >> [0]PETSC ERROR: KSPSetUp() line 278 in /packages/petsc-3.3-p1/src/ksp/ksp/interface/itfunc.c >> [0]PETSC ERROR: main() line 479 in src/ksp/ksp/examples/tutorials/ex2.c >> application called MPI_Abort(MPI_COMM_WORLD, 73) - process 0 >> [cli_0]: aborting job: >> application called MPI_Abort(MPI_COMM_WORLD, 73) - process 0 >> >> how can I handle these problems? >> >> On Fri, Jul 20, 2012 at 12:58 AM, Zahra Seifollahi <zahraseifollahi2009(a)gmail.com> wrote: >> that error is ok now >> >> >> On Fri, Jul 20, 2012 at 12:36 AM, Zahra Seifollahi <zahraseifollahi2009(a)gmail.com> wrote: >> I used MatDuplicate and it's ok now, but again I've got this: >> $ /packages/petsc-3.3-p1/win32-gnu/bin/mpiexec -bootstrap fork -n 1 ./ex2 >> [0]PETSC ERROR: --------------------- Error Message ------------------------------------ >> [0]PETSC ERROR: Argument out of range! >> [0]PETSC ERROR: Invalid stage -1 should be in [0,1)! >> [0]PETSC ERROR: ------------------------------------------------------------------------ >> [0]PETSC ERROR: Petsc Release Version 3.3.0, Patch 1, Fri Jun 15 09:30:49 CDT 2012 >> [0]PETSC ERROR: See docs/changes/index.html for recent updates. >> [0]PETSC ERROR: See docs/faq.html for hints about trouble shooting. >> [0]PETSC ERROR: See docs/index.html for manual pages. >> [0]PETSC ERROR: ------------------------------------------------------------------------ >> [0]PETSC ERROR: ./ex2 on a win32-gnu named ZAHRA-VAIO by Zahra Fri Jul 20 00:33:01 2012 >> [0]PETSC ERROR: Libraries linked from /packages/petsc-3.3-p1/win32-gnu/lib >> [0]PETSC ERROR: Configure run at Wed Jul 11 23:45:53 2012 >> [0]PETSC ERROR: Configure options PETSC_ARCH=win32-gnu --with-cc=gcc --with-cxx=g++ --with-fc=0 --with-clanguage=cxx --download-f2cblaslapack=1 --download-mpich=1 --download-parmetis=1 --downloa >> d-boost=1 --download-metis=1 --useThreads=0 >> [0]PETSC ERROR: ------------------------------------------------------------------------ >> [0]PETSC ERROR: PetscStageLogGetClassPerfLog() line 353 in /packages/petsc-3.3-p1/src/sys/plog/utils/stageLog.c >> [0]PETSC ERROR: PetscLogObjCreateDefault() line 309 in /packages/petsc-3.3-p1/src/sys/plog/utils/classLog.c >> [0]PETSC ERROR: MatCreate() line 74 in /packages/petsc-3.3-p1/src/mat/utils/gcreate.c >> [0]PETSC ERROR: MatDuplicate_SeqAIJ() line 3868 in /packages/petsc-3.3-p1/src/mat/impls/aij/seq/aij.c >> [0]PETSC ERROR: MatDuplicate() line 3963 in /packages/petsc-3.3-p1/src/mat/interface/matrix.c >> [0]PETSC ERROR: --------------------- Error Message ------------------------------------ >> [0]PETSC ERROR: Null argument, when expecting valid pointer! >> [0]PETSC ERROR: Null Object: Parameter # 1! >> [0]PETSC ERROR: ------------------------------------------------------------------------ >> [0]PETSC ERROR: Petsc Release Version 3.3.0, Patch 1, Fri Jun 15 09:30:49 CDT 2012 >> [0]PETSC ERROR: See docs/changes/index.html for recent updates. >> [0]PETSC ERROR: See docs/faq.html for hints about trouble shooting. >> [0]PETSC ERROR: See docs/index.html for manual pages. >> [0]PETSC ERROR: ------------------------------------------------------------------------ >> [0]PETSC ERROR: ./ex2 on a win32-gnu named ZAHRA-VAIO by Zahra Fri Jul 20 00:33:01 2012 >> [0]PETSC ERROR: Libraries linked from /packages/petsc-3.3-p1/win32-gnu/lib >> [0]PETSC ERROR: Configure run at Wed Jul 11 23:45:53 2012 >> [0]PETSC ERROR: Configure options PETSC_ARCH=win32-gnu --with-cc=gcc --with-cxx=g++ --with-fc=0 --with-clanguage=cxx --download-f2cblaslapack=1 --download-mpich=1 --download-parmetis=1 --downloa >> d-boost=1 --download-metis=1 --useThreads=0 >> [0]PETSC ERROR: ------------------------------------------------------------------------ >> [0]PETSC ERROR: MatSetUp() line 630 in /packages/petsc-3.3-p1/src/mat/interface/matrix.c >> [0]PETSC ERROR: main() line 352 in src/ksp/ksp/examples/tutorials/ex2.c >> application called MPI_Abort(MPI_COMM_WORLD, 85) - process 0 >> [cli_0]: aborting job: >> application called MPI_Abort(MPI_COMM_WORLD, 85) - process 0 >> >> I don't know what is about! wouldn't these errors finish?! >> >> On Fri, Jul 20, 2012 at 12:29 AM, Zahra Seifollahi <zahraseifollahi2009(a)gmail.com> wrote: >> I need a copy of K matrix with all of these elements, should I use MatDuplicate()? >> >> >> On Thu, Jul 19, 2012 at 11:49 PM, Barry Smith <bsmith(a)mcs.anl.gov> wrote: >> >> You need to create and setup the second matrix before copying into it. That is kk must be a valid matrix. Are you sure you do not want MatDuplicate()? >> On Jul 19, 2012, at 2:08 PM, Zahra Seifollahi wrote: >> >>> thank you. And then what's wrong with this? >>> >>> *[0]PETSC ERROR: --------------------- Error Message >>> ------------------------------------* >>> *[0]PETSC ERROR: Invalid argument!* >>> *[0]PETSC ERROR: Wrong type of object: Parameter # 2!* >>> *[0]PETSC ERROR: >>> ------------------------------------------------------------------------* >>> *[0]PETSC ERROR: Petsc Release Version 3.3.0, Patch 1, Fri Jun 15 09:30:49 >>> CDT 2012* >>> *[0]PETSC ERROR: See docs/changes/index.html for recent updates.* >>> *[0]PETSC ERROR: See docs/faq.html for hints about trouble shooting.* >>> *[0]PETSC ERROR: See docs/index.html for manual pages.* >>> *[0]PETSC ERROR: >>> ------------------------------------------------------------------------* >>> *[0]PETSC ERROR: ./ex2 on a win32-gnu named ZAHRA-VAIO by Zahra Thu Jul 19 >>> 23:34:11 2012* >>> *[0]PETSC ERROR: Libraries linked from /packages/petsc-3.3-p1/win32-gnu/lib* >>> *[0]PETSC ERROR: Configure run at Wed Jul 11 23:45:53 2012* >>> *[0]PETSC ERROR: Configure options PETSC_ARCH=win32-gnu --with-cc=gcc >>> --with-cxx=g++ --with-fc=0 --with-clanguage=cxx --download-f2cblaslapack=1 >>> --download-mpich=1 --download-parmetis=1 --downloa* >>> *d-boost=1 --download-metis=1 --useThreads=0* >>> *[0]PETSC ERROR: >>> ------------------------------------------------------------------------* >>> *[0]PETSC ERROR: MatCopy() line 3614 in >>> /packages/petsc-3.3-p1/src/mat/interface/matrix.c* >>> *[0]PETSC ERROR: --------------------- Error Message >>> ------------------------------------* >>> *[0]PETSC ERROR: Invalid argument!* >>> *[0]PETSC ERROR: Wrong type of object: Parameter # 1!* >>> *[0]PETSC ERROR: >>> ------------------------------------------------------------------------* >>> *[0]PETSC ERROR: Petsc Release Version 3.3.0, Patch 1, Fri Jun 15 09:30:49 >>> CDT 2012* >>> *[0]PETSC ERROR: See docs/changes/index.html for recent updates.* >>> *[0]PETSC ERROR: See docs/faq.html for hints about trouble shooting.* >>> *[0]PETSC ERROR: See docs/index.html for manual pages.* >>> *[0]PETSC ERROR: >>> ------------------------------------------------------------------------* >>> *[0]PETSC ERROR: ./ex2 on a win32-gnu named ZAHRA-VAIO by Zahra Thu Jul 19 >>> 23:34:11 2012* >>> *[0]PETSC ERROR: Libraries linked from /packages/petsc-3.3-p1/win32-gnu/lib* >>> *[0]PETSC ERROR: Configure run at Wed Jul 11 23:45:53 2012* >>> *[0]PETSC ERROR: Configure options PETSC_ARCH=win32-gnu --with-cc=gcc >>> --with-cxx=g++ --with-fc=0 --with-clanguage=cxx --download-f2cblaslapack=1 >>> --download-mpich=1 --download-parmetis=1 --downloa* >>> *d-boost=1 --download-metis=1 --useThreads=0* >>> *[0]PETSC ERROR: >>> ------------------------------------------------------------------------* >>> *[0]PETSC ERROR: MatSetUp() line 630 in >>> /packages/petsc-3.3-p1/src/mat/interface/matrix.c* >>> *[0]PETSC ERROR: main() line 352 in src/ksp/ksp/examples/tutorials/ex2.c* >>> *application called MPI_Abort(MPI_COMM_WORLD, 62) - process 0* >>> *[cli_0]: aborting job:* >>> *application called MPI_Abort(MPI_COMM_WORLD, 62) - process 0* >>> * >>> * >>> I copied K matrix into K matrix in my program by this command: >>> ierr = MatCopy(K,KK,SAME_NONZERO_PATTERN); >>> Is it ok then? >>> >>> and I've got this during making: >>> *ex2.c:348:52: warning: ?KK? may be used uninitialized in this function* >>> >>> On Thu, Jul 19, 2012 at 11:07 PM, Barry Smith <bsmith(a)mcs.anl.gov> wrote: >>> >>>> >>>> It only needs to be called once for each Mat. >>>> >>>> Barry >>>> >>>> On Jul 19, 2012, at 1:12 PM, Zahra Seifollahi wrote: >>>> >>>>> Actually I've done, should I call it before all setting values in the >>>> matrix or just one time? >>>>> >>>>> On Thu, Jul 19, 2012 at 10:35 PM, Barry Smith <bsmith(a)mcs.anl.gov> >>>> wrote: >>>>> >>>>> You need to call either MatSetUp() or >>>> http://www.mcs.anl.gov/petsc/petsc-current/docs/manualpages/Mat/MatXAIJSetP… before setting values in the matrix. >>>>> >>>>> This was a change introduced into petsc-3.3 >>>>> >>>>> Barry >>>>> >>>>> On Jul 19, 2012, at 12:36 PM, Zahra Seifollahi wrote: >>>>> >>>>>> Hello. >>>>>> I've written a program based on petsc but when I tried to run this >>>> program >>>>>> I've got this error: >>>>>> >>>>>> *ackages/petsc-3.3-p1/win32-gnu/bin/mpiexec -bootstrap fork -n 1 ./ex2 >>>>>> -mat_type mpiaij* >>>>>> *ETSC ERROR: --------------------- Error Message >>>>>> ------------------------------------* >>>>>> *ETSC ERROR: Object is in wrong state!* >>>>>> *ETSC ERROR: Mat object's type is not set: Argument # 1!* >>>>>> *ETSC ERROR: >>>>>> >>>> ------------------------------------------------------------------------* >>>>>> *ETSC ERROR: Petsc Release Version 3.3.0, Patch 1, Fri Jun 15 09:30:49 >>>> CDT >>>>>> 2012* >>>>>> *ETSC ERROR: See docs/changes/index.html for recent updates.* >>>>>> *ETSC ERROR: See docs/faq.html for hints about trouble shooting.* >>>>>> *ETSC ERROR: See docs/index.html for manual pages.* >>>>>> *ETSC ERROR: >>>>>> >>>> ------------------------------------------------------------------------* >>>>>> *ETSC ERROR: ./ex2 on a win32-gnu named ZAHRA-VAIO by Zahra Thu Jul 19 >>>>>> 22:02:55 2012* >>>>>> *ETSC ERROR: Libraries linked from >>>> /packages/petsc-3.3-p1/win32-gnu/lib* >>>>>> *ETSC ERROR: Configure run at Wed Jul 11 23:45:53 2012* >>>>>> *ETSC ERROR: Configure options PETSC_ARCH=win32-gnu --with-cc=gcc >>>>>> --with-cxx=g++ --with-fc=0 --with-clanguage=cxx >>>> --download-f2cblaslapack=1 >>>>>> --download-mpich=1 --download-parmetis=1 --downloa* >>>>>> *ost=1 --download-metis=1 --useThreads=0* >>>>>> *ETSC ERROR: >>>>>> >>>> ------------------------------------------------------------------------* >>>>>> *ETSC ERROR: MatSetValues() line 988 in >>>>>> /packages/petsc-3.3-p1/src/mat/interface/matrix.c* >>>>>> *ETSC ERROR: main() line 129 in src/ksp/ksp/examples/tutorials/ex2.c* >>>>>> *ication called MPI_Abort(MPI_COMM_WORLD, 73) - process 0* >>>>>> *_0]: aborting job:* >>>>>> *ication called MPI_Abort(MPI_COMM_WORLD, 73) - process 0 * >>>>>> >>>>>> What's wrong with it? What should I do to figure out this problem? >>>> help me >>>>>> please. >>>>>> -- >>>>>> Zahra Seifollahi >>>>>> KNTU University of Technology >>>>>> Aerospace engineering >>>>>> >>>>> >>>>> >>>>> >>>>> >>>>> -- >>>>> Zahra Seifollahi >>>>> KNTU University of Technology >>>>> Aerospace engineering >>>>> >>>> >>>> >>> >>> >>> -- >>> Zahra Seifollahi >>> KNTU University of Technology >>> Aerospace engineering >>> >> >> >> >> >> -- >> Zahra Seifollahi >> KNTU University of Technology >> Aerospace engineering >> >> >> >> >> -- >> Zahra Seifollahi >> KNTU University of Technology >> Aerospace engineering >> >> >> >> >> -- >> Zahra Seifollahi >> KNTU University of Technology >> Aerospace engineering >> >> >> >> >> -- >> Zahra Seifollahi >> KNTU University of Technology >> Aerospace engineering >> >> <ex2.c> >

1 0

Re: [petsc-dev] A small bug
by Barry Smith 19 Jul '12

19 Jul '12

Adam, Good eye for detecting this; it was an error. I have pushed a fix for 3.3 and dev but rather than changing the diagonal entry I have changed the rhs for those locations. The reason is that for the geometric multigrid to converge well it is better to scale the "zeroed-out rows" diagonal associated with Dirichlet boundary conditions with a value near the rest of the diagonal entries. Barry On Jul 19, 2012, at 12:59 PM, Adam Rose wrote: > To whom it may concern, > > I believe that I've found a small error in some example code located in $PETSC_DIR/src/ksp/ksp/examples/tutorials/ex45.c, and I just wanted to bring it to your attention. It's a Laplace equation equation example with Dirichlet boundary conditions u=1 on the square [0,1]x[0,1], but on running the program I noticed that the boundary condition wasn't consistent with the solution. This seems due to line 119 which sets the diagonal values of our discretization matrix in the case of boundary points. It currently reads "v[0] = 2.0*(HxHydHz + HxHzdHy + HyHzdHx);" which is the same as the diagonal away from the boundary, but a quick change to "v[0] = Hx*Hy*Hz;" produces the desired result. > > Thanks for reading, > Adam Rose (Intern, Tech-X Corporation)

1 0

Re: [petsc-dev] DMGetNamedGlobalVector() implementation?
by Barry Smith 19 Jul '12

19 Jul '12

On Jul 19, 2012, at 1:54 PM, Jed Brown wrote: > On Thu, Jul 19, 2012 at 1:50 PM, Barry Smith <bsmith(a)mcs.anl.gov> wrote: > Is there a way to fix the reference loop problem in PetscObjectCompose()? (Since you have a magic fix for DMGetNamedGlobalVector() can we use that everywhere) > > Look at the ugly code in DMDestroy that takes care of the work vectors and named list. Note that the API maintains that these named vectors have reverse links. If you allow a mix, it gets messier. What about indirect loops? > > > You have to admit that introducing in a bunch of places separate constructs for containment rather than using a single universal construct is not good software development. > > I agree, but I could not find an easy way. Can you find a difficult way ? We chose to make the DM both a factory for creating stuff and a place to stash stuff that people can pull out later, this will lead to more and more of these loops. We could 1) do proper garbage collection always (yuck from the person who has to write that perspective) 2) have separate factories that the user manage; so basically split the DM into two parts that do not reference each other. (Yuck from the user perspective) 3) continue with all these hacks to prevent loops 4) something else?

1 0

Re: [petsc-dev] DMGetNamedGlobalVector() implementation?
by Jed Brown 19 Jul '12

19 Jul '12

On Thu, Jul 19, 2012 at 1:50 PM, Barry Smith <bsmith(a)mcs.anl.gov> wrote: > Is there a way to fix the reference loop problem in PetscObjectCompose()? > (Since you have a magic fix for DMGetNamedGlobalVector() can we use that > everywhere) > Look at the ugly code in DMDestroy that takes care of the work vectors and named list. Note that the API maintains that these named vectors have reverse links. If you allow a mix, it gets messier. What about indirect loops? > > You have to admit that introducing in a bunch of places separate > constructs for containment rather than using a single universal construct > is not good software development. > I agree, but I could not find an easy way.

1 0