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

Re: [petsc-dev] YAML patch
by Sean Farley 08 Aug '12

08 Aug '12

> This needs to be fixed too, then. Yes, hence my line "both approaches should be using Petscgetline()".

1 0

Re: [petsc-dev] YAML patch
by Blaise Bourdin 08 Aug '12

08 Aug '12

On Aug 8, 2012, at 9:39 AM, Sean Farley <sean(a)mcs.anl.gov> wrote: >> The existing, but non working, function is called PetscOptionsInsertFile_YAML. I just kept its name without checking the petsc developer style guide. My mistake. >> >> The attached patch fixes this issue. > > "There is a limit of 128,000 char on the file length." … What? The old > function had a (temporary) limit of 128,000 characters *per line* but > both approaches should be using Petscgetline(). Unless I am proven wrong, it seems that the old function also had a limit of 0 char it would _actually_ parse. src/sys/objects/options.c:388 /* Warning: assume a maximum size for all options in a string */ ierr = PetscMalloc(128000*sizeof(char),&vstring);CHKERRQ(ierr); vstring[0] = 0; This needs to be fixed too, then. Blaise -- Department of Mathematics and Center for Computation & Technology Louisiana State University, Baton Rouge, LA 70803, USA Tel. +1 (225) 578 1612, Fax +1 (225) 578 4276 http://www.math.lsu.edu/~bourdin

1 0

Re: [petsc-dev] YAML patch
by Sean Farley 08 Aug '12

08 Aug '12

> The existing, but non working, function is called PetscOptionsInsertFile_YAML. I just kept its name without checking the petsc developer style guide. My mistake. > > The attached patch fixes this issue. "There is a limit of 128,000 char on the file length." … What? The old function had a (temporary) limit of 128,000 characters *per line* but both approaches should be using Petscgetline().

1 0

Re: [petsc-dev] YAML patch
by Blaise Bourdin 08 Aug '12

08 Aug '12

Hi, The existing, but non working, function is called PetscOptionsInsertFile_YAML. I just kept its name without checking the petsc developer style guide. My mistake. The attached patch fixes this issue. Blaise > > Home come there is an _ before the YAML in the function name? We put in _ in function names for private functions and generally user called public functions (like this) would not have that. > > Barry > > On Aug 7, 2012, at 9:29 AM, Blaise Bourdin <bourdin(a)lsu.edu> wrote: > >> Hi, >> >> My YAML patch did not exactly raise much enthusiasm so I tried to clean it up a bit... >> The attached patch is another attempt at providing partial support for parsing YAML files: -options_file_yaml now works from C and fortran. It's logic was changed so that it does not require reinserting the whole option string. Aliases and sequences are still not supported. >> >> Is this patch worth being pushed to petsc-dev and petsc-3.3? >> >> Blaise >> >> <PetscOptionsInsertFile_YAML.patch> >> -- >> Department of Mathematics and Center for Computation & Technology >> Louisiana State University, Baton Rouge, LA 70803, USA >> Tel. +1 (225) 578 1612, Fax +1 (225) 578 4276 http://www.math.lsu.edu/~bourdin >> >> >> >> >> >> >> > -- Department of Mathematics and Center for Computation & Technology Louisiana State University, Baton Rouge, LA 70803, USA Tel. +1 (225) 578 1612, Fax +1 (225) 578 4276 http://www.math.lsu.edu/~bourdin

1 0

Re: [petsc-dev] YAML patch
by Barry Smith 07 Aug '12

07 Aug '12

Home come there is an _ before the YAML in the function name? We put in _ in function names for private functions and generally user called public functions (like this) would not have that. Barry On Aug 7, 2012, at 9:29 AM, Blaise Bourdin <bourdin(a)lsu.edu> wrote: > Hi, > > My YAML patch did not exactly raise much enthusiasm so I tried to clean it up a bit... > The attached patch is another attempt at providing partial support for parsing YAML files: -options_file_yaml now works from C and fortran. It's logic was changed so that it does not require reinserting the whole option string. Aliases and sequences are still not supported. > > Is this patch worth being pushed to petsc-dev and petsc-3.3? > > Blaise > > <PetscOptionsInsertFile_YAML.patch> > -- > Department of Mathematics and Center for Computation & Technology > Louisiana State University, Baton Rouge, LA 70803, USA > Tel. +1 (225) 578 1612, Fax +1 (225) 578 4276 http://www.math.lsu.edu/~bourdin > > > > > > >

1 0

Test example for PCBDDC
by Stefano Zampini 07 Aug '12

07 Aug '12

Finally I had some time to write a basic test example for PCBDDC ( http://petsc.cs.iit.edu/petsc/petsc-dev/rev/848c8bdc7bcb) It illustrates the basic features of MATIS assembling (which to my knowledge was not present in PETSc examples) and BDDC customization by solving a simple Laplacian discretized by the spectral element method with Gauss-Lobatto-Legendre quadrature rule. -- Stefano

1 0

YAML patch
by Blaise Bourdin 07 Aug '12

07 Aug '12

Hi, My YAML patch did not exactly raise much enthusiasm so I tried to clean it up a bit... The attached patch is another attempt at providing partial support for parsing YAML files: -options_file_yaml now works from C and fortran. It's logic was changed so that it does not require reinserting the whole option string. Aliases and sequences are still not supported. Is this patch worth being pushed to petsc-dev and petsc-3.3? Blaise -- Department of Mathematics and Center for Computation & Technology Louisiana State University, Baton Rouge, LA 70803, USA Tel. +1 (225) 578 1612, Fax +1 (225) 578 4276 http://www.math.lsu.edu/~bourdin

1 0

Slides of Elemental Interface
by Xuan Zhou 02 Aug '12

02 Aug '12

Hi Jeff, Here are the slides. Thanks. -- Best Regards, Xuan Zhou Ph.D. Candidate, Department of Applied Mathematics, Illinois Institute of Technology President, IIT SIAM Student Chapter

1 0

catching Fortran users forgetting ierr arg
by Barry Smith 02 Aug '12

02 Aug '12

Can we somehow use the PetscCheckPointer() in ALL our Fortran stubs to make sure the user has passed a reasonable ierr arg? Barry Begin forwarded message: > From: Jed Brown <jedbrown(a)mcs.anl.gov> > Subject: Re: [petsc-maint #126615] Problem, please help me > Date: August 2, 2012 9:51:11 AM CDT > To: petsc-maint(a)mcs.anl.gov, Vahid Nekouie <V.Nekouie2(a)lboro.ac.uk> > Reply-To: petsc-maint(a)mcs.anl.gov, Jed Brown <jedbrown(a)mcs.anl.gov> > > On Thu, Aug 2, 2012 at 6:56 AM, Vahid Nekouie <V.Nekouie2(a)lboro.ac.uk>wrote: > >> (gdb) run >> Starting program: /home/mmvn-extra/programs/navid/update.out >> [Thread debugging using libthread_db enabled] >> g_1 1 224 0 >> g_2 224 224 0 >> Came IN STEP: 1 >> Size u,X,Z,alpha: 761 1701 762 1822 >> ______ INCREMENT = 1 >> current time= 1.0000000000000000 dtime= 1.0000000000000000 >> >> Program received signal SIGSEGV, Segmentation fault. >> 0x0000000000450b70 in iscreategeneral_ (comm=0x14b55e0, n=0x7ffffffa79e8, >> idx=0x191d5b0, mode=0x7ffffffa79f8, is=0x7ffffffa79ec, >> __ierr=0x80000000) >> at generalf.c:48 >> 48 *__ierr = ISCreateGeneral( >> > > The address of ierr does not look valid. What does the call site (in your > Fortran) look like? >

1 0

Re: [petsc-dev] Semismooth Newton
by Todd Munson 02 Aug '12

02 Aug '12

> How does one precondition a system like that? Presumably, the user may at > best be able to provide a preconditioner for JF. > > If JF responds well to multigrid, then it may be possible to use AMG-KKT > to coarsen the constraint Hessian, but how does MG perform for the sum > JF + Hc? There is also the problem of smoothing, although, if JF and Jc > are sparse, then DD may be an option with a direct solve on subdomains. > > Anything more definitive on preconditioning these systems? If the constraints are linear, then there are no modifications to JF, so your preconditioner for JF should be fine. Then you either construct a preconditioner for the KKT system or you use something like what the Sandian's use for their optimization methods. For nonlinear; not a clue. If the symmetric part can be approximated well by a low-rank matrix, then you might combine the preconditioner for JF with a quasi-newton approximation to Hc? Cheers, Todd.

1 0