[petsc-users] How to define blocks for PCFIELDSPLIT?

Mon Mar 6 17:05:17 CST 2017

I tried to split my matrix (attached drawing of the non-null elements) by
calling PCFieldSplitSetIS() twice:

PCFieldSplitSetIS(pc, "p", zero_to_n);
PCFieldSplitSetIS(pc, "u", n_to_m);

But upon running with options:

-ksp_type preonly -pc_type fieldsplit

I get the following error:

[0]PETSC ERROR: --------------------- Error Message
--------------------------------------------------------------
[0]PETSC ERROR: Petsc has generated inconsistent data
[0]PETSC ERROR: Unhandled case, must have at least two fields, not 1
[0]PETSC ERROR: See http://www.mcs.anl.gov/petsc/documentation/faq.html for
trouble shooting.
[0]PETSC ERROR: Petsc Release Version 3.7.3, Jul, 24, 2016
[0]PETSC ERROR:
/home/lvella/src/higtree/examples/read-amr-solve-ns-cfl-example-2d on a
x86_64-linux-gnu-real named lvella-workstation by lvella Mon Mar  6
20:01:47 2017
[0]PETSC ERROR: Configure options --build=x86_64-linux-gnu --prefix=/usr
--includedir=${prefix}/include --mandir=${prefix}/share/man
--infodir=${prefix}/share/info --sysconfdir=/etc --localstatedir=/var
--with-silent-rules=0 --libdir=${prefix}/lib/x86_64-linux-gnu
--libexecdir=${prefix}/lib/x86_64-linux-gnu --with-maintainer-mode=0
--with-dependency-tracking=0 --with-debugging=0
--shared-library-extension=_real --with-hypre=1 --with-hypre-dir=/usr
--with-clanguage=C++ --with-shared-libraries --with-pic=1 --useThreads=0
--with-fortran-interfaces=1 --with-mpi-dir=/usr/lib/openmpi
--with-blas-lib=-lblas --with-lapack-lib=-llapack --with-blacs=1
--with-blacs-lib="-lblacsCinit-openmpi -lblacs-openmpi" --with-scalapack=1
--with-scalapack-lib=-lscalapack-openmpi --with-mumps=1
--with-mumps-include="[]" --with-mumps-lib="-ldmumps -lzmumps -lsmumps
-lcmumps -lmumps_common -lpord" --with-suitesparse=1
--with-suitesparse-include=/usr/include/suitesparse
--with-suitesparse-lib="-lumfpack -lamd -lcholmod -lklu" --with-spooles=1
--with-spooles-include=/usr/include/spooles --with-spooles-lib=-lspooles
--with-ptscotch=1 --with-ptscotch-include=/usr/include/scotch
--with-ptscotch-lib="-lptesmumps -lptscotch -lptscotcherr" --with-fftw=1
--with-fftw-include="[]" --with-fftw-lib="-lfftw3 -lfftw3_mpi"
--with-superlu=0 --CXX_LINKER_FLAGS=-Wl,--no-as-needed
--prefix=/usr/lib/petscdir/3.7.3/x86_64-linux-gnu-real
PETSC_DIR=/build/petsc-fA70UI/petsc-3.7.3.dfsg1
--PETSC_ARCH=x86_64-linux-gnu-real CFLAGS="-g -O2
-fdebug-prefix-map=/build/petsc-fA70UI/petsc-3.7.3.dfsg1=.
-fstack-protector-strong -Wformat -Werror=format-security -fPIC"
CXXFLAGS="-g -O2 -fdebug-prefix-map=/build/petsc-fA70UI/petsc-3.7.3.dfsg1=.
-fstack-protector-strong -Wformat -Werror=format-security -fPIC"
FCFLAGS="-g -O2 -fdebug-prefix-map=/build/petsc-fA70UI/petsc-3.7.3.dfsg1=.
-fstack-protector-strong -fPIC" FFLAGS="-g -O2
-fdebug-prefix-map=/build/petsc-fA70UI/petsc-3.7.3.dfsg1=.
-fstack-protector-strong -fPIC" CPPFLAGS="-Wdate-time -D_FORTIFY_SOURCE=2"
LDFLAGS="-Wl,-Bsymbolic-functions -Wl,-z,relro -fPIC" MAKEFLAGS=w
[0]PETSC ERROR: #1 PCFieldSplitSetDefaults() line 474 in
/build/petsc-fA70UI/petsc-3.7.3.dfsg1/src/ksp/pc/impls/fieldsplit/fieldsplit.c
[0]PETSC ERROR: #2 PCSetUp_FieldSplit() line 491 in
/build/petsc-fA70UI/petsc-3.7.3.dfsg1/src/ksp/pc/impls/fieldsplit/fieldsplit.c
[0]PETSC ERROR: #3 PCSetUp() line 968 in
/build/petsc-fA70UI/petsc-3.7.3.dfsg1/src/ksp/pc/interface/precon.c
[0]PETSC ERROR: #4 KSPSetUp() line 390 in
/build/petsc-fA70UI/petsc-3.7.3.dfsg1/src/ksp/ksp/interface/itfunc.c
[0]PETSC ERROR: #5 KSPSolve() line 599 in
/build/petsc-fA70UI/petsc-3.7.3.dfsg1/src/ksp/ksp/interface/itfunc.c

What is the proper way to setup PCFIELDSPLIT in this case?

2017-03-01 16:26 GMT-03:00 Barry Smith <bsmith at mcs.anl.gov>:

>
> > On Mar 1, 2017, at 12:59 PM, Lucas Clemente Vella <lvella at gmail.com>
> wrote:
> >
> > I have a parallel AIJ matrix and I know exactly which element belongs to
> each one of the 4 submatrices blocks I want to use to solve the linear
> system. The blocks are not strided, because they have different number of
> elements.
> >
> > I understand that I must use PCFieldSplitSetIS(), since
> PCFieldSplitSetFields() is only for strided blocks. What I don't understand
> is how to create the IS structure I must pass to it.
> >
> > Each matrix coefficient is identified by a pair (i, j), but on IS
> creation functions, like ISCreateGeneral() and ISCreateBlock(), I am
> supposed to provide a one dimension set of indices. How does these indices
> relates to the matrix coefficients?
>
>    PCFieldSplitSetIS() always indicates SQUARE blocks along the diagonal
> of the original matrix. Hence you need only one IS to define a block, you
> don't need one for the columns and one for the rows.  The IS is telling
> what rows AND columns you want in the block.
>
>
> > Also, ISCreateGeneral() seems to create a single block, and
> ISCreateBlock() seems to create multiple blocks of the same size.
>
>     ISCreateBlock() does not create multi blocks, it creates a single IS
> that has "block structure", for example 0,1, 3, 4, 6, 7, 9,10, ....
> > How to create multiple blocks with different sizes?
>
>    ISCreateGeneral().
>
>
> >
> > Thanks.
> >
> > --
> > Lucas Clemente Vella
> > lvella at gmail.com
>
>

-- 
Lucas Clemente Vella
lvella at gmail.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20170306/743f690b/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: matrix.png
Type: image/png
Size: 24869 bytes
Desc: not available
URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20170306/743f690b/attachment-0001.png>