[petsc-dev] asm / gasm

Thu Jun 23 00:08:29 CDT 2016

[adding Garth]

On Thu, Jun 23, 2016 at 12:52 AM, Barry Smith <bsmith at mcs.anl.gov> wrote:

>
>   Mark,
>
>    I think there is a misunderstanding here. With GASM an individual block
> problem is __solved__ (via a parallel KSP) in parallel by several
> processes, with ASM each block is "owned" by and solved on a single process.
>

Ah, OK, so this is for multiple processors in a block. Yes, we are looking
at small, smother, blocks.

>
>    With both the "block" can come from any unknowns on any processes. You
> can have, for example a block that comes from a region snaking across
> several processes if you like (or it makes sense due to coupling in the
> matrix).
>
>    By default if you use ASM it will create one non-overlapping block
> defined by all unknowns owned by a single process and then extend it by
> "one level" (defined by the nonzero structure of the matrix) to get
> overlapping.

The default in ASM is one level of overlap? That is new.  (OK, I have not
looked at ASM in like over 10 years)

> If you use multiple blocks per process it defines the non-overlapping
> blocks within a single process's unknowns

I assume this still chops the matrix and does not call a partitioner.

> and extends each of them to have overlap (again by the non-zero structure
> of the matrix). The default is simple because the user only need indicate
> the number of blocks per process, the drawback is of course that it does
> depend on the process layout, number of processes etc and does not take
> into account particular "coupling information" that the user may know about
> with their problem.
>
>   If the user wishes to defined the blocks themselves that is also
> possible with PCASMSetSubLocalSubdomains(). Each process provides 1 or more
> index sets for the subdomains it will solve on. Note that the index sets
> can contain any unknowns in the entire problem so the blocks do not have to
> "line up" with the parallel decomposition at all.

Oh, OK, this is what I want. (I thought this worked).

> Of course determining and providing good such subdomains may not always be
> clear.
>

In smoothed aggregation there is an argument that the aggregates are good,
but the scale is fixed obviously.  On a regular grid smoothed aggregation
wants 3^D sized aggregates, which is obviously wonderful for AMS.  And for
anisotropy you want your ASM blocks to be on strongly connected components,
which is what smoothed aggregation wants (not that I do this very well).

>
>   I see in GAMG you have PCGAMGSetUseASMAggs

But the code calls PCGASMSetSubdomains and the command line
is -pc_gamg_use_agg_gasm, so this is all messed up.  (more below)

> which sadly does not have an explanation in the users manual and sadly
> does not have a matching options data base name -pc_gamg_use_agg_gasm
> following the rule of drop the word set, all lower case, and put _ between
> words the option should be -pc_gamg_use_asm_aggs.
>

BUT, THIS IS THE WAY IT WAS!  It looks like someone hijacked this code and
made it gasm.  I never did this.

Barry: you did this apparently in 2013.

>    In addition to this one you could also have one that uses the aggs but
> use the PCASM to manage the solves instead of GASM, it would likely be less
> buggy and more efficient.

yes

>
>   Please tell me exactly what example you tried to run with what options
> and I will debug it.

We got an error message:

** Max-trans not allowed because matrix is distributed

Garth: is this from your code perhaps? I don't see it in PETSc.

> Note that ALL functionality that is included in PETSc should have tests
> that test that functionality then we will find out immediately when it is
> broken instead of two years later when it is much harder to debug. If this
> -pc_gamg_use_agg_gasm had had a test we won't be in this mess now. (Jed's
> damn code reviews sure don't pick up this stuff).
>

First we need to change gasm to asm.

We could add this argument pc_gamg_use_agg_asm  to ksp/ex56 (runex56 or
make a new test).  The SNES version (also ex56) is my current test that I
like to refer to as recommended parameters for elasticity. So I'd like to
keep that clean, but we can add junk to ksp/ex56.

I've done this in a branch mark/gamg-agg-asm.  I get an error (appended).
It looks like the second coarsest grid, which has 36 dof on one processor
has an index 36 in the block on every processor. Strange.  I can take a
look at it later.

Mark

> [3]PETSC ERROR: [4]PETSC ERROR: --------------------- Error Message
--------------------------------------------------------------
> [4]PETSC ERROR: Petsc has generated inconsistent data
> [4]PETSC ERROR: ith 0 block entry 36 not owned by any process, upper
bound 36
> [4]PETSC ERROR: See http://www.mcs.anl.gov/petsc/documentation/faq.html
for trouble shooting.
> [4]PETSC ERROR: Petsc Development GIT revision: v3.7.2-630-g96e0c40  GIT
Date: 2016-06-22 10:03:02 -0500
> [4]PETSC ERROR: ./ex56 on a arch-macosx-gnu-g named MarksMac-3.local by
markadams Thu Jun 23 06:53:27 2016
> [4]PETSC ERROR: Configure options COPTFLAGS="-g -O0" CXXOPTFLAGS="-g -O0"
FOPTFLAGS="-g -O0" --download-hypre=1 --download-parmetis=1
--download-metis=1 --download-ml=1 --download-p4est=1 --download-exodus=1
--download-triangle=1 --with-hdf5-dir=/Users/markadams/Codes/hdf5
--with-x=0 --with-debugging=1 PETSC_ARCH=arch-macosx-gnu-g --download-chaco
> [4]PETSC ERROR: #1 VecScatterCreate_PtoS() line 2348 in
/Users/markadams/Codes/petsc/src/vec/vec/utils/vpscat.c
> [4]PETSC ERROR: #2 VecScatterCreate() line 1552 in
/Users/markadams/Codes/petsc/src/vec/vec/utils/vscat.c
> [4]PETSC ERROR: Petsc has generated inconsistent data
> [3]PETSC ERROR: ith 0 block entry 36 not owned by any process, upper
bound 36
> [3]PETSC ERROR: See http://www.mcs.anl.gov/petsc/documentation/faq.html
for trouble shooting.
> [3]PETSC ERROR: Petsc Development GIT revision: v3.7.2-630-g96e0c40  GIT
Date: 2016-06-22 10:03:02 -0500
> [3]PETSC ERROR: ./ex56 on a arch-macosx-gnu-g named MarksMac-3.local by
markadams Thu Jun 23 06:53:27 2016
> [3]PETSC ERROR: Configure options COPTFLAGS="-g -O0" CXXOPTFLAGS="-g -O0"
FOPTFLAGS="-g -O0" --download-hypre=1 --download-parmetis=1
--download-metis=1 --download-ml=1 --download-p4est=1 --download-exodus=1
--download-triangle=1 --with-hdf5-dir=/Users/markadams/Codes/hdf5
--with-x=0 --with-debugging=1 PETSC_ARCH=arch-macosx-gnu-g --download-chaco
> [3]PETSC ERROR: #1 VecScatterCreate_PtoS() line 2348 in
/Users/markadams/Codes/petsc/src/vec/vec/utils/vpscat.c
> [3]PETSC ERROR: #2 VecScatterCreate() line 1552 in
/Users/markadams/Codes/petsc/src/vec/vec/utils/vscat.c
> [3]PETSC ERROR: #3 PCSetUp_ASM() line 279 in
/Users/markadams/Codes/petsc/src/ksp/pc/impls/asm/asm.c

>
>    Barry
>
>
>
>
>
>
> > On Jun 22, 2016, at 5:20 PM, Mark Adams <mfadams at lbl.gov> wrote:
> >
> >
> >
> > On Wed, Jun 22, 2016 at 8:06 PM, Barry Smith <bsmith at mcs.anl.gov> wrote:
> >
> >    I suggest focusing on asm.
> >
> > OK, I will switch gasm to asm, this does not work anyway.
> >
> > Having blocks that span multiple processes seems like over kill for a
> smoother ?
> >
> > No, because it is a pain to have the math convolved with the parallel
> decompositions strategy (ie, I can't tell an application how to partition
> their problem). If an aggregate spans processor boundaries, which is fine
> and needed, and let's say we have a pretty uniform problem, then if the
> block gets split up, H is small in part of the domain and convergence could
> suffer along processor boundaries.  And having the math change as the
> parallel decomposition changes is annoying.
> >
> > (Major league overkill) in fact doesn't one want multiple blocks per
> process, ie. pretty small blocks.
> >
> > No, it is just doing what would be done in serial.  If the cost of
> moving the data across the processor is a problem then that is a tradeoff
> to consider.
> >
> > And I think you are misunderstanding me.  There are lots of blocks per
> process (the aggregates are say 3^D in size).  And many of the
> aggregates/blocks along the processor boundary will be split between
> processors, resulting is mall blocks and weak ASM PC on processor
> boundaries.
> >
> > I can understand ASM not being general and not letting blocks span
> processor boundaries, but I don't think the extra matrix communication
> costs are a big deal (done just once) and the vector communication costs
> are not bad, it probably does not include (too many) new processors to
> communicate with.
> >
> >
> >    Barry
> >
> > > On Jun 22, 2016, at 7:51 AM, Mark Adams <mfadams at lbl.gov> wrote:
> > >
> > > I'm trying to get block smoothers to work for gamg.  We (Garth) tried
> this and got this error:
> > >
> > >
> > >  - Another option is use '-pc_gamg_use_agg_gasm true' and use
> '-mg_levels_pc_type gasm'.
> > >
> > >
> > > Running in parallel, I get
> > >
> > >      ** Max-trans not allowed because matrix is distributed
> > >  ----
> > >
> > > First, what is the difference between asm and gasm?
> > >
> > > Second, I need to fix this to get block smoothers. This used to work.
> Did we lose the capability to have blocks that span processor subdomains?
> > >
> > > gamg only aggregates across processor subdomains within one layer, so
> maybe I could use one layer of overlap in some way?
> > >
> > > Thanks,
> > > Mark
> > >
> >
> >
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20160623/cb9343a5/attachment.html>