Alejandro:
I can reproduce this error. I'm investigating it.
I made minor update of MatGetSubMatrices_SeqSBAIJ() in v3.7 which only
removed requirement of isrow=iscol...do not know what leads to this error
yet.
Hong
Hi,
>
> After updating petsc from version 3.6.2 to 3.7, I had th
Alejandro :
Fixed
https://bitbucket.org/petsc/petsc/commits/0043d7fe17a5a8a1e4768ab560023392b3862352
It will merged to petsc-release after our nightly tests.
Your detailed report is very helpful, thanks!
Hong
Hi,
>
> After updating petsc from version 3.6.2 to 3.7, I had the following
>
is a subclass of DMPlex.
Hong
>
>
>> On Fri, Jun 24, 2016 at 3:45 PM, Matthew Knepley
>> wrote:
>> > On Fri, Jun 24, 2016 at 1:58 AM, Patrick Sanan > >
>> > wrote:
>> >>
>> >> I'm working on some groundwork to improve the PETSc
Efficient and scalable MatGetSubmatrix() and assemble submatrices into a
matrix -- used for multiphysic simulation, e.g., wash project we are doing
now.
Hong
On Sat, Jul 2, 2016 at 2:46 AM, Patrick Sanan
wrote:
> Maybe a general class of opportunities for PETSc could be wrapped up
>
KSPGetPC(ksp,&pc);
ierr = PCFactorSetMatSolverPackage(pc,MATSOLVERPETSC);
ierr = PCFactorSetUpMatSolverPackage(pc); /* call MatGetFactor() to
create F */
ierr = PCFactorGetMatrix(pc,&F);
ierr = KSPSetUp(ksp);
ierr = MatSolve(F,b,x);
Hong
> [0]PETSC ERROR: See http:
Pierre:
See petsc/src/ksp/ksp/examples/tests/ex27.c on how to use MatMatSolve().
Hong
On Wed, Aug 17, 2016 at 9:34 AM, Hong wrote:
> Pierre:
>
>>
>> [0]PETSC ERROR: Object is in wrong state
>> [0]PETSC ERROR: Unfactored matrix
>>
>
> indicates that your i
Stefano:
The bug is fixed on branch hzhang/fix-ptap-seqaij
https://bitbucket.org/petsc/petsc/commits/4bcca364924c81402b2f7ad375877c5546b7966b
Satish,
Can you help merge this branch to next for testing?
Hong
I have found a bug in MatPtAPNumeric_SeqAIJ_SeqAIJ_SparseAxpy.
> Attached is a
Dear all,
I'm the older Hong Zhang (张红, hzh...@mcs.anl.gov). A young man named Hong
Zhang (张宏?hongzh...@mcs.anl.gov) joined our team one-two year ago. He is an
expert on time-stepping methods :-)
Hong
On Wed, Oct 5, 2016 at 10:27 AM, Matthew Knepley wrote:
> On Wed, Oct 5, 2016 at
Sherry,
When I applied the graduate school at MSU many years ago, there were 6
applicants from China named Hong Zhang, two eventually came, me and a
man. "This
is a deficiency of English", indeed :-(
Don't worry, I can easily distinguish emails from the context. I've
for
i.e., MatGetSubMatrices_MPIAIJ_SingleIS() is called when ismax=1 for each
process.
General case is handled by MatGetSubMatrices_MPIAIJ_Local(C,...)
I've optimized this routine and will merge it to master.
Hong
It comes from here:
>
> https://bitbucket.org/petsc/petsc/commits/c10200c1442b553b
Matt,
By default, the flag C->submat_singleis = false.
In PCSetUp_ASM(), we set it as 'true' to
use MatGetSubMatrices_MPIAIJ_SingleIS().
Can you check the value of this flag in your case?
Hong
>
> It comes from here:
>>
>> https://bitbucket.org/petsc
Matt,
This bug is fixed and merged to master. Please give it a try and let me
know if there is a problem.
Hong
On Mon, Jan 23, 2017 at 7:04 PM, Zhang, Hong wrote:
> OK, I removed flag MAT_SUBMAT_SINGLEIS
> https://bitbucket.org/petsc/petsc/commits/65b5f10fe03da0cc383d7dd0f31fad
> f
ransposeMatMult() tends to use much larger ctable than MatMatMult.
Is your matrix symmetric? If so, MatMatMult() will use smaller ctable and
require less communication.
Hong
&Gmat2);CHKERRQ(ierr);
+} else {
+ ierr = MatTransposeMatMult(Gmat1, Gmat1, MAT_INITIAL_MATRIX,
PETSC_DEFAULT, &Gmat2);CHKERRQ(ierr);
+}
In this way, MatTransposeMatMult() can be replace with MatMatMult() when
Gmat1 is symmetric.
Hong
>
>>>>
>> MatTranspos
_MATRIX,
>> PETSC_DEFAULT, &Gmat2);CHKERRQ(ierr);
>> +}
>>
>>
> That would be great thanks. When you do that I can change ex56 to use
> this, test it, and update the test.
>
Where is Gmat1 created, in user's code (e.g., ex56) or inside PET
ierr = PetscCalloc1(B->cmap->N,&ab_dense);
which is global column size P->N.
I'll check if we can avoid this non-scalable approach.
Meanwhile, can you try option '-matrap false' to use P^T*A*P?
Hong
We are getting this error on Edison. This problem has run before. Thi
Mark:
Try '-matmatmult_via scalable' first. If this works, should we set it as
default?
Hong
>
>
> On Thu, Feb 2, 2017 at 10:53 AM, Satish Balay wrote:
>
>> > [3404]PETSC ERROR: Memory allocated 0 Memory used by process 2537885696
>> > [3404]PETSC ERROR:
inear solvers in PETSc are usually used by the nonlinear
>> solvers and ODE integrators that have recovery methods for failure in a
>> linear solve we moved away from the "crash and burn" on failed linear
>> solver approach; since that makes the recovery more difficult.
>>
rface to Hypre's PtAP. Comparing the two,
Petsc MatPtAP() is approx 3x faster than Hypre's.
I'm writing a scalable MatPtAP() now.
Hong
On Thu, Feb 2, 2017 at 2:54 PM, Stefano Zampini
wrote:
>
>
> Il 02 Feb 2017 23:43, "Mark Adams" ha scritto:
>
>
I'll check it tomorrow and fix it.
Hong
On Thu, Feb 9, 2017 at 4:27 PM, Barry Smith wrote:
>
>When called with a NULL set of numerical values
> MatCreateSeqAIJWithArrays() should turn off the function pointers for
> things that don't make sense for matrices without
Change below fixed the problem. I'll turn off additional ops later.
Hong
diff --git a/src/mat/impls/aij/seq/aij.c b/src/mat/impls/aij/seq/aij.c
index 1e0b0b8..5cee782 100644
--- a/src/mat/impls/aij/seq/aij.c
+++ b/src/mat/impls/aij/seq/aij.c
@@ -4169,6 +4169,11 @@ PetscErro
Pierre :
The bug is fixed and pushed to petsc-maint
https://bitbucket.org/petsc/petsc/branch/maint
Thanks for reporting it.
Hong
Oh my, this example also segfaults (at least on my Mac) when running:
> mpirun -np 1 ./ex1 -pc_type gamg -mat_view
>
> [0]PETSC ERROR: [0] MatView_SeqAIJ_A
AP 4.6129e+01
hypre1.9389e+02
I'm merging this work to next, then to master soon.
Hong
On Wed, Feb 8, 2017 at 7:25 PM, Mark Adams wrote:
> Thanks Hong, that sounds great.
>
> I am weary of silent optimizations like you suggest but 2x is big! and
>
This work is merged to master
https://bitbucket.org/petsc/petsc/commits/57c33550d09512dabc9444743ef186e1d13ee746
Hong
On Tue, Feb 28, 2017 at 8:04 AM, Mark Adams wrote:
> Yes, this looks great. Good to get some data on this. Hard problem.
>
> On Mon, Feb 27, 2017 at 3:31 PM, Barry Smi
I fixed typos in maint branch:
https://bitbucket.org/petsc/petsc/commits/d4ed72bbe47df8d3eb83fd589394594abed431db
I'll check the error produced by './ex12 -pc_type gasm -ksp_view'.
Hong
On Thu, Mar 16, 2017 at 3:31 AM, Pierre Jolivet
wrote:
> Hello,
> Using master:
>
Ce Qin:
This is fixed in branch hzhang/matmatops_mpiaij4np1.
I consider it as a new feature - support mpiaij for np=1, not a bugfix.
Once it passes our regression tests, I'll merge it to petsc master branch.
You may test it. Let us know you see any problem.
Thanks for reporting it.
Hong
This is in master now.
Hong
On Wed, Mar 22, 2017 at 9:43 PM, Ce Qin wrote:
> Dear Hong,
>
> I have tested the new feature, it works now.
> Thanks.
>
> Best,
> Ce Qin
>
>
> 2017-03-17 23:58 GMT+08:00 Hong :
>
>> Ce Qin:
>> This is fixed in branch hzh
bi_1() (not 2*m).
I'll fix them in petsc-maint.
Thanks for reporting it!
Hong
Fixed
https://bitbucket.org/petsc/petsc/commits/79cd6d079a316048b19f539cf806a8cfa87064d9
Hong
On Mon, Mar 27, 2017 at 9:39 AM, Hong wrote:
> Pierre:
>
>> Hello,
>> PCApply_PBJacobi FLOPs complexity is bs*(2*bs-1)*m.
>> How do you come up with PetscLogFlops(80.0*m) in PC
Brad:
You must call MatDestroySubMatrices(), not MatDestroyMatrices().
See petsc/src/mat/examples/tests/ex54.c
I'll add it to the help menu of MatCreateSubMatrices().
Hong
>
>
> In PyLith we call MatCreateSubMatrices() and MatDestroyMatrices() with
> potentially different numbe
Brad:
I updated the help menu
https://bitbucket.org/petsc/petsc/commits/7526cb21428e6e453a7c940d78502e96d2c8a4bd
Thanks for your report.
Hong
>
> On 03/27/2017 12:57 PM, Hong wrote:
>
>>
>>
>> Brad:
>> You must call MatDestroySubMatrices(), not MatDestroyMatrices
Stefano :
>
> I would be inclined to have MatMatTransposeMult to dispatch the same way
> MatTransposeMatMult does. Comments?
>
I agree. This is the way that MatMatMult() does.
All matmat-ops should follow this pattern.
Will you fix MatMatTransposeMult()?
Hong
Adrian,
Adding option '-mat_type aij', do you still see this problem?
Is possible to send us a simple code that produces this error?
I cannot reproduce it using petsc examples.
Hong
On Wed, May 17, 2017 at 10:50 PM, Barry Smith wrote:
>
>Adrian,
>
> There is a
Adrian,
Ignore my previous email.
I can reproduce it using
petsc/src/snes/examples/tutorials
mpiexec -n 2 ./ex19 -pc_type asm -dm_mat_type baij
I'll let you the fix.
Hong
On Thu, May 18, 2017 at 4:33 PM, Hong wrote:
> Adrian,
> Adding option '-mat_type aij', do you s
Adrian,
It is a bug in our ilu0 for seqbaij matrix
https://bitbucket.org/petsc/petsc/commits/2eb5df998544f7f4e52ff7268aa93a795e17258c
I'll add some tests to petsc, then push it to next for regression tests,
finally the fix will be merged to master.
Thanks for your report.
Hong
On Thu, M
2 because it is
> trying to cast a matrix to Mat_MPIAIJ instead of Mat_MPIBAIJ. Any chance
> you could let me know how to dispatch the call (*(*C)->ops->matmultnumeric)
> to the proper implementation (depending on the type of A)?
>
You define
(*C)->ops->matmultnumeric = MatMatMultNumeric_MPIBAIJ_MPIDense;
in your MatMatMultSymbolic_MPIBAIJ_MPIDense
Hong
IBAIJ_MPIDense.
You may follow a debugging process
using petsc/src/mat/examples/tests/ex109.c
Are you working on a branch of petsc? If so, I may take a look and see what
is the problem.
Hong
>
>
> 2) I'm having trouble when scall == MAT_REUSE_MATRIX. Here,
>>>
>
What is your calling procedure for using MatMatMultNumeric_MPIBAIJ_MPIDense?
Hong
On Tue, Jun 6, 2017 at 1:52 PM, Pierre Jolivet
wrote:
> On Tue, 6 Jun 2017 13:44:06 -0500, Hong wrote:
>
>> Pierre :
>>
>> Yes, of course I defined (
Pierre :
I'm not aware the "dummy function" MatMatMultNumeric_MPIDense().
Good to know you find a solution.
Hong
> Barry,
> There is no such error when you do:
> --- a/src/mat/examples/tests/ex109.c
> +++ b/src/mat/examples/tests/ex109.c
> @@ -57,7 +57,8 @@
test that shows this error?
Hong
On Mon, Jun 19, 2017 at 5:38 AM, Lisandro Dalcin wrote:
> Dear Hong, any chance this is related to your work around commit
> 8b3fa1f7362a032c26473e2c80cb80f09a5a21c7
>
> This test is working just fine with PETSc 3.5, 3.6, and 3.7, but fails
> with pe
Lisandro :
I fixed bug in branch hzhang/fix-ao2petscis:
https://bitbucket.org/petsc/petsc/commits/260dbea8cadd3eeca7e59945a235ff92ceae1333
I tested "-iga_dof 3" as well.
Let me know if you still encounter problems, or have any comments on this
fix.
Hong
On 20 June 2017 at 01:46, H
I add support for MatRARt_MPIAIJ
https://bitbucket.org/petsc/petsc/commits/a3c14138ce0daf4ee55c7c10f1b4631a8ed2f13e
It is in branch hzhang/mpirart.
Let me know your comments.
Hong
On Wed, Jun 21, 2017 at 12:53 PM, Barry Smith wrote:
>
> > On Jun 21, 2017, at 8:00 AM, Franck Houssen
t,scall,fill,C);CHKERRQ(ierr);
ierr = MatDestroy(&Rt);CHKERRQ(ierr);
}
This does NOT work for most matrix formats because we do not have fallbacks
for MatTranspose() and MatMatMult().
Hong
>
> >
> > On Wed, Jun 21, 2017 at 12:53 PM, Barry Smith
> wrote:
> >
> >&g
r in either
MatTranspose() or MatMatMatMult() depends on input matrix format?
If so, we need add this type of 'default' to all mat operations --
currently, all routines do
if (!mat->ops-> )
SETERRQ1(PetscObjectComm((PetscObject)mat),PETSC_ERR_SUP,"Mat type
%s",((PetscObject)mat)->type_name);
Hong
port.
I'll check the manual on MatRARt/PtAP to make it clear to users.
Hong
On Fri, Jun 23, 2017 at 7:13 AM, Stefano Zampini
wrote:
>
> > On Jun 23, 2017, at 10:10 AM, Franck Houssen
> wrote:
> >
> > By the way, I noticed this is exactly the same for
> MatRARt/PtAPNu
I added following to the help menu of MatRARt():
"This routine is currently only implemented for pairs of SeqAIJ matrices
and classes
which inherit from SeqAIJ. For parallel matrices, it is recommended to
use MatPtAP()."
On Fri, Jun 23, 2017 at 9:59 AM, Hong wrote:
> Due to
Franck:
See
https://bitbucket.org/petsc/petsc/commits/f3dcd100076caffe0af33a424f0d06cb26a76989
Hong
>
> --
>
> *De: *"Hong"
> *À: *"Stefano Zampini"
> *Cc: *"Franck Houssen" , "petsc-dev" <
> petsc-dev@
of R, which could
be very expensive.
We recommend using MatPtAP().
Thanks for your request and suggestion!
Hong
On Sun, Jun 25, 2017 at 7:36 AM, Franck Houssen
wrote:
> Great !
>
> --
>
> *De: *"Hong"
> *À: *"Franck Houssen"
hich
you can add option '-mg_coarse_pc_redundant_number 2' to have two
subcommunicators at coarse-grid level.
Can you use pcredundant for gamg?
Hong
On Tue, Jun 27, 2017 at 9:46 AM, Mark Adams wrote:
>
>
> On Tue, Jun 27, 2017 at 8:35 AM, Matthew Knepley
> wrote:
>
>&
MatCreateSubMatrices_xxx() calloc an array of size n+1, n=ismax and
submat[n]=null as default.
For MatCreateSubMatrices_MPIAIJ(), I calloc ismax+nstages and use
submat[n] ... to store some datastructures for reuse.
When submat[n] != null, these data structures are destroyed.
Hong
On Wed, Jul
Stefano :
Can you give me a detailed instruction on how to reproduce this error?
Hong
Hong,
>
> I think what you wrote is conceptually wrong, being in the interface part
> of the library.
> The extra allocation is not done consistently throughout all the specific
> imple
; The extra allocation is not done consistently throughout all the
>>> specific implementations that MatCreateSubMatrices can use.
>>>
>>
How about adding 'destroysubmatrices' into _MatOps table, then move the
extra stuff into MatDestroySubmatrices_xxx()?
Hong
&
This example works well on maint-branch.
There is a bug in master branch.
I'll fix it after I'm back from the vacation (Thursday).
Hong
On Sat, Jul 22, 2017 at 1:13 PM, Jose E. Roman wrote:
> Attached is an example that shows that MatCreateRedundantMatrix() fails if
&g
Jose,
Bug is fixed in branch hzhang/bugfix-RedundantMat/master
https://bitbucket.org/petsc/petsc/commits/7bbdc51d16de20c2a4daada3a4bf77c9346d6e84
Let me know if you have any comments.
Hong
On Mon, Jul 24, 2017 at 9:02 AM, Hong wrote:
> This example works well on maint-branch.
>
> T
mumps stores LDLt using its private internal data structure, which is not
exposed to user and can be used only via mumps API.
Either ask mumps developer about it or use a debugger to step into mumps
source code.
Hong
On Thu, Aug 24, 2017 at 9:46 AM, Franck Houssen
wrote:
> How to get L (a
Jed :
> ~jedbrown/ceres_solver_iteration_001_A.petsc on mcs.anl.gov.
>
I got this file. How many matrices in this file? How to load them?
It seems the first matrix is rectangular.
Hong
>
> The issue is that MatTransposeMatMult is implemented using sparse outer
> products
Jed,
I reproduced what you observed.
30x time is spent on numerical C = At*B.
We compute A^T*B using outer product C=(A^T)[:,i]*B[i,:].
For your large matrix A: rows=574902, cols=184446,
outer product takes long long time to insert values to C.
We may use At=A^T and C=At*B instead.
Hong
On Thu
Jed:
> Hong writes:
>
> > Jed,
> > I reproduced what you observed.
> > 30x time is spent on numerical C = At*B.
> > We compute A^T*B using outer product C=(A^T)[:,i]*B[i,:].
> > For your large matrix A: rows=574902, cols=184446,
> > outer product
Jed :
New implementation for seqaij is in branch hzhang/opt-mattransposematmult.
petsc/src/mat/examples/tests/ex209.c is a test.
Please give it a try and let me know if you have any comment.
I'll remove previous implementation and optimize MatTransposeMatMult() for
mpiaij matrices.
Hong
Fande,
'-pc_factor_shift_amount 1' is huge.
Try use petsc default shift amount '1.e-14' by
'-pc_factor_shift_type nonzero' only.
Hong
On Tue, Oct 10, 2017 at 9:40 AM, Matthew Knepley wrote:
> On Tue, Oct 10, 2017 at 10:36 AM, Kong, Fande wrote:
>
>&g
This should be cleaned by
https://bitbucket.org/petsc/petsc/commits/a0d1c92d1d6734b184005d635c14bf9895961849
Will merge it to master once it passes nightly tests.
Hong
On Tue, Oct 24, 2017 at 10:41 PM, Barry Smith wrote:
>
> ftp://ftp.mcs.anl.gov/pub/petsc/nightlylogs/archive/
> 2
Stefano:
Sorry, I do not understand your question. Where comes matrix B?
Can you give me an example of this error?
Hong
Hong,
>
> can you explain what’s the rationale behind calling explicitly the Layout
> and header destroy for the submit case in the loop here
> https://bitbucket.org
trix out of submatrices array, and
use/destroy it as a regular single matrix?
Hong
On Sat, Jan 13, 2018 at 1:09 PM, Stefano Zampini
wrote:
> Cc'ing petsc-dev
> -- Messaggio inoltrato --
> Da: "Stefano Zampini"
> Data: 12 Gen 2018 9:32 PM
> Oggetto: Re: M
these type of usage.
Hong
>
> On Jan 14, 2018, at 7:16 AM, Hong wrote:
>
> Stefano,
> MatCreateSubMatrices() must be destroyed by MatDestroySubMatrices()
> because all submatrices in the array share common internal data structure
> for reuse.
>
> + ierr = PetscObj
LayoutDestroy(&C->rmap);CHKERRQ(ierr);
+ierr = PetscLayoutDestroy(&C->cmap);CHKERRQ(ierr);
+ierr = PetscHeaderDestroy(&C);CHKERRQ(ierr);
+ }
} else {
should fix the problem. Let me know your thought.
Hong
On Sun, Jan 14, 2018 at 9:09 PM, Hong wrote:
> Stefano:
Barry :
>
> I've fixed this in the branch.
>
Thanks! It has a clean build now.
Hong
>
>
> > On Feb 23, 2018, at 9:00 AM, PETSc checkBuilds <
> petsc-checkbui...@mcs.anl.gov> wrote:
> >
> >
> >
> > Dear PETSc developer,
> >
>
+ requires one to pass in as submat a Mat (integer) array of size at
least n+1.
If you agree, I'll push it to the petsc released version.
Hong
Tim:
It is in the release branch
https://bitbucket.org/petsc/petsc/commits/f79229c73db120b69d05e6d49c19f429e09e633f
Thanks for reporting it.
Hong
Hong,
> we are fine with that. Thanks!
> Volker
>
> 2018-04-18 17:32 GMT+02:00 Hong :
> > Tim :
> >>
> >> Hi all,
&
ve
>
Can you send us a stand-alone code for this type of application.
We will investigate how to support it.
Hong
I cannot change this calling sequence, because I’d like to nest 3) and 4)
> inside a loop.
> However, I end up with this error during 4):
> [0]PETSC ERROR: Nonconforming ob
Pierre:
Thanks for creating this use case. I'll check it and get back to you
sometime next week.
Hong
> Hong,
> Thanks for your reply.
> It’s not so minimalist, but at least it is based on one of your example.
> diff --git a/src/ksp/ksp/examples/tutorials/ex42.c b/src
helpful, thanks!
Hong
On Fri, May 4, 2018 at 10:15 AM, Hong wrote:
> Pierre:
> Thanks for creating this use case. I'll check it and get back to you
> sometime next week.
> Hong
>
> Hong,
>> Thanks for your reply.
>> It’s not so minimalist, but at least it is ba
Pierre:
> Hong,
> Thank you for the patch.
> Our application is not crashing anymore.
> Hopefully this will be merged quickly in master.
>
Merged to master.
Hong
>
>
> On 7 May 2018, at 10:07 PM, Hong wrote:
>
> Pierre:
> I added this support
> https
Stefano :
>
>
> why didn't you merge it to master and main yet? is the wrong patch?
>
Just merged.
Sorry, I thought it was merged.
Thanks once again,
Hong
>
> 2015-02-12 20:26 GMT+03:00 h...@aspiritech.org :
>
>> Barry:
>>
>>>
>>>
branch from master and test
your patch.
Hong
Sascha :
Thanks for your contribution!
I have patched both to petsc-master in the branch hzhang/ex-update
https://bitbucket.org/petsc/petsc/commits/0c500c50befac1baa76f423dafbb038fecb748c3
Once it passes our regression tests, I'll merge it to the petsc master.
Hong
Hello,
>
> I prepa
Sascha:
I'll replace *.txt file with a C-array to simply ex56.c.
Please ignore my previous request.
Hong
Hong,
> I do not have an immediate explanation for this. In particular, I wonder
> why
> there is no line break after 'sp=/' in this line:
> > file = 0x0, sp=
The line 1376 has been in the file for long time -- at least I did not add
it:
if (Petsc_Elemental_keyval == MPI_KEYVAL_INVALID) {
ierr =
MPI_Keyval_create(MPI_NULL_COPY_FN,MPI_NULL_DELETE_FN,&Petsc_Elemental_keyval,(void*)0);
}
Would recent change in other places cause this problem?
Hong
Patrick:
We have replaced this segment of code (reading a data file)
with initialized C arrays in petsc master branch. Are you using
petsc-release version?
Hong
I tested this out and the strange behavior is very likely because the C
> string “sp" is not null-terminated, so there is u
Stefano:
Your patch is pushed to maint branch. Thanks!
Hong
On Thu, Apr 23, 2015 at 10:40 AM, h...@aspiritech.org
wrote:
> Stefano:
>
>> It seems to me that the current code for PetscSubcommSetTypeGeneral
>> assumes that all colors will be provided within a range
>> [0,
Stefano:
> I would like to merge
>
>
> https://bitbucket.org/petsc/petsc/branch/stefano_zampini/matmumps-schursupport
>
> into next. Any objection? I will open a PR in that case and we can discuss.
>
This is fine with me.
Hong
. The frustration we encountered is that the code seems
working well, until I checked the solutions at each time steps and found
them never change values. Then it took us hours to find out it was caused
by shift.
At least for LU, when zero-pivot occurs, we should throw an error.
Hong
>
&
low the program to keep running.
>
> Obviously there are some issues in the new model that have to be worked
> out.
>
Running 1,003,415 time steps with 'nan', and still continue ... ?
Hong
>
>
> > On Jun 10, 2015, at 8:49 PM, Hong wrote:
> >
> >
Here is an updated (python3 and a few other new features) version of Jed's
script.
https://github.com/caidao22/petscplot
Hong (Mr.)
On Aug 8, 2018, at 10:51 PM, Jed Brown
mailto:j...@jedbrown.org>> wrote:
Fande Kong mailto:fdkong...@gmail.com>> writes:
Hi all,
If we
eoint/week2.pdf .
Other types of symplectic methods such as symplectic Runge-Kutta use different
tableaus and cannot be implemented in the same framework as the basic one. So
when naming this particular type of symplectic methods, we think it is better
to be specific than general.
Thanks,
Hong
On S
this happen before, thus want to check.
Hong
From: Smith, Barry F.
Sent: Wednesday, October 10, 2018 6:41:20 PM
To: Zhang, Junchao
Cc: petsc-dev
Subject: Re: [petsc-dev] MUMPS silent errors
I looked at the code and it is handled in the PETSc way. The user should
Junchao:
Hong,
The user's example code reads a matrix, calls KSPSolve, then over. From his
log_view file, I saw long MatLUFactorNum time and short MatSolve time. Now I
know that is because MatSolve was skipped. Thanks.
This is intended.
Hong
From:
.
Hong
t terms in the RHS (function) but
> not in the G Jacobian, then this is like treating these terms with
> forward Euler, is that right?
Yes. If you only want to use an explicit method, you can just provide the RHS
function without the RHS Jacobian.
Hong
> I that right?
>
> On Wed, A
The problem is in line 5216
\noindent\textbf{\trl{TSAdapt} basic.}
If you change the word TSAdapt (e.g. add a random character), the error would
disappear. Probably it is conflicting with something else.
Hong
> On Sep 12, 2016, at 10:01 AM, Barry Smith wrote:
>
>
> Ru
> On Nov 15, 2016, at 10:56 AM, Emil Constantinescu
> wrote:
>
>
>
> On 11/15/16 7:35 AM, Stefano Zampini wrote:
>> Emil,
>>
>> I have modifed the example code to also include the convection matrix in the
>> lhs.
>>
>> Plain arkimex with -ts_arkimex_fully_implicit false and M*udot - K*u in
tscTableCreate(A->cmap->n,aN,&ta);
Mark: can you make this change and let me know if it fix the problem?
I'll check similar problems in other parallel matrix product routines.
Hong
From: Barry Smith [bsm...@mcs.anl.gov]
Sent: Wednesday, Jan
master still uses
PetscTableCreate(aN,aN,&ta);
I guess Satish latest change in PetscTable uses more memory than previous one,
thus input aN
starts causing problem.
I believe replacing 1st aN with A->cmap->n will fix the problem.
Please give it a t
/petsc/petsc/src/ed945043d5c70928faa72750bc5cf0ad23ecc966/src/docs/tex/manual/part2.tex?at=hongzh%2Fadd-eclipse-instructions&fileviewer=file-view-default#part2.tex-8145
It was only tested with Eclipse Neon(4.6.2) on Mac. Any feedback/suggestion
would be welcome.
Hong (Mr.)
> On Jan 7, 201
this paragraph. I can do it
if nobody else objects.
Hong (Mr.)
On Sun, Jan 15, 2017 at 3:39 AM, Zhang, Hong
mailto:hongzh...@anl.gov>> wrote:
I am always itching to add something in this section. Finally got the change to
do that.
Since Patrick has already added the part on how to use
gt;10k.
MatGetSubMatrices() works without this flag turns on.
I just merged the improved MatGetSubMatrices_MPIAIJ() which only calls
MPI_Allreduce once
for repeated use. Previously, it calls MPI_Allreduce twice for each call of
MatGetSubMatrices_MPIAIJ().
Hong
___
ie., after first call of MatGetSubMatrices_MPIAIJ_SingleIS, this flag is set in
saved context of submatrices
and is reset as 'false' to the original matrix C.
Do you agree with this fix? Is so, I'll push it in a branch for nightly test.
Hong
From: Barry
What if a MPI process has no submatrices, how does it save the flag?
case of 'SingleIS' is for ismax=1 in all processes. For PCASM, all process has
ismax=1, isn't it?
Hong
> On Jan 22, 2017, at 8:58 PM, Zhang, Hong wrote:
>
> The 'saved' context is in s
We allocate array submats[] with length 'ismax+1'. When a process has ismax=0,
submats[0] is still generated with a 0-dimension matrix, so 'saved context' can
be attached for reuse.
Hong
____
From: Zhang, Hong
Sent: Monday, January 23, 201
Sorry, checking the code, we actually require user to provide 0-length IS's
with ismax=1 as input
in the case of (MAT_REUSE_MATRIX && !ismax).
Hong
From: Zhang, Hong
Sent: Monday, January 23, 2017 10:54 AM
To: Barry Smith
Cc: Matthew Knepley;
on(pc->pmat,MAT_SUBMAT_SINGLEIS,PETSC_TRUE);CHKERRQ(ierr);
}
which is free of communication, then set the flag to avoid any MPI_Allreduce
call in
MatGetSubMatrices(). Without this flag, it calls general impl of
MatGetSubMatrices() which calls reduce once.
Hong
__
1 - 100 of 383 matches
Mail list logo