both are slightly different questions. But I understand now that the
whole thing boils down to obtaining the nullspaces.
Thank you,
Nidish
On 8/17/20 1:33 PM, zakaryah wrote:
Hi Nidish,
I may not fully understand your problem, but it sounds like you could
benefit from continuation methods. Have y
Thankfully for this step, the matrix is not dense. But thank you.
Nidish
On 8/17/20 8:52 AM, Jose E. Roman wrote:
El 17 ago 2020, a las 14:27, Barry Smith escribió:
Nidish,
Your matrix is dense, correct? MUMPS is for sparse matrices.
Then I guess you could use Scalapack
teAIJ(size=M.T.shape, csr =(M.indptr,
M.indices, M.data))
i.e., I need to use the shape of the transpose of the matrix.
Does this have something to do with the way the csr format is
interpreted in the petsc4py wrappings or in PETSc itself?
--
Nidish
Oh damn. Alright, I'll keep trying out the different options.
Thank you,
Nidish
On 8/16/20 8:05 PM, Barry Smith wrote:
SVD is enormously expensive, needs to be done on a full dense matrix
so completely impractical. You need the best tuned iterative method,
Jose is the by far the
ably). I'm not sure what factorization or
solver I should use in this case.
TLDR: I have K.v=lam*M.v where BOTH K and M are singular. The nullspace
of M is a subset of the nullspace of K.
Thank you,
Nidish
On 8/16/20 1:50 AM, Jose E. Roman wrote:
Nothing special is required for solv
t, but I'm wondering if you could provide me insights into
whether this will.
Thanks,
Nidish
On 8/16/20 2:50 PM, Barry Smith wrote:
If you know part of your null space explicitly (for example the
rigid body modes) I would recommend you always use that information
explicitly since
Could you have me the suggested solver
configurations to get this type of eigenvectors in slepc?
Nidish
On Aug 16, 2020, 00:17, at 00:17, Jed Brown wrote:
>It's possible to use this or a similar algorithm in SLEPc, but keep in
>mind that it's more expensive to compute these eigenv
I just use the standard eigs function
(https://www.mathworks.com/help/matlab/ref/eigs.html) as a black box. I think
it uses a lanczos type method under the hood.
Nidish
On Aug 15, 2020, 21:42, at 21:42, Barry Smith wrote:
>
>Exactly what algorithm are you using in Matlab to get
itive semi-definite (contains a few "rigid body
modes") and M is strictly positive definite.
I'd appreciate any assistance you may provide with this.
Thank you,
Nidish
o much.
Is there any example code for such a case with a singular matrix where
these procedures are carried out? Or could you provide
references/guidances for approaching the problem?
Thank you,
Nidish
Thank you,
Nidish
On 8/13/20 11:51 PM, Jed Brown wrote:
Nidish writes:
Does that mean one can choose to develop code using the petc4py wrappers
without having to sacrifice any performance?
Roughly, yes. Though assembly of residuals and Jacobians is still your
business and its performance
Does that mean one can choose to develop code using the petc4py wrappers
without having to sacrifice any performance?
Apologies if this question is too basic, I'm just trying to understand
what I would be sacrificing if I chose to completely write my
application on Python.
Nidish
On
r for code
development.
Thank you,
Nidish
Oh that makes sense - I just need to use the first element's address.
Thank you!
On 8/10/20 6:00 PM, Jed Brown wrote:
Nidish writes:
Urgh I must've been blind - flycheck was throwing warnings only if I
didn't cast elstiff[][] into const PetscScalar* .
You don't have
Urgh I must've been blind - flycheck was throwing warnings only if I
didn't cast elstiff[][] into const PetscScalar* .
Thanks for the responses!
Nidish
On 8/10/20 5:35 PM, Jed Brown wrote:
Nidish writes:
Ah I get it now, MatSetBlocked has to be set node-wise. I tried this an
Ah I get it, thanks!
On 8/10/20 5:40 PM, Matthew Knepley wrote:
On Mon, Aug 10, 2020 at 6:26 PM Nidish <mailto:n...@rice.edu>> wrote:
Ah I get it now, MatSetBlocked has to be set node-wise. I tried
this and
it works, thank you.
The other question I had was wh
typecast there
so my flycheck doesn't keep throwing me warnings on emacs ;)
Thank You,
Nidish
On 8/10/20 5:16 PM, Jed Brown wrote:
Nidish writes:
It's a 1D model with displacements and rotations as DoFs at each node.
I couldn't find much in the manual on MatSetBlockSize - co
a stencil width of 1 there), the matrix object should
have the nonzero elements preallocated. Here's the call to DMDACreate1d:
DMDACreate1d(PETSC_COMM_WORLD, DM_BOUNDARY_NONE, N, 2, 1, NULL, &mshdm);
Thank you,
Nidish
On 8/10/20 4:55 PM, Jed Brown wrote:
It looks like you haven't
works, while
MatSetValuesBlocked(jac, 4, (const PetscInt*)idx, 4, (const
PetscInt*)idx,(const PetscScalar*)elstiff, ADD_VALUES);
does NOT work.
Also from the documentation I could not really understand what the
difference was between these two.
Thank you,
Nidish
static char help[] = &quo
Update: It was a type error (idx and idy below were declared as
PetscScalar* instead of PetscInt*), I'm sorry.
Guess I'm being punished for not paying attention to types.
Nidish
On 8/9/20 10:05 PM, Nidish wrote:
Update: I had made some mistakes in the assembly in the previou
Update: I had made some mistakes in the assembly in the previous
version. They're fixed now (see below).
On 8/9/20 9:13 PM, Nidish wrote:
I'm trying to write a simple 1D finite element test case (linear
elasticity). It was given in the manual that using an appropriate DM
a
I'm trying to write a simple 1D finite element test case (linear
elasticity). It was given in the manual that using an appropriate DM and
calling MatSetValuesBlocked(). I'm however unable to do this for cases
where I want to take runtime inputs for physical constants in the matrices.
One thing
Alright, thank you.
Nidish
On 8/9/20 1:43 PM, Matthew Knepley wrote:
On Sun, Aug 9, 2020 at 2:37 PM Nidish <mailto:n...@rice.edu>> wrote:
On 8/9/20 1:45 AM, Barry Smith wrote:
On Aug 9, 2020, at 1:14 AM, Nidish <mailto:n...@rice.edu>
wrote:
Hello,
My ques
On 8/9/20 1:45 AM, Barry Smith wrote:
On Aug 9, 2020, at 1:14 AM, Nidish wrote:
Hello,
My question is related to solving a generic nonlinear algebraic system in
parallel. I'm using the word generic to point to the fact that the system may
or may not be from a finite element/mesh
with this email.
TL;DR Version: What modifications must be made in
"$PETSC_DIR/src/snes/tutorials/ex1.c" to make it work in Parallel? Is it
possible to make minimal modifications to make it run in parallel too?
Thank You,
Nidish
static char help[] = "Nonlinear Solution
On 8/7/20 12:55 PM, Barry Smith wrote:
On Aug 7, 2020, at 12:26 PM, Nidish <mailto:n...@rice.edu>> wrote:
On 8/7/20 8:52 AM, Barry Smith wrote:
On Aug 7, 2020, at 1:25 AM, Nidish <mailto:n...@rice.edu>> wrote:
Indeed - I was just using the default solver (GMRES with
On 8/7/20 8:52 AM, Barry Smith wrote:
On Aug 7, 2020, at 1:25 AM, Nidish <mailto:n...@rice.edu>> wrote:
Indeed - I was just using the default solver (GMRES with ILU).
Using just standard LU (direct solve with "-pc_type lu -ksp_type
preonly"), I find elemental to be extr
$> mpirun -n 4 ./ksps -N 1000 -mat_type mpiaij -pc_type lu -ksp_type
preonly
$> mpirun -n 4 ./ksps -N 1000 -mat_type elemental -pc_type lu
-ksp_type preonly
elemental seems to perform poorly in both cases.
Nidish
On 8/7/20 12:50 AM, Barry Smith wrote:
What is the output of -ksp_
lemental. I have not been able to make the elemental version finish up
for 2e4 so far (my patience runs out faster).
What's going on here? I thought elemental was supposed to be superior for dense
matrices.
I can share the code if that's appropriate for this forum (sorry, I'm new h
ly shoots up.
2. Are there any specific guidelines on when I can expect elemental to
perform better in parallel than in serial?
Of course, I'm interesting in any other details that may be important in
this regard.
Thank you,
Nidish
30 matches
Mail list logo
