Yep - vector_mult, reciprocal, and pointwise_mult are how we do explicit
updating in MOOSE:
https://github.com/idaholab/moose/blob/next/framework/src/timeintegrators/ActuallyExplicitEuler.C#L182
Derek
On Mon, May 20, 2019 at 2:38 PM Yuxiang Wang wrote:
> Thank you Jed & John! That's extremely
MOOSE has had this for a long time - you can see how we do it here:
https://github.com/idaholab/moose/blob/devel/framework/src/problems/FEProblemBase.C#L4651
Basically:
1. Indicate/Mark
2. Adapt
3. Reinit and re-project
4. Repeat
Derek
On Sun, Jul 15, 2018 at 1:55 PM Vinicius C. Reis
wrot
Also: name you should name your files like:
something.e
something.e-0002
something.e-0003
Then Paraview will pick up the series automatically simply by opening
"something.e".
Derek
On Wed, May 2, 2018 at 12:20 PM John Peterson wrote:
> On Wed, May 2, 2018 at 11:56 AM, Daniel Vasconcelos <
> d
nging.
Thanks for the reply - I was just making sure I wasn't missing anything.
Derek
On Fri, Apr 27, 2018 at 7:21 PM David Knezevic
wrote:
> On Fri, Apr 27, 2018 at 6:40 PM, Derek Gaston wrote:
>
>> What is up with SolverConfiguration? It's abstract... and I can't s
What is up with SolverConfiguration? It's abstract... and I can't see a
single implementation of it anywhere in libMesh. In addition - everything
is public.
Shouldn't there be a few implementations of this (like for PETSc, etc.)?
What am I missing?
Derek
---
Ok - sounds reasonable. Just checking! :-)
Derek
On Wed, Jun 28, 2017 at 1:24 PM Harshad Sahasrabudhe
wrote:
> Wait: can you describe what you're actually trying to do?
>>
>> Why do you think you need a new mesh made up of a subset of elements?
>>
>
> I'm doing self-consistent Schrodinger-Poi
Wait: can you describe what you're actually trying to do?
Why do you think you need a new mesh made up of a subset of elements?
Derek
On Wed, Jun 28, 2017 at 12:11 PM Harshad Sahasrabudhe
wrote:
> >
> > It would be quite useful to be able to instead create a range of
> > iterators with any arb
Check out MeshTools::build_node_to_elem_map()
Derek
On Sun, Jun 25, 2017 at 4:27 PM Shayan Hoshyari
wrote:
> Hi,
>
> If it is feasible, you can create an array of pointers to elements with the
> size of your vertices, vert2el[]. Then, loop over all elements, e, and loop
> over all the vertices o
Cody: she should be able to use the "ghosted_boundaries" option in the Mesh
block to specify boundary IDs of elements that should be fully ghosted to
all processors... that should enable contact to work.
Derek
On Thu, Mar 9, 2017 at 11:16 AM Cody Permann wrote:
> Roy, before we spend a lot of t
Be aware that the version of GetPot in libMesh is fairly different from the
original project. We forked the original one many years ago and have added
many of our own features.
Derek
On Fri, Feb 17, 2017 at 11:20 AM Xikai Jiang wrote:
> Thanks, Shayan. Good to know these features, and it works
splitter also has the ability to do M->N parittioning. What this means is
that splitter itself can be run using MPI... and can generate partitions
for any number of processors. In fact, you can give it a list of n_procs
you want it partition for and it will do all of them simultaneously.
If you
d, Sep 28, 2016 at 7:19 AM David Knezevic
wrote:
> On Wed, Sep 28, 2016 at 12:13 AM, Derek Gaston wrote:
>
>> Yeah - coupling interior boundary elements is what we do with MOOSE.
>> Makes things simple.
>>
>
> Just to confirm, do you mean that you add extra "dim
There is no determinism in the PointLocator system. Everything is done
using floating point math and dynamically built structures. You're going to
need to put your own system in place for guaranteeing determinism.
Derek
On Wed, Sep 28, 2016 at 11:09 AM Harshad Sahasrabudhe
wrote:
> HI John,
>
>
Yeah - coupling interior boundary elements is what we do with MOOSE. Makes
things simple.
On Tue, Sep 27, 2016 at 9:55 PM David Knezevic
wrote:
> On Tue, Sep 27, 2016 at 9:47 PM, Roy Stogner
> wrote:
>
> >
> > On Tue, 27 Sep 2016, David Knezevic wrote:
> >
> > Quick question about the new Ghosti
On Mon, Aug 22, 2016 at 1:08 PM Mike Marchywka
wrote:
> I tried that and the compiler did not like it because of the way you
> designed the classes...
>
Compiler didn't like what? You have to write C++ that compiles. That has
nothing to do with the "class design". Pay attention to the Doxygen
Mike: you seem overly concerned with optimization even though you are just
getting started. Optimization is something better handled after you have
your application working. Try to write good solid code that makes sense
first.
Right now you are just guessing at what's going to be slow... the only
Normally you don't need to "iterate over components". Normally you would
just use the overloaded operators to do vector arithmetic.
For instance, for a PDE term that looks like "div grad u" the volume
portion of the weak form looks like "grad u * grad phi_i" (where phi_i is
your test function).
That warning is from PETSc and can be ignored (PETSc doesn't understand
that you're passing libMesh command-line options).
--keep-cout is the correct one.
Derek
On Thu, Aug 18, 2016 at 1:29 PM Jayaraman, Vegnesh
wrote:
> Sorry.
>
> I did a typo while writing the previous mail
>
> I did
> ./exa
A little late to the party - but we use all of the options you listed.
However: the one I would recommend is using ReplicatedMesh and just use
less cores per node in your cluster so that it fits. It will save you an
immense amount of development time... and most likely a lot of run time as
well.
Actually - in this case... I would just make a copy of the coarse
mesh/solution into a new EquationSystems and then refine it to get the fine
one.
Derek
On Fri, Jul 22, 2016 at 12:50 PM Roy Stogner
wrote:
>
> On Fri, 22 Jul 2016, Harriet Li wrote:
>
> > I'm trying to transfer a solution between
All packages have their strengths and weaknesses. libMesh's strength (one
of them anyway) is that you (can) control EVERYTHING.
If you're willing to give up a little bit of control and gain some code
brevity you might try out Diffsystem in libMesh.
If you're willing to give up more control and wr
cator(point) searches
> the entire mesh space, and point_value() uses the current_local_solution
> (which means the entire dof domain).
> So is that possible to access a value in a parallel vector which does not
> exist on a certain processor?
>
>
> On Jul 1, 2016, at 3:56 PM, Derek Gasto
How much memory do you have on your machine? Do you have more than 48GB?
Are you using a 32bit or 64bit operating system?
Derek
On Fri, Jul 1, 2016 at 2:34 PM 张江 wrote:
> Hi,
>
> I am trying to read a large data (11.7GB) with libmesh and use it for my
> application. The program runs well when
am limited to first order Lagrange elements
> and probably a trapezoidal integration rule.
>
> If I call renumber_nodes_and_elements after using LinearPartitioner, will
> the resulting DOFs make a block tri-diagonal matrix?
>
> Harshad
>
> On Fri, Jul 1, 2016 at 2:38 PM, Dere
Renumbering and Partitioning are two different things.
You can partition along an axis using:
https://github.com/libMesh/libmesh/blob/master/include/partitioning/linear_partitioner.h
Derek
On Fri, Jul 1, 2016 at 2:26 PM Harshad Sahasrabudhe
wrote:
> Hi,
>
> Is it possible to partition a 3D mes
We actually use it daily and have many regression tests for it in MOOSE. It
works just fine :-)
On Fri, Jul 1, 2016 at 11:02 AM Roy Stogner
wrote:
>
> On Fri, 1 Jul 2016, Harshad Sahasrabudhe wrote:
>
> > I'm trying to have 2 meshes, one which is parallelized over n processes
> and
> > another wh
Nothing can be more official than the source code! Those doxygen pages are
just made from the source code anyway...
Derek
On Wed, Jun 29, 2016 at 7:29 PM Shashank Acharya <
f2009...@goa.bits-pilani.ac.in> wrote:
> Dear all,
>
> I couldn't find the element ID listing. The page from sourceforge i
The right one is:
./configure --enable-all-static --enable-static --disable-shared
But "make test" may not work in your environment (because you probably need
to interact with the queue, etc.).
Derek
On Wed, Jun 29, 2016 at 4:53 PM 张江 wrote:
> Hi,
>
> I am trying to build libmesh on BG/Q Vest
Back of the envelope:
(Assuming you're using HEX27 elements... but the analysis won't be off by
much if you're using HEX20)
~60 first order nodes in each direction: 216,000 first order nodes
~120 second order nodes in each direction: 1,728,000 second order nodes
One first order variable: 216,000
If you don't have a bifurcation you can just write a loop around your
solver and ramp up your parameter each "step".
John: does the ContinuationSystem offer any advantages in this case?
Derek
On Mon, Jun 6, 2016 at 11:27 AM John Peterson wrote:
> On Fri, Jun 3, 2016 at 1:34 PM, Harriet Li wrot
Don't treat them as separate meshes. Get the values you want on the coarse
mesh and then uniform_refine() it in libMesh as many times as you want...
libMesh will automatically project the solution on the the finer meshes
each time you call EquationSystems::reinit(). Then start solving on that
fin
It is supported... but one thing to keep in mind is that the output format
we like to use (ExodusII) has a requirement that separate element types
must be in separate element "blocks" (in libMesh that means they must have
separate "subdomain_ids").
Note: that is NOT a requirement in libMesh. libM
, 2016 at 10:18 PM John Peterson wrote:
>
>
> On Mar 21, 2016, at 9:04 PM, Derek Gaston wrote:
>
> On Mon, Mar 21, 2016 at 9:48 PM John Peterson
> wrote:
>
>> This mistake should only even be possible if your UniquePtr is still a
>> std::auto_ptr. Are you con
On Mon, Mar 21, 2016 at 9:48 PM John Peterson wrote:
> This mistake should only even be possible if your UniquePtr is still a
> std::auto_ptr. Are you configured with --enable-unique-ptr? I don't think
> it's an issue with the libmesh interface, it's a well-known issue with
> std::auto_ptr that
On Mon, Mar 21, 2016 at 8:50 PM Cody Permann wrote:
> On Mon, Mar 21, 2016 at 6:34 PM Derek Gaston wrote:
>
>>
>> There are lots of uses for unique pointer. I don't want to start a debate
> on the list but just Google around a bit, you'll find a few. Shared
>
> enough but if you had actually typed "UniquePtr solution_vector ..." we all
> would have spotted that error sooner!
>
>
> On Mon, Mar 21, 2016 at 6:22 PM Kirk, Benjamin (JSC-EG311) <
> benjamin.k...@nasa.gov> wrote:
>
>>
>
Thanks for looking! The code was so small and simple that I thought I was
going crazy...
Derek
On Mon, Mar 21, 2016 at 8:22 PM Kirk, Benjamin (JSC-EG311) <
benjamin.k...@nasa.gov> wrote:
>
> On Mar 21, 2016, at 7:12 PM, Derek Gaston wrote:
>
> Fixed it.
>
>
> Ni
think this is an interface problem... it should never be
possible to nullify a pointer inside of a libMesh object so easily!)
Changing it to:
auto & solution_vector = output_system.solution;
Fixed it.
Good old one character bug...
Sorry for the list spam!
Derek
On Mon, Mar 21, 2016 at 7:37
Pulling my hair out here. I'm sure I'm missing something simple...
I was just creating a simple app to output partitionings... and it
segfaults. It's the simplest code ever... here:
https://github.com/friedmud/simple_libmesh_app
Here is a backtrace with a dbg build:
* thread #1: tid = 0x57c149
It's not everything... but we can also "promote" elements
programmatically. We already support mesh.all_second_order() to promote
first order mesh to second order mesh. We could do the same thing for
these more exotic types... read in first (or second) order geometry and
then promote them (by add
Cool link John, I may need something like that myself... I wonder if it has
a volume preserving meshing capability...
On Sat, Mar 5, 2016 at 9:00 AM John Peterson wrote:
>
>
> > On Mar 5, 2016, at 1:12 AM, Giorgio Bornia
> wrote:
> >
> > My starting point is a fine mesh of a real geometry that w
Actually, libMesh does compile in the "nanoflann" utility for quickly
finding approximate nearest neighbors... so maybe we're not THAT far off...
Derek
On Sun, Feb 28, 2016 at 11:06 AM Roy Stogner
wrote:
>
> On Fri, 26 Feb 2016, Wen Jiang wrote:
>
> > Could anyone tell me if libmesh supports mes
Odd. One workaround for now may be to use OpenMP instead. Try configuring
libMesh with "--enable-openmp".
Can you tell us more about your OS/compiler?
Derek
On Thu, Feb 25, 2016 at 4:18 PM Zhang, Hong wrote:
> Dear all,
>
> I am using the latest version of libmesh from the github repository
Yep, you got down to the right thing: with SerialMesh and Exodus we do
partition independent looping on processor 0... so the files always come
out the same.
However, if you use ParallelMesh with Exodus (yes, that doesn't quite makes
sense to do... but WE do it a lot... especially in our test suit
Are the fine and coarse meshes actually part of the same Mesh and therefore
part of the same System? Why not break up the problems into two separate
Meshes and two separate Systems? Then the indexing will be straight forward.
Derek
On Wed, Feb 10, 2016 at 7:34 AM Giacomo Rosilho de Souza <
giacomo
Also: it can depend on integration error in the integration of the L2
Error. We're still using quadrature to integrate... so if that quadrature
is poor you can "miss" solution features on a coarse grid that then show up
as you refine the mesh (which refines the quadrature too) leading to higher
err
You may be interested in the "Checkpoint" mesh format:
https://github.com/libMesh/libmesh/blob/master/include/mesh/checkpoint_io.h
It is still XDR based... but basically just writes out the raw mesh as it
sits on each processor and then reads it back in during a restart. Because
of this you have
Cool - I wasn't arguing against this in any way... just wanted to hear your
thoughts on the trade offs. Thanks!
Derek
On Sun, Jan 10, 2016 at 11:28 AM David Knezevic
wrote:
> On Sun, Jan 10, 2016 at 10:01 AM, Derek Gaston wrote:
>
>> Just a question: why the "isolated n
Just a question: why the "isolated node" instead of scalar variables for
this constraint?
I'm just wondering what you see as the trade offs between both.
Derek
On Fri, Jan 8, 2016 at 4:16 PM David Knezevic
wrote:
> On Fri, Jan 8, 2016 at 4:07 PM, Roy Stogner
> wrote:
>
> >
> > On Fri, 8 Jan 20
On Fri, Nov 27, 2015 at 9:57 AM John Peterson wrote:
> A question for people that know more about Eigen than me: Could
> DenseMatrix just be changed to an Eigen::MatrixXf or something similar?
> There might be some trickiness with interfacing with DenseSubMatrix, but it
> would be nice to get all
Instruments on Mac OSX. Simple. Works.
Derek
On Mon, Nov 9, 2015 at 9:51 PM Roy Stogner wrote:
>
> On Mon, 9 Nov 2015, Miguel Angel Salazar de Troya wrote:
>
> > My intentions are to profile my libmesh application. Which tools do you
> > guys usually employ for this purpose?
>
> kcachegrind for
--enable-all-static CXX=mpicxx CC=mpicc F77=mpif77 FC=mpif90
>>
>>
>>
>> So you were using --enable-all-static at some point. If you just want to
>> see whether dynamic li
. What if the elem pointer is not available,
> i.e., how can I
> obtain the i-th element in a parallel mesh without iterating over the
> mesh? Shall I use
> mesh.query_elem()?
>
> Cheers,
> Dafang
>
> On 9/1/2015 5:17 PM, Derek Gaston wrote:
> > if (elem->processo
if (elem->processor_id() == mesh.processor_id())
On Tue, Sep 1, 2015 at 5:12 PM Dafang Wang wrote:
> Hi,
>
> I am wondering how to query whether a processor owns a given element in a
> parallel mesh,
> assuming that the finite-element mesh is partitioned across multiple
> processors. One way I
>
Are you using std::cout to print a message out?
libMesh (by default) eats all std::cout output on all processors other than
processor zero. You can use the command-line option of "--keep-cout" to
keep that from happening... or you can print to std::err (which is
normally what I do in these cases
On Wed, Aug 12, 2015 at 2:10 PM David Knezevic
wrote:
>
> Cool, thanks for pointing that out. I remember this being discussed on the
> list, but I haven't used it before. It only works in the case that you
> write and read with the same number of processors, right?
>
Yep - that's right.
Well -
Right - that's one of the reasons I created the CheckpointIO Mesh writer:
http://libmesh.github.io/doxygen/classlibMesh_1_1CheckpointIO.html
Derek
On Wed, Aug 12, 2015 at 1:55 PM David Knezevic
wrote:
> On Wed, Aug 12, 2015 at 1:47 PM, Derek Gaston wrote:
>
>> How are yo
How are you writing out the solution? This certainly doesn't happen when
using Exodus (Hilbert keys aren't involved there at all).
Derek
On Wed, Aug 12, 2015 at 1:45 PM David Knezevic
wrote:
> On Wed, Aug 12, 2015 at 1:32 PM, John Peterson
> wrote:
>
> > Sorry it has taken a while for me to r
:33 PM Junchao Zhang
wrote:
> I used GMVIO(mesh).write("output.gmv");
>
> --Junchao Zhang
>
> On Wed, Aug 5, 2015 at 2:32 PM, Derek Gaston wrote:
>
>> What file format were you using to write the mesh out and view it in
>> Paraview?
>>
>
What file format were you using to write the mesh out and view it in
Paraview?
Derek
On Wed, Aug 5, 2015 at 12:14 PM Junchao Zhang
wrote:
> <>
>
> In addition, I did more experiments. I increased the input mesh size, let
> it has 64 elements and ran with 12 processes. It looks libmesh keeps th
Why do you need to serialize the mesh? If you're writing out any of the
serial formats (Exodus, VTK, Tecplot, etc.) then the mesh is serialized
internally... you don't need to do it yourself.
Derek
On Sat, Apr 18, 2015 at 12:03 PM Harshad Sahasrabudhe
wrote:
> Hi Roy,
>
> Sorry I forgot to pos
Remember: localize_to_one() only pulls the vector down to one processor
(processor 0 by default). If you need a full copy of the vector on every
processor then you should use localize().
Derek
On Thu, Apr 16, 2015 at 11:25 AM John Peterson wrote:
> On Thu, Apr 16, 2015 at 8:52 AM, ernestol wrot
Emacs
On Mon, Mar 30, 2015 at 4:31 PM John Peterson wrote:
> On Mon, Mar 30, 2015 at 2:23 PM, Miguel Angel Salazar de Troya <
> salazardetr...@gmail.com> wrote:
>
> > Hi all
> >
> > I was wondering which IDE or text editor you guys use to develop libmesh.
> > I've been using eclipse for one year
Forgot to send this to the list..
-- Forwarded message -
From: Derek Gaston
Date: Wed, Feb 11, 2015 at 10:02 PM
Subject: Re: [Libmesh-users] Getting xyz coordinates from a global dof id
To: Roy Stogner
On Wednesday, February 11, 2015, Roy Stogner
wrote:
> If you'r
That question doesn't compute. Do you mean get the xyz position of the
node associated with a Lagrange DoF?
On Wed Feb 11 2015 at 8:25:57 PM EST Vikram Garg
wrote:
> Can someone tell me how one can obtain the xyz coordinates corresponding to
> a global dof id ?
>
> Thanks.
> --
> Vikram Garg
>
If you're trying to solve a different problem on each processor you'll want
to create an MPI sub-communicator for each processor and then initialize
the Mesh with that (Note the Mesh constructor that takes a
Parallel::Communicator:
http://libmesh.github.io/doxygen/classlibMesh_1_1Mesh.html ). I be
one to explosion, I want to keep
> monitoring it and to interrupt SNES when the residual explodes.
>
> Cheers,
> Dafang
>
>
> On 12/10/2014 11:43 PM, Derek Gaston wrote:
>
> If you're using PETSc and doing a "normal" Newton solve then you can try
> to us
If you're using PETSc and doing a "normal" Newton solve then you can try to
use SNESSetFunctionDomainError() (
http://www.mcs.anl.gov/petsc/petsc-current/docs/manualpages/SNES/SNESSetFunctionDomainError.html
) and then return... and hopefully PETSc will clean up for you.
However, if you're doing J
For those following along you can take a look at my first cut here:
https://github.com/idaholab/moose/pull/4346
Derek
On Wed, Dec 3, 2014 at 10:29 PM, Derek Gaston wrote:
> It will be a MOOSE PR. It's mostly pure libMesh code but it is using a
> MOOSE object: LineSegmen
put it in MOOSE first and then we can see if we want to move it.
Derek
On Wed, Dec 3, 2014 at 9:56 PM, Roy Stogner
wrote:
>
> On Wed, 3 Dec 2014, Derek Gaston wrote:
>
> It's not done yet... just pushed up to save my work.
>>
>> Building tests for it now. Look for
It's not done yet... just pushed up to save my work.
Building tests for it now. Look for a PR a little later tonight...
Derek
On Wed, Dec 3, 2014 at 9:34 PM, Benjamin W Spencer wrote:
> That was easy! Thanks!
>
> Ben
>
> On Dec 3, 2014, at 7:15 PM, Derek Gaston wrote:
I'm going to do this for you. Coding now.
Derek
On Wed, Dec 3, 2014 at 7:02 PM, Benjamin W Spencer wrote:
> Just a few. I want to write a VectorPostprocessor like the
> LineValueSampler, but for material data.
>
> On Dec 3, 2014, at 5:52 PM, Derek Gaston wrote:
>
> AKA
AKA "ray tracing". It's a difficult problem on unstructured meshes.
How many "rays" are you going to have? Tracing _one_ vs millions will lead
to different algorithms...
Derek
On Wed, Dec 3, 2014 at 5:48 PM, Benjamin Spencer
wrote:
> I need to find the set of elements that a straight line pa
On Fri, Oct 17, 2014 at 9:54 AM, Roy Stogner
wrote:
> Should we just create libmesh.github.io and put the docs up there?
>
Yep - it's needed to happen for a while anyway. Time to move
completely
--
Comprehensive Ser
Just a heads up: we have been using "--enable-static --disable-shared"
exclusively on one of our clusters... so it's getting tested :-)
Miguel: I would still recommend going with shared libraries... generally
clusters have slow filesystems and link times for huge static executables
can be atrociou
Just a quick note here... I've successfully used SerialMesh on over 10,000
processors... and ParallelMesh on over 100,000
That "128 processors" quote from the paper is quaint ;-)
Derek
On Mon, Jul 14, 2014 at 1:30 PM, John Peterson wrote:
> On Mon, Jul 14, 2014 at 11:32 AM, Hodge, Neil E. wr
I would find a bit more fuzziness here to be useful myself
Derek
On Tue, Jul 15, 2014 at 8:53 PM, David Knezevic
wrote:
>
> On 07/15/2014 10:51 PM, Roy Stogner wrote:
> >
> > On Tue, 15 Jul 2014, David Knezevic wrote:
> >
> >> I use MeshFunction a lot, it's very helpful. But sometimes due
solution is not ghosted. current_local_solution is.
"Ghosting" in this case means that it contains extra entries in it that
hold off-processor values that border elements _this_ processor owns (those
values are necessary/useful for a finite element assembly).
current_local_solution is updated wh
s to the
> false argument. Given this, wouldn’t MeshOputput::write_equation_system()
> then serialize the whole mesh for ExodusII and not for Nemesis?
>
> Or am I missing something?
>
> Manav
>
>
> On May 10, 2014, at 12:11 PM, Derek Gaston wrote:
>
> Just a head
gt;
> Now, in the last step, I am getting the error that I described in my
> previous email, which I am trying to decipher.
>
> Manav
>
>
> On May 10, 2014, at 11:00 AM, Derek Gaston wrote:
>
> I'm not sure how (or why?) you're using SerialMesh with Nemesis...
I'm not sure how (or why?) you're using SerialMesh with Nemesis... Nemesis
is for reading Parallel Meshes...
What exactly are you trying to do here?
Derek
On Sat, May 10, 2014 at 8:52 AM, Manav Bhatia wrote:
> Hi,
>
> I am curious if SerialMesh with AMR uses the RemoteElem. In the
> f
ing "git submodule update"... again through a script).
Nice and tidy - and no waiting on libMesh releases.
On Fri, May 9, 2014 at 4:55 PM, Derek Gaston wrote:
> Sorry for one more email on this topic... but if you want to see how we
> track libMesh revisions look at our rep
hash of the version of
libMesh we're using and all of our customers get the new revisions of
libMesh automatically when they update their version of our software (and
after running "git submodule update").
Nice and tidy - and no waiting on libMesh releases.
Derek
On Fri, May 9, 2
To be clear: we update libMesh about monthly... each time just updating to
HEAD in the Git repo...
Derek
On Fri, May 9, 2014 at 4:50 PM, Derek Gaston wrote:
> Michael - can you really not use the Git version? A libMesh "release"
> really isn't much more special than a
Michael - can you really not use the Git version? A libMesh "release"
really isn't much more special than any version in the Git repo... in fact
all it means is that it's usually behind on bug fixes
We've never used a "release" and we distribute libMesh to hundreds of
customers daily without
We're hiring on the MOOSE Framework team! Come join a high-intensity
computational science team that is devoted to open source and innovative
development methods!
The MOOSE Framework ( http://www.mooseframework.org ) is a high-level,
parallel, multiscale, multiphysics, PDE solution framework buil
If you are going to be in Western Australia in May, swing by and catch Cody
Permann and myself doing a free and open training course for the MOOSE
Framework! The course is sponsored by iVec and CSIRO.
http://www.ivec.org/ai1ec_event/moose-3-day-course/
The MOOSE Framework is a nonlinear, multip
ortantly the DofMap is updated to
> find vol/surface coupling (not just vol/vol through a face) I think that'd
> be close.
>
> > On Mar 27, 2014, at 5:43 PM,
27, 2014, at 5:16 PM, "Roy Stogner"
> wrote:
> >
> >
> >> On Thu, 27 Mar 2014, Derek Gaston wrote:
> >>
> >> Is this the place to use BoundaryMesh?
> >
> > That's what we do, yeah.
> >
> >> Can you get the normals for the ele
How about restricting the variable to the boundary - is there any path
forward on that?
Derek
On Thu, Mar 27, 2014 at 4:16 PM, Roy Stogner wrote:
>
> On Thu, 27 Mar 2014, Derek Gaston wrote:
>
> Is this the place to use BoundaryMesh?
>>
>
> That's what we do,
I need to do a solve for a different set of variables on the side of a 3D
domain. Meaning that I need a variable that only exists on the side of a
3D domain (boundary restricted variable?). Also - I still need the outward
facing normal (this "side" is going to be highly irregular).
Any ideas her
Thanks Ben! Collaboration will be MUCH easier now!
Derek
On Wed, Mar 19, 2014 at 10:38 PM, Kirk, Benjamin (JSC-EG311) <
benjamin.k...@nasa.gov> wrote:
> This is great news, and I'm looking forward to continued collaboration!
>
> -Ben
>
>
> > On Mar 19, 2014, at 6:56 PM, "Gaston, Derek R"
> wr
BTW - VTK reads Exodus files just fine if you would rather stick to a
format that is a little more mature in libMesh
Derek
On Wed, Feb 26, 2014 at 5:17 AM, Roy Stogner wrote:
>
> On Wed, 26 Feb 2014, Erich Gaertig wrote:
>
> > vtkUnstructuredGrid *grid;
> > grid = VTKIO (mesh).get_vtk_grid(
The doxygen is out of date.
Also - we really need to move it to GitHub... it's one of the few things
left back at Sourceforge
Derek
On Fri, Feb 7, 2014 at 9:35 AM, Michael Povolotskyi wrote:
> On 02/07/2014 11:27 AM, Derek Gaston wrote:
>
>> Michael - I can vouch tha
Michael - I can vouch that using different communicators for Mesh (and
everything else) does work in libMesh... it's something we have done quite
a bit of.
If you are struggling, please send along a small example.
Thanks,
Derek
On Thu, Feb 6, 2014 at 4:20 PM, Michael Povolotskyi wrote:
> On 0
This is almost definitely not a libMesh problem. It sounds like an
uninitialized variable. I recommend running your application using
valgrind ( http://valgrind.org ) . Use a debug (dbg) build of your
application and valgrind will point you right to the problem.
Derek
On Thu, Jan 2, 2014 at 7
It's funny - I was just getting ready to push a patch for this as it was a
compile error on BG/Q. Did you already fix it John or do you want me to
push my patch?
Derek
On Tue, Dec 10, 2013 at 2:30 PM, Roy Stogner wrote:
>
>
> On Tue, 10 Dec 2013, Santiago Alonso Pérez Rubiano wrote:
>
> I'm g
Actually - those look really good. What preconditioner are you using now?
If you're happy with the speed I would stop there...
But - yes, after recompiling PETSc you would want to recompile libMesh...
Derek
On Tue, Dec 10, 2013 at 8:58 AM, Lorenzo Zanon
wrote:
> Hi,
>
> After re-installing p
.417540e-04
> NL step 2, |residual|_2 = 6.174706e-08
> NL step 3, |residual|_2 = 3.577768e-12
> NL step 4, |residual|_2 = 7.687278e-17
> StVen system solved at nonlinear iteration 4 , final nonlinear residual
> norm: 7.687278e-17
>
>
> I think it would be fine to
How is it failing in parallel? Is it not converging, or something else?
What preconditioner are you using? How well are the linear solves going
(use -ksp_monitor)?
Derek
On Mon, Dec 9, 2013 at 9:59 AM, subramanya sadasiva wrote:
> Hi,
> i am trying to run a nonlinear solver (a multiphase ca
1 - 100 of 436 matches
Mail list logo
