Mark, should we provide some more flexible way to label fields? It
will be more complicated than the present code and I think packing into
interlaced format is faster anyway.
I'm thinking this would entail reordering the matrix and inverting this
permutation for the fine grid R P. This
It is difficult to understand what you are doing here. What is dim? What is
NX and NY? Is the length of inpx and inpw 256*256 ? Are you using a PETSc
Mat like AIJ to apply the “fast convolution” or some custom MATSHELL? Is the
“fast convolution” the same for each dim, i and j or is it
I am moving my code to have the DM create the matrix and am hitting this
error in DMGetLocalToGlobalMapping:
if (!dm-ops-getlocaltoglobalmapping)
SETERRQ(PetscObjectComm((PetscObject)dm),PETSC_ERR_SUP,DM can not create
LocalToGlobalMapping);
This happens when I create the matrix like this:
adding a subject ...
On Wed, Aug 6, 2014 at 12:00 PM, Mark Adams mfad...@lbl.gov wrote:
I am moving my code to have the DM create the matrix and am hitting this
error in DMGetLocalToGlobalMapping:
if (!dm-ops-getlocaltoglobalmapping)
It looks like DMShellCreate does not set dm-ops-getlocaltoglobalmapping,
which is needed for DMCreateMatrix with a composite DM. I do set a global
and local vector and a matrix. It looks like I could call
DMShellSetLocalToGlobal.
Should I be doing this? Seems low level and the kind of thing
Hi Barry,
Thanks for your email. sorry i did not make it clear. Here is a more detailed
one:
int dim, i, j;
int NDOF=3, NX=5, NY=5;
for(dim=0; dimNDOF; dim++)
{
for(i=0; iNX; i++)
{
for(j=0; jNY; j++)
{
//compute inpx
Set
Now that I look at DMShellSetLocalToGlobal, I see that it sets the
begin/end functions. But these are already set.
It looks like DMShell's SetType must set something for
dm-ops-getlocaltoglobalmapping
for DMShell to be usable to get a matrix from a Composite DM.
On Wed, Aug 6, 2014 at 2:34 PM,
Hello,
I am trying to implement a Flux Correction Transport method for advection
equations (e.g. http://www.mathematik.uni-dortmund.de/~kuzmin/linfct.pdf).
One of the steps consists of calculating a matrix D which is such that
d_ij = max{ 0 , -k_ij , -kJi }
where k_ij are the components of a
Jed, it looks like DM_DA creates its l2gmap in Setup, but DM_Shell does
not. This seems to be causing errors when DM_Shell is used in DM_Composit
and I make a matrix.
I'm thinking I could add this construction to DM_Shell. DM_Shell does not
have a Setup method (like DM_DA) but I could create
I see there is a DM_Create_Shell that I can out this in.
On Wed, Aug 6, 2014 at 6:31 PM, Mark Adams mfad...@lbl.gov wrote:
Jed, it looks like DM_DA creates its l2gmap in Setup, but DM_Shell does
not. This seems to be causing errors when DM_Shell is used in DM_Composit
and I make a matrix.
Mathieu MORLIGHEM mathieu.morlig...@uci.edu writes:
Hello,
I am trying to implement a Flux Correction Transport method for advection
equations (e.g. http://www.mathematik.uni-dortmund.de/~kuzmin/linfct.pdf).
One of the steps consists of calculating a matrix D which is such that
d_ij =
Hi,
So in ex62 of SNES it solves the stokes equation with P2-P1 elements. I
want to solve the darcy equation with P2-P1 elements in a similar fashion.
The difference between stokes and darcy is that in the weak form of darcy,
the velocity-velocity term is v*u whereas in stokes it's
Hello,
I am trying to figure out a faster way of doing in petsc4py and thought I would
ask to see how this should be done in petsc. If I can figure that out then I
think I can figure out how to translate it but I am having issues with the
basic algorithm I should be using.
I am solving a
On Wed, Aug 6, 2014 at 5:55 PM, Justin Chang jychan...@gmail.com wrote:
Hi,
So in ex62 of SNES it solves the stokes equation with P2-P1 elements. I
want to solve the darcy equation with P2-P1 elements in a similar fashion.
The difference between stokes and darcy is that in the weak form of
On Wed, Aug 6, 2014 at 6:50 PM, Francis Poulin fpou...@uwaterloo.ca wrote:
Hello,
I am trying to figure out a faster way of doing in petsc4py and thought I
would ask to see how this should be done in petsc. If I can figure that
out then I think I can figure out how to translate it but I am
Hello Matt,
Thanks for the help.
I had some some tests in Matlab using the two methods and because I am trying
to compute the growth rates of a system that are very small, it seemed to be
faster to find the inverse but I will try both and see what works best.
Yes, my inexperience let me
16 matches
Mail list logo