Re: [petsc-users] Number of levels of multigrid : 2-3 is sufficient ??

On Wed, Oct 14, 2015 at 9:32 PM, Timothée Nicolas < timothee.nico...@gmail.com> wrote: > Thank you Barry for pointing this out. Indeed on a system with no > debugging the Jacobian evaluations no longer dominate the time (less than > 10%). However the rest is similar, except the improvement from 2

Re: [petsc-users] Diagnosing Poisson Solver Behavior

Hello Barry, The full error message is [0]PETSC ERROR: - Error Message -- [0]PETSC ERROR: Argument out of range [0]PETSC ERROR: Inserting a new nonzero at (25,27) in the matrix [0]PETSC ERROR: See

Re: [petsc-users] Diagnosing Poisson Solver Behavior

On Thu, Oct 15, 2015 at 10:43 AM, K. N. Ramachandran wrote: > Hello Barry, > > The full error message is > > [0]PETSC ERROR: - Error Message > -- > [0]PETSC ERROR: Argument out of range > [0]PETSC

[petsc-users] MatPtAP for involving MPIDENSE and MPIAIJ matrices

HI, I'm trying to evaluate B = P^T * A * P, where P is an MPIDENSE matrix and A is an MPIAIJ matrix. Neither MatTransposeMatMult nor MatPtAP support such an operation for the given type of matrices. I tried to look into MATELEMENTAL class and didn't find any MatTransposeMatMult function for a

Re: [petsc-users] Load balancing problem with DMDA

On Thu, Oct 15, 2015 at 8:55 AM, Xiao, Jianjun (IKET) wrote: > Matt, > > > Yes. I would like to try. Where should I start? > I would take a small 2D problem that you care about and try to make an equivalent Plex code to the DMDA code. I think it will be slower than DMDA,

Re: [petsc-users] MatPtAP for involving MPIDENSE and MPIAIJ matrices

On Thu, Oct 15, 2015 at 8:20 AM, Bikash Kanungo wrote: > HI, > > I'm trying to evaluate B = P^T * A * P, where P is an MPIDENSE matrix and > A is an MPIAIJ matrix. Neither MatTransposeMatMult nor MatPtAP support such > an operation for the given type of matrices. I tried to

Re: [petsc-users] Load balancing problem with DMDA

Matt, Yes. I would like to try. Where should I start? From: Matthew Knepley Sent: Thursday, October 15, 2015 3:43 PM To: Barry Smith Cc: Xiao, Jianjun (IKET); petsc-users@mcs.anl.gov Subject: Re: [petsc-users] Load balancing problem with

Re: [petsc-users] Number of levels of multigrid : 2-3 is sufficient ??

OK, perfect, thank you. 2015-10-16 6:01 GMT+09:00 Barry Smith : > > I guess the -pc_mg_type full is the best you are going to get. In > parallel the coarser grids are problematic because they have little local > work but still communication. > > Barry > > > > On Oct 15,

[petsc-users] HDF5 vec output with DMPlex

hi, I am wanting to output time-stepped simulation results to an HDF5 file, from a PETSc Vec created on a DMPlex using DMCreateGlobalVector(). I tried doing it the way it's done in src/vec/vec/examples/tutorials/ex19.c, using PetscViewerHDF5Open(), PetscViewerHDF5PushGroup() and

Re: [petsc-users] HDF5 vec output with DMPlex

On Thu, Oct 15, 2015 at 10:51 PM, Adrian Croucher wrote: > hi, > > I am wanting to output time-stepped simulation results to an HDF5 file, > from a PETSc Vec created on a DMPlex using DMCreateGlobalVector(). > > I tried doing it the way it's done in >

Re: [petsc-users] Number of levels of multigrid : 2-3 is sufficient ??

I guess the -pc_mg_type full is the best you are going to get. In parallel the coarser grids are problematic because they have little local work but still communication. Barry > On Oct 15, 2015, at 12:26 AM, Timothée Nicolas > wrote: > > OK, > > I ran an

Re: [petsc-users] MatPtAP for involving MPIDENSE and MPIAIJ matrices

r<#secure method=pgpmime mode=sign> Hong writes: > I plan to implement MatTransposeMatMult_MPIDense_MPIDense via > > 1. add MatTransposeMatMult_elemental_elemental() > 2. C_dense = P_dense^T * B_dense > via MatConvert_dense_elemental() and MatConvert_elemental_dense()

Re: [petsc-users] HDF5 vec output with DMPlex

hi On 16/10/15 16:54, Matthew Knepley wrote: No, that should work. This is how we do it in PyLith. There must be some setup problem. When you look at the HDF5 file. Does it have a dimension for timestep? No, it doesn't. Is there any extra setup I might need to do to get it to work? I've

Re: [petsc-users] Diagnosing Poisson Solver Behavior

On Thu, Oct 15, 2015 at 12:10 PM, K. N. Ramachandran wrote: > Hello Matt, > > On Thu, Oct 15, 2015 at 12:18 PM, Matthew Knepley > wrote: > >> >> The bad design decision is using a storage format designed for high speed >> MatMult() to use for data

Re: [petsc-users] MatPtAP for involving MPIDENSE and MPIAIJ matrices

> On Oct 15, 2015, at 8:40 AM, Matthew Knepley wrote: > > On Thu, Oct 15, 2015 at 8:20 AM, Bikash Kanungo wrote: > HI, > > I'm trying to evaluate B = P^T * A * P, where P is an MPIDENSE matrix and A > is an MPIAIJ matrix. Neither MatTransposeMatMult nor

Re: [petsc-users] MatPtAP for involving MPIDENSE and MPIAIJ matrices

I plan to implement MatTransposeMatMult_MPIDense_MPIDense via 1. add MatTransposeMatMult_elemental_elemental() 2. C_dense = P_dense^T * B_dense via MatConvert_dense_elemental() and MatConvert_elemental_dense() Let me know if you have better suggestions. Hong On Thu, Oct 15, 2015 at 1:49

Re: [petsc-users] MatPtAP for involving MPIDENSE and MPIAIJ matrices

On Thu, Oct 15, 2015 at 1:36 PM, Jed Brown wrote: > Matthew Knepley writes: > > Jed, is TAIJ the way to do this? > > No, TAIJ is unrelated. > I do not understand "unrelated". My understanding was that TAIJ could be used (with T = I) to get the action of A

Re: [petsc-users] MatPtAP for involving MPIDENSE and MPIAIJ matrices

Matthew Knepley writes: > I do not understand "unrelated". My understanding was that TAIJ could be > used (with T = I) > to get the action of A on a set of vectors, which I think would be A P. Why > specifically would > you not use it? MAIJ is enough for that; it would involve

Re: [petsc-users] Diagnosing Poisson Solver Behavior

"K. N. Ramachandran" writes: > I am trying to simulate the movement of charged particles in a domain, and > these particles become collimated along a particular direction. So one way > to partition might be 1D slabs perpendicular to the collimated direction, > so each rank can