[petsc-users] Parallel TS for ODE
Stefano Zampini
stefano.zampini at gmail.com
Tue May 4 03:57:23 CDT 2021
Using DMDA to get automatic parallelization of a 0-D code is nonsense.
Ghost points are meant for spatially distributed data, not for 0D.
If you really want to use DM, you should use DMRedundant and call
DMGlobalToLocal to go to your distributed to local full representation (it
calls MPI_Bcast internally).
My advice is to read the many examples we have.
Il giorno mar 4 mag 2021 alle ore 10:54 Francesco Brarda <
brardafrancesco at gmail.com> ha scritto:
> Thank you very much everyone. I do have one more question.
>
> For what you want to do you can use 3,1,2. This says three "spatial"
> points, 1 dof at each "spatial" point and 2 ghost points. In your
> case "spatial" does not mean spatial in space it is just three abstract
> points.
>
> In this case, since I have 2 ghost points, should I change the
> DMBoundaryType?
> For one process this works, but with 2 or 3 procs it doesn’t. The error I
> have is the following:
>
> Solving a non-linear TS problem, number of processors = 2
> [1]PETSC ERROR: --------------------- Error Message
> --------------------------------------------------------------
> [1]PETSC ERROR: Argument out of range
> [1]PETSC ERROR: Local x-width of domain x 1 is smaller than stencil width
> s 2
> [1]PETSC ERROR: See https://www.mcs.anl.gov/petsc/documentation/faq.html
> for trouble shooting.
> [1]PETSC ERROR: Petsc Release Version 3.14.4, unknown
> [1]PETSC ERROR: ./test_ic on a arch-debug named srvulx13 by fbrarda Mon
> May 3 17:46:37 2021
> [1]PETSC ERROR: Configure options --with-cc=gcc --with-cxx=g++
> --with-fc=gfortran --with-openblas-dir=/opt/packages/openblas/0.2.13-gcc
> --download-mpich PETSC_ARCH=arch-debug
> [1]PETSC ERROR: #1 DMSetUp_DA_1D() line 199 in
> /home/fbrarda/petsc/src/dm/impls/da/da1.c
> [1]PETSC ERROR: #2 DMSetUp_DA() line 20 in
> /home/fbrarda/petsc/src/dm/impls/da/dareg.c
> [1]PETSC ERROR: #3 DMSetUp() line 787 in
> /home/fbrarda/petsc/src/dm/interface/dm.c
> [1]PETSC ERROR: #4 main() line 232 in test_ic.c
> [1]PETSC ERROR: No PETSc Option Table entries
> [1]PETSC ERROR: ----------------End of Error Message -------send entire
> error message to petsc-maint at mcs.anl.gov----------
>
> Il giorno 2 mag 2021, alle ore 18:54, Barry Smith <bsmith at petsc.dev> ha
> scritto:
>
>
> [0]PETSC ERROR: Wrong subtype object:Parameter # 1 must have
> implementation da it is shell
>
> Are you calling TSSetDM() to supply your created DMDA to the TS? Based on
> the error message you are not, it is using a default shell DM, which is
> what TS does if you do not provide them with a DM. You need to call
> TSSetDM() after you create the TS and the DMDA.
>
>
> On Apr 29, 2021, at 2:57 AM, Francesco Brarda <brardafrancesco at gmail.com>
> wrote:
>
> I defined the DM as follows
> ierr =
> DMDACreate1d(PETSC_COMM_WORLD,DM_BOUNDARY_NONE,3,3,3,NULL,&da);CHKERRQ(ierr);
>
>
> If you truly want one "spatial" point then you would want to use 1,3,0.
> This says 1 location in space, three degrees of freedom at that point and 0
> ghost values (since there is only one spatial point there can be no ghost
> spatial values).
>
> BUT DMDA ALWAYS puts all degree of freedom at a point on the same process
> so this will not give you parallelism. All 3 dof will be on the same MPI
> rank.
>
> For what you want to do you can use 3,1,2. This says three "spatial"
> points, 1 dof at each "spatial" point and 2 ghost points. In your
> case "spatial" does not mean spatial in space it is just three abstract
> points.
>
> The global vectors (from DMCreate or Get GlobalVector) will have 1 value
> on each rank. The local vector from DMCreate or Get LocalVector) will have
> three values on each rank. Your initial conditions code would be something
> like
>
> if (rank == 0) {
>
> x[0] = appctx->N - appctx->p[2];
>
> } else if (rank == 1) {
>
> x[1] = appctx->p[2];
>
> } else {
>
> x[2] = 0.0;
>
> }
>
>
> Your TSSetRHSFunction() would make a call to DMGetLocalVector(...&localX),
> do a DMGlobalToLocalBegin/End() from the input global X to localX, you
> would call DMDAVecGetArray(...,&xarray) on the localX and access all three
> values in xarray. The resulting computation of f the output vector would be
> something like
>
> if (rank == 0) {
>
> farray[0] = your code that can use xarray[0], xarray[1], xarray[2]
>
> } else if (rank == 1) {
>
> farray[1] = your code that can use xarray[0], xarray[1], xarray[2]
>
> } else {
>
> farray[2] = your code that can use xarray[0], xarray[1], xarray[2]
>
> }
>
> There are many examples of this pattern in the example tutorials.
>
> When you implement a code with a spatial distribution you would use a dof
> of 3 at each point and not parallelize over the dof at each point. Likely
> you want to use DMNETWORK to manage the spatial distribution since it has a
> simple API and allows any number of different number of neighbors for each
> point. DMDA would not make sense for true spatial distribution except in
> some truly trivial neighbor configurations.
>
> Barry
>
>
>
>
> I am not sure whether I correctly understood this command properly. The
> vector should have 3 components (S, I, R) and 3 DOF as it is defined only
> when the three coordinates have been set.
> Then I create a global vector X. When I set the initial conditions as
> below
>
> static PetscErrorCode InitialConditions(TS ts,Vec X, void *ctx)
> {
> PetscErrorCode ierr;
> AppCtx *appctx = (AppCtx*) ctx;
> PetscScalar *x;
> DM da;
>
> PetscFunctionBeginUser;
> ierr = TSGetDM(ts,&da);CHKERRQ(ierr);
>
> /* Get pointers to vector data */
> ierr = DMDAVecGetArray(da,X,(void*)&x);CHKERRQ(ierr);
>
> x[0] = appctx->N - appctx->p[2];
> x[1] = appctx->p[2];
> x[2] = 0.0;
>
> ierr = DMDAVecRestoreArray(da,X,(void*)&x);CHKERRQ(ierr);
> PetscFunctionReturn(0);
> }
>
> I have the error:
>
> [0]PETSC ERROR: Invalid argument
> [0]PETSC ERROR: Wrong subtype object:Parameter # 1 must have
> implementation da it is shell
> [0]PETSC ERROR: See https://www.mcs.anl.gov/petsc/documentation/faq.html for
> trouble shooting.
> [0]PETSC ERROR: Petsc Release Version 3.14.4, unknown
> [0]PETSC ERROR: ./par_sir_model on a arch-debug named srvulx13 by fbrarda
> Thu Apr 29 09:36:17 2021
> [0]PETSC ERROR: Configure options --with-cc=gcc --with-cxx=g++
> --with-fc=gfortran --with-openblas-dir=/opt/packages/openblas/0.2.13-gcc
> --download-mpich PETSC_ARCH=arch-debug
> [0]PETSC ERROR: #1 DMDAVecGetArray() line 48 in
> /home/fbrarda/petsc/src/dm/impls/da/dagetarray.c
> [0]PETSC ERROR: #2 InitialConditions() line 175 in par_sir_model.c
> [0]PETSC ERROR: #3 main() line 295 in par_sir_model.c
> [0]PETSC ERROR: No PETSc Option Table entries
>
> I would be very happy to receive any advices to fix the code.
> Best,
> Francesco
>
> Il giorno 20 apr 2021, alle ore 21:35, Matthew Knepley <knepley at gmail.com>
> ha scritto:
>
> On Tue, Apr 20, 2021 at 1:17 PM Francesco Brarda <
> brardafrancesco at gmail.com> wrote:
> Thank you for the advices, I would just like to convert the code I already
> have to see what might happen once parallelized.
> Do you think it is better to put the 3 equations into a 1d Distributed
> Array with 3 dofs and run the job with multiple procs regardless of how
> many equations I have? Is it possible?
>
> If you plan in the end to use a structured grid, this is a great plan. If
> not, this is not a good plan.
>
> Thanks,
>
> Matt
>
> Thank you,
> Francesco
>
> Il giorno 20 apr 2021, alle ore 17:57, Stefano Zampini <
> stefano.zampini at gmail.com> ha scritto:
>
> It does not make sense to parallelize to 1 equation per process, unless
> that single equation per process is super super super costly.
> Is this work you are doing used to understand PETSc parallelization
> strategy? if so, there are multiple examples in the sourcetree that you can
> look at to populate matrices and vectors in parallel
>
> Il giorno mar 20 apr 2021 alle ore 17:52 Francesco Brarda <
> brardafrancesco at gmail.com> ha scritto:
> In principle the entire code was for 1 proc only. The functions were built
> with VecGetArray(). While adapting the code for multiple procs I thought
> using VecGetOwnershipRange was a possible way to allocate the equations in
> the vector using multiple procs. What do you think, please?
>
> Thank you,
> Francesco
>
> Il giorno 20 apr 2021, alle ore 16:43, Matthew Knepley <knepley at gmail.com>
> ha scritto:
>
> On Tue, Apr 20, 2021 at 10:41 AM Francesco Brarda <
> brardafrancesco at gmail.com> wrote:
> I was trying to follow Barry's advice some time ago, but I guess that's
> not the way he meant it. How should I refer to the values contained in x?
> With Distributed Arrays?
>
> That is how you get values from x. However, I cannot understand at all
> what you are doing with "mybase".
>
> Matt
>
> Thanks
> Francesco
>
> Even though it will not scale and will deliver slower performance it is
> completely possible for you to solve the 3 variable problem using 3 MPI
> ranks. Or 10 mpi ranks. You would just create vectors/matrices with 1
> degree of freedom for the first three ranks and no degrees of freedom for
> the later ranks. During your function evaluation (and Jacobian evaluation)
> for TS you will need to set up the appropriate communication to get the
> values you need on each rank to evaluate the parts of the function
> evaluation needed by that rank. This is true for parallelizing any
> computation.
>
> Barry
>
>
>
>
> Il giorno 20 apr 2021, alle ore 15:40, Matthew Knepley <knepley at gmail.com>
> ha scritto:
>
> On Tue, Apr 20, 2021 at 9:36 AM Francesco Brarda <
> brardafrancesco at gmail.com> wrote:
> Hi!
> I tried to implement the SIR model taking into account the fact that I
> will only use 3 MPI ranks at this moment.
> I built vectors and matrices following the examples already available. In
> particular, I defined the functions required similarly (RHSFunction,
> IFunction, IJacobian), as follows:
>
> I don't think this makes sense. You use "mybase" to distinguish between 3
> procs, which would indicate that each procs has only
> 1 degree of freedom. However, you use x[1] on each proc, indicating it has
> at least 2 dofs.
>
> Thanks,
>
> Matt
>
> static PetscErrorCode RHSFunction(TS ts,PetscReal t,Vec X,Vec F,void *ctx)
> {
> PetscErrorCode ierr;
> AppCtx *appctx = (AppCtx*) ctx;
> PetscScalar f;//, *x_localptr;
> const PetscScalar *x;
> PetscInt mybase;
>
> PetscFunctionBeginUser;
> ierr = VecGetOwnershipRange(X,&mybase,NULL);CHKERRQ(ierr);
> ierr = VecGetArrayRead(X,&x);CHKERRQ(ierr);
> if (mybase == 0) {
> f = (PetscScalar) (-appctx->p1*x[0]*x[1]/appctx->N);
> ierr = VecSetValues(F,1,&mybase,&f,INSERT_VALUES);
> }
> if (mybase == 1) {
> f = (PetscScalar) (appctx->p1*x[0]*x[1]/appctx->N-appctx->p2*x[1]);
> ierr = VecSetValues(F,1,&mybase,&f,INSERT_VALUES);
> }
> if (mybase == 2) {
> f = (PetscScalar) (appctx->p2*x[1]);
> ierr = VecSetValues(F,1,&mybase,&f,INSERT_VALUES);
> }
> ierr = VecRestoreArrayRead(X,&x);CHKERRQ(ierr);
> ierr = VecAssemblyBegin(F);CHKERRQ(ierr);
> ierr = VecAssemblyEnd(F);CHKERRQ(ierr);
> PetscFunctionReturn(0);
> }
>
>
> Whilst for the Jacobian I did:
>
>
> static PetscErrorCode IJacobian(TS ts,PetscReal t,Vec X,Vec Xdot,PetscReal
> a,Mat A,Mat B,void *ctx)
> {
> PetscErrorCode ierr;
> AppCtx *appctx = (AppCtx*) ctx;
> PetscInt mybase, rowcol[] = {0,1,2};
> const PetscScalar *x;
>
> PetscFunctionBeginUser;
> ierr = MatGetOwnershipRange(B,&mybase,NULL);CHKERRQ(ierr);
> ierr = VecGetArrayRead(X,&x);CHKERRQ(ierr);
> if (mybase == 0) {
> const PetscScalar J[] = {a + appctx->p1*x[1]/appctx->N,
> appctx->p1*x[0]/appctx->N, 0};
> ierr =
> MatSetValues(B,1,&mybase,3,rowcol,J,INSERT_VALUES);CHKERRQ(ierr);
> }
> if (mybase == 1) {
> const PetscScalar J[] = {- appctx->p1*x[1]/appctx->N, a -
> appctx->p1*x[0]/appctx->N + appctx->p2, 0};
> ierr =
> MatSetValues(B,1,&mybase,3,rowcol,J,INSERT_VALUES);CHKERRQ(ierr);
> }
> if (mybase == 2) {
> const PetscScalar J[] = {0, - appctx->p2, a};
> ierr =
> MatSetValues(B,1,&mybase,3,rowcol,J,INSERT_VALUES);CHKERRQ(ierr);
> }
> ierr = VecRestoreArrayRead(X,&x);CHKERRQ(ierr);
>
> ierr = MatAssemblyBegin(A,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr);
> ierr = MatAssemblyEnd(A,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr);
> if (A != B) {
> ierr = MatAssemblyBegin(B,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr);
> ierr = MatAssemblyEnd(B,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr);
> }
> PetscFunctionReturn(0);
> }
>
> This code does not provide the correct result, that is, the solution is
> the initial condition, either using implicit or explicit methods. Is the
> way I defined these objects wrong? How can I fix it?
> I also tried to print the Jacobian with the following commands but it does
> not work (blank rows and error message). How should I print the Jacobian?
>
> ierr = TSGetIJacobian(ts,NULL,&K, NULL, NULL); CHKERRQ(ierr);
> ierr = MatAssemblyBegin(K,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr);
> ierr = MatAssemblyEnd(K,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr);
> ierr = MatView(K,PETSC_VIEWER_STDOUT_WORLD);CHKERRQ(ierr);
>
>
> I would very much appreciate any kind of help or advice.
> Best,
> Francesco
>
> Il giorno 2 apr 2021, alle ore 04:45, Barry Smith <bsmith at petsc.dev> ha
> scritto:
>
>
>
> On Apr 1, 2021, at 9:17 PM, Zhang, Hong via petsc-users <
> petsc-users at mcs.anl.gov> wrote:
>
>
>
> On Mar 31, 2021, at 2:53 AM, Francesco Brarda <brardafrancesco at gmail.com>
> wrote:
>
> Hi everyone!
>
> I am trying to solve a system of 3 ODEs (a basic SIR model) with TS.
> Sequentially works pretty well, but I need to switch it into a parallel
> version.
> I started working with TS not very long time ago, there are few questions
> I’d like to share with you and if you have any advices I’d be happy to hear.
> First of all, do I need to use a DM object even if the model is only time
> dependent? All the examples I found were using that object for the other
> variable when solving PDEs.
>
>
> Are you considering SIR on a spatial domain? If so, you can parallelize
> your model in the spatial domain using DM. Splitting the three variables in
> the ODE among processors would not scale.
>
>
> Even though it will not scale and will deliver slower performance it is
> completely possible for you to solve the 3 variable problem using 3 MPI
> ranks. Or 10 mpi ranks. You would just create vectors/matrices with 1
> degree of freedom for the first three ranks and no degrees of freedom for
> the later ranks. During your function evaluation (and Jacobian evaluation)
> for TS you will need to set up the appropriate communication to get the
> values you need on each rank to evaluate the parts of the function
> evaluation needed by that rank. This is true for parallelizing any
> computation.
>
> Barry
>
>
>
>
>
> Hong (Mr.)
>
> When I preallocate the space for the Jacobian matrix, is it better to
> decide the local or global space?
>
> Best,
> Francesco
>
>
>
>
> --
> What most experimenters take for granted before they begin their
> experiments is infinitely more interesting than any results to which their
> experiments lead.
> -- Norbert Wiener
>
> https://www.cse.buffalo.edu/~knepley/
>
>
>
>
> --
> What most experimenters take for granted before they begin their
> experiments is infinitely more interesting than any results to which their
> experiments lead.
> -- Norbert Wiener
>
> https://www.cse.buffalo.edu/~knepley/
>
>
>
>
> --
> Stefano
>
>
>
>
> --
> What most experimenters take for granted before they begin their
> experiments is infinitely more interesting than any results to which their
> experiments lead.
> -- Norbert Wiener
>
> https://www.cse.buffalo.edu/~knepley/
>
>
>
>
>
--
Stefano
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20210504/7266a606/attachment-0001.html>
More information about the petsc-users
mailing list