[petsc-users] Loading a PETsc matrix with matrix data in CSR format?

The matrix import function looks like this: void csr2pet (     const Foam::lduMatrix & matrix,     petsc_declaration & petsc_matrix  // How to declare the PETsc matrix to be filled? ) {     int n = matrix.diag().size(); // small case n = 40800     int nnz = matrix.lower().size() + matrix.upper().

Re: [petsc-users] Loading a PETsc matrix with matrix data in CSR format?

then saves it with MatView() and a binary viewer. You can then load the matrix easily and efficiently in PETSc in parallel with MatLoad   3) If you have matrix already in CSR format you can use MatCreateSeqAIJWithArrays()   Barry > On Sep 17, 2017, at 9:25 AM, Klaus Burkart wrote:

[petsc-users] Which matrix object type should I use?

I am confused with matrix object types and which one to use in my case: I want to solve a linear system using multiple processes. The steps I am struggling with is the import of the global matrix and global rhs vector into applicable PETSc objects. The global matrix is currently stored in CSR

[petsc-users] preallocate space for matrices if only the global number of nonzeros is known

Hi, How can I preallocate space for matrices (some symmetric, others asymmetric) if I have the global number of nonzeros (NNZ) but not the number of nonzeros per row? I could compute the NNZ for the upper or lower part separately if this would be useful for symmetric matrices. I create the mat

Re: [petsc-users] Matrix and vector type selection & memory allocation for efficient matrix import?

gt; On Apr 20, 2018, at 10:05 AM, Klaus Burkart wrote: > > I think I understood the matrix structure for parallel computation with the > rows, diagonal (d) and off-diagonal (o) structure, where I have problems is > how to do the setup including memory allocation in PETSc: >

[petsc-users] Can't initialize petsc from within application code

Hello, I am trying to initialize petsc from within application code but I keep getting the following error. symbol lookup error: /home/klaus/OpenFOAM/klaus-5.0/platforms/linux64GccDPInt32Opt/lib/libmyPCG2.so: undefined symbol: PetscInitialize The code is just:     PetscInitialize(0,0,PETSC_N

[petsc-users] Incorrect local row ranges allocated to the processes i.e. rstart and rend are not what I expected

Hello, I am trying to compute the local row ranges allocated to the processes i.e. rstart and rend of each process, needed as a  prerequisite for MatMPIAIJSetPreallocation using d_nnz and o_nnz. I tried the following: ...     PetscInitialize(0,0,PETSC_NULL,PETSC_NULL);     MPI_Comm_size(PETS

Re: [petsc-users] Incorrect local row ranges allocated to the processes i.e. rstart and rend are not what I expected

The output is: local_size   =  25, on process 0 rstart =  0, on process 0 rend =  25, on process 0 local_size   =  25, on process 1 rstart =  0, on process 1 rend =  25, on process 1 local_size   =  25, on process 2 rstart =  0, on process 2 rend =  25, on process 2 local_size   =  25, on process

Re: [petsc-users] Incorrect local row ranges allocated to the processes i.e. rstart and rend are not what I expected

e PetscSynchronizedPrintf() to have each process print its own values.     Barry > On Nov 23, 2018, at 6:44 AM, Klaus Burkart via petsc-users > wrote: > > Hello, > > I am trying to compute the local row ranges allocated to the processes i.e. > rstart and rend of each

[petsc-users] How can I update a global application vector with local petsc results?

Hello, I use the following routine to transfer the petsc result back to my application which works fine using one core as the local parts of the linear system == global parts:     VecGetArray(petsc_x, &array);     for (i=rstart; i < rend; ++i) {     application_x[i] = array[i];     }    

[petsc-users] How to solve multiple linear systems in parallel, one on each process?

Hello, I want to solve a cfd case, after decomposition, I get a sub matrix allocated to each process. The example below shows how the data is allocated to the processes (the sample data includes only the lower parts of the matrices). Row and column addresses are local. What petsc program setup

[petsc-users] Is it possible to start MPI with n processes from within the code so it takes only effect on the petsc code?

Hello, I try to integrate petsc into an application and I think it would be much simpler if I could bypass the applications original MPI functionality by starting MPI with n processes when initializing Petsc and stopping it when PetscFinalize(); is called. The standard  mpirun -np 4 application

Re: [petsc-users] How to solve multiple linear systems in parallel, one on each process?

8, 15:12:58 MEZ hat Dave May Folgendes geschrieben: On Mon, 3 Dec 2018 at 13:52, Klaus Burkart via petsc-users wrote: Hello, I want to solve a cfd case, after decomposition, I get a sub matrix allocated to each process. The example below shows how the data is allocated to the processes

[petsc-users] What's the correct syntax for PetscOptionsSetValue("-pc_hypre_boomeramg_no_CF","true"); ?

Hello, I try to use the Hypre boomeramg preconditioner but I keep getting type conversion errors like this one for PetscOptionsSetValue(...): Fehler: »const char*« kann nicht nach »PetscOptions« {aka »_n_PetscOptions*«} umgewandelt werden PetscOptionsSetValue("-pc_hypre_boomeramg_no_CF","tru

[petsc-users] Loading only upper + MatSetOption(A,MAT_SYMMETRIC,PETSC_TRUE);

Hello, I want to solve many symmetric linear systems one after another in parallel using boomerAMG + KSPCG  and need to make the matrix transfer more efficient. Matrices are symmetric in structure and values. boomerAMG + KSPCG work fine. So far I have been loading the entire matrices but I read