xref: /petsc/src/mat/tests/ex128.c (revision 732aec7a18f2199fb53bb9a2f3aef439a834ce31)

static char help[] = "Tests ILU and ICC factorization with and without matrix ordering on seqsbaij format. Modified from ex30.c\n\
  Input parameters are:\n\
  -lf <level> : level of fill for ILU (default is 0)\n\
  -lu : use full LU or Cholesky factorization\n\
  -m <value>,-n <value> : grid dimensions\n\
Note that most users should employ the KSP interface to the\n\
linear solvers instead of using the factorization routines\n\
directly.\n\n";

#include <petscmat.h>

int main(int argc, char **args)
{
  Mat           C, sC, sA;
  PetscInt      i, j, m = 5, n = 5, Ii, J, lf = 0;
  PetscBool     CHOLESKY = PETSC_FALSE, TRIANGULAR = PETSC_FALSE, flg;
  PetscScalar   v;
  IS            row, col;
  MatFactorInfo info;
  Vec           x, y, b, ytmp;
  PetscReal     norm2, tol = 100 * PETSC_MACHINE_EPSILON;
  PetscRandom   rdm;
  PetscMPIInt   size;

  PetscFunctionBeginUser;
  PetscCall(PetscInitialize(&argc, &args, NULL, help));
  PetscCallMPI(MPI_Comm_size(PETSC_COMM_WORLD, &size));
  PetscCheck(size == 1, PETSC_COMM_WORLD, PETSC_ERR_WRONG_MPI_SIZE, "This is a uniprocessor example only!");
  PetscCall(PetscOptionsGetInt(NULL, NULL, "-m", &m, NULL));
  PetscCall(PetscOptionsGetInt(NULL, NULL, "-n", &n, NULL));
  PetscCall(PetscOptionsGetInt(NULL, NULL, "-lf", &lf, NULL));

  PetscCall(MatCreate(PETSC_COMM_SELF, &C));
  PetscCall(MatSetSizes(C, m * n, m * n, m * n, m * n));
  PetscCall(MatSetFromOptions(C));
  PetscCall(MatSetUp(C));

  /* Create matrix C in seqaij format and sC in seqsbaij. (This is five-point stencil with some extra elements) */
  for (i = 0; i < m; i++) {
    for (j = 0; j < n; j++) {
      v  = -1.0;
      Ii = j + n * i;
      J  = Ii - n;
      if (J >= 0) PetscCall(MatSetValues(C, 1, &Ii, 1, &J, &v, INSERT_VALUES));
      J = Ii + n;
      if (J < m * n) PetscCall(MatSetValues(C, 1, &Ii, 1, &J, &v, INSERT_VALUES));
      J = Ii - 1;
      if (J >= 0) PetscCall(MatSetValues(C, 1, &Ii, 1, &J, &v, INSERT_VALUES));
      J = Ii + 1;
      if (J < m * n) PetscCall(MatSetValues(C, 1, &Ii, 1, &J, &v, INSERT_VALUES));
      v = 4.0;
      PetscCall(MatSetValues(C, 1, &Ii, 1, &Ii, &v, INSERT_VALUES));
    }
  }
  PetscCall(MatAssemblyBegin(C, MAT_FINAL_ASSEMBLY));
  PetscCall(MatAssemblyEnd(C, MAT_FINAL_ASSEMBLY));

  PetscCall(MatIsSymmetric(C, 0.0, &flg));
  PetscCheck(flg, PETSC_COMM_SELF, PETSC_ERR_SUP, "C is non-symmetric");
  PetscCall(MatConvert(C, MATSEQSBAIJ, MAT_INITIAL_MATRIX, &sC));

  /* Create vectors for error checking */
  PetscCall(MatCreateVecs(C, &x, &b));
  PetscCall(VecDuplicate(x, &y));
  PetscCall(VecDuplicate(x, &ytmp));
  PetscCall(PetscRandomCreate(PETSC_COMM_SELF, &rdm));
  PetscCall(PetscRandomSetFromOptions(rdm));
  PetscCall(VecSetRandom(x, rdm));
  PetscCall(MatMult(C, x, b));

  PetscCall(MatGetOrdering(C, MATORDERINGNATURAL, &row, &col));

  /* Compute CHOLESKY or ICC factor sA */
  PetscCall(MatFactorInfoInitialize(&info));

  info.fill          = 1.0;
  info.diagonal_fill = 0;
  info.zeropivot     = 0.0;

  PetscCall(PetscOptionsHasName(NULL, NULL, "-cholesky", &CHOLESKY));
  if (CHOLESKY) {
    PetscCall(PetscPrintf(PETSC_COMM_SELF, "Test CHOLESKY...\n"));
    PetscCall(MatGetFactor(sC, MATSOLVERPETSC, MAT_FACTOR_CHOLESKY, &sA));
    PetscCall(MatCholeskyFactorSymbolic(sA, sC, row, &info));
  } else {
    PetscCall(PetscPrintf(PETSC_COMM_SELF, "Test ICC...\n"));
    info.levels = lf;

    PetscCall(MatGetFactor(sC, MATSOLVERPETSC, MAT_FACTOR_ICC, &sA));
    PetscCall(MatICCFactorSymbolic(sA, sC, row, &info));
  }
  PetscCall(MatCholeskyFactorNumeric(sA, sC, &info));

  /* test MatForwardSolve() and MatBackwardSolve() with matrix reordering on aij matrix C */
  if (CHOLESKY) {
    PetscCall(PetscOptionsHasName(NULL, NULL, "-triangular_solve", &TRIANGULAR));
    if (TRIANGULAR) {
      PetscCall(PetscPrintf(PETSC_COMM_SELF, "Test MatForwardSolve...\n"));
      PetscCall(MatForwardSolve(sA, b, ytmp));
      PetscCall(PetscPrintf(PETSC_COMM_SELF, "Test MatBackwardSolve...\n"));
      PetscCall(MatBackwardSolve(sA, ytmp, y));
      PetscCall(VecAXPY(y, -1.0, x));
      PetscCall(VecNorm(y, NORM_2, &norm2));
      if (norm2 > tol) PetscCall(PetscPrintf(PETSC_COMM_SELF, "MatForwardSolve and BackwardSolve: Norm of error=%g\n", (double)norm2));
    }
  }

  PetscCall(MatSolve(sA, b, y));
  PetscCall(MatDestroy(&sC));
  PetscCall(MatDestroy(&sA));
  PetscCall(VecAXPY(y, -1.0, x));
  PetscCall(VecNorm(y, NORM_2, &norm2));
  if (lf == -1 && norm2 > tol) PetscCall(PetscPrintf(PETSC_COMM_SELF, " reordered SEQAIJ:   Cholesky/ICC levels %" PetscInt_FMT ", residual %g\n", lf, (double)norm2));

  /* Free data structures */
  PetscCall(MatDestroy(&C));
  PetscCall(ISDestroy(&row));
  PetscCall(ISDestroy(&col));
  PetscCall(PetscRandomDestroy(&rdm));
  PetscCall(VecDestroy(&x));
  PetscCall(VecDestroy(&y));
  PetscCall(VecDestroy(&ytmp));
  PetscCall(VecDestroy(&b));
  PetscCall(PetscFinalize());
  return 0;
}

/*TEST

   test:
      output_file: output/ex128.out

   test:
      suffix: 2
      args: -cholesky -triangular_solve

TEST*/