xref: /petsc/src/mat/tests/ex121.c (revision 834855d6effb0d027771461c8e947ee1ce5a1e17)

static char help[] = "Test sequential FFTW convolution\n\n";

/*
  Compiling the code:
    This code uses the complex numbers, so configure must be given --with-scalar-type=complex to enable this
*/

#include <petscmat.h>

int main(int argc, char **args)
{
  typedef enum {
    RANDOM,
    CONSTANT,
    TANH,
    NUM_FUNCS
  } FuncType;
  const char  *funcNames[NUM_FUNCS] = {"random", "constant", "tanh"};
  Mat          A;
  PetscMPIInt  size;
  PetscInt     n = 10, N, ndim = 4, dim[4], DIM, i, j;
  Vec          w, x, y1, y2, z1, z2;
  PetscScalar *a, *a2, *a3;
  PetscScalar  s;
  PetscRandom  rdm;
  PetscReal    enorm;
  PetscInt     func     = 0;
  FuncType     function = RANDOM;
  PetscBool    view     = PETSC_FALSE;

  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!");
  PetscOptionsBegin(PETSC_COMM_WORLD, NULL, "FFTW Options", "ex112");
  PetscCall(PetscOptionsEList("-function", "Function type", "ex121", funcNames, NUM_FUNCS, funcNames[function], &func, NULL));
  PetscCall(PetscOptionsBool("-vec_view draw", "View the functions", "ex112", view, &view, NULL));
  function = (FuncType)func;
  PetscOptionsEnd();

  for (DIM = 0; DIM < ndim; DIM++) dim[DIM] = n; /* size of transformation in DIM-dimension */
  PetscCall(PetscRandomCreate(PETSC_COMM_SELF, &rdm));
  PetscCall(PetscRandomSetFromOptions(rdm));

  for (DIM = 1; DIM < 5; DIM++) {
    /* create vectors of length N=n^DIM */
    for (i = 0, N = 1; i < DIM; i++) N *= dim[i];
    PetscCall(PetscPrintf(PETSC_COMM_SELF, "\n %d-D: FFTW on vector of size %d \n", DIM, N));
    PetscCall(VecCreateSeq(PETSC_COMM_SELF, N, &x));
    PetscCall(PetscObjectSetName((PetscObject)x, "Real space vector"));
    PetscCall(VecDuplicate(x, &w));
    PetscCall(PetscObjectSetName((PetscObject)w, "Window vector"));
    PetscCall(VecDuplicate(x, &y1));
    PetscCall(PetscObjectSetName((PetscObject)y1, "Frequency space vector"));
    PetscCall(VecDuplicate(x, &y2));
    PetscCall(PetscObjectSetName((PetscObject)y2, "Frequency space window vector"));
    PetscCall(VecDuplicate(x, &z1));
    PetscCall(PetscObjectSetName((PetscObject)z1, "Reconstructed convolution"));
    PetscCall(VecDuplicate(x, &z2));
    PetscCall(PetscObjectSetName((PetscObject)z2, "Real space convolution"));

    if (function == RANDOM) {
      PetscCall(VecSetRandom(x, rdm));
    } else if (function == CONSTANT) {
      PetscCall(VecSet(x, 1.0));
    } else if (function == TANH) {
      PetscCall(VecGetArray(x, &a));
      for (i = 0; i < N; ++i) a[i] = tanh((i - N / 2.0) * (10.0 / N));
      PetscCall(VecRestoreArray(x, &a));
    }
    if (view) PetscCall(VecView(x, PETSC_VIEWER_DRAW_WORLD));

    /* Create window function */
    PetscCall(VecGetArray(w, &a));
    for (i = 0; i < N; ++i) {
      /* Step Function */
      a[i] = (i > N / 4 && i < 3 * N / 4) ? 1.0 : 0.0;
      /* Delta Function */
      /*a[i] = (i == N/2)? 1.0: 0.0; */
    }
    PetscCall(VecRestoreArray(w, &a));
    if (view) PetscCall(VecView(w, PETSC_VIEWER_DRAW_WORLD));

    /* create FFTW object */
    PetscCall(MatCreateFFT(PETSC_COMM_SELF, DIM, dim, MATFFTW, &A));

    /* Convolve x with w*/
    PetscCall(MatMult(A, x, y1));
    PetscCall(MatMult(A, w, y2));
    PetscCall(VecPointwiseMult(y1, y1, y2));
    if (view && i == 0) PetscCall(VecView(y1, PETSC_VIEWER_DRAW_WORLD));
    PetscCall(MatMultTranspose(A, y1, z1));

    /* Compute the real space convolution */
    PetscCall(VecGetArray(x, &a));
    PetscCall(VecGetArray(w, &a2));
    PetscCall(VecGetArray(z2, &a3));
    for (i = 0; i < N; ++i) {
      /* PetscInt checkInd = (i > N/2-1)? i-N/2: i+N/2;*/

      a3[i] = 0.0;
      for (j = -N / 2 + 1; j < N / 2; ++j) {
        PetscInt xpInd   = (j < 0) ? N + j : j;
        PetscInt diffInd = (i - j < 0) ? N - (j - i) : (i - j > N - 1) ? i - j - N : i - j;

        a3[i] += a[xpInd] * a2[diffInd];
      }
    }
    PetscCall(VecRestoreArray(x, &a));
    PetscCall(VecRestoreArray(w, &a2));
    PetscCall(VecRestoreArray(z2, &a3));

    /* compare z1 and z2. FFTW computes an unnormalized DFT, thus z1 = N*z2 */
    s = 1.0 / (PetscReal)N;
    PetscCall(VecScale(z1, s));
    if (view) PetscCall(VecView(z1, PETSC_VIEWER_DRAW_WORLD));
    if (view) PetscCall(VecView(z2, PETSC_VIEWER_DRAW_WORLD));
    PetscCall(VecAXPY(z1, -1.0, z2));
    PetscCall(VecNorm(z1, NORM_1, &enorm));
    if (enorm > 1.e-11) PetscCall(PetscPrintf(PETSC_COMM_SELF, "  Error norm of |z1 - z2| %g\n", (double)enorm));

    /* free spaces */
    PetscCall(VecDestroy(&x));
    PetscCall(VecDestroy(&y1));
    PetscCall(VecDestroy(&y2));
    PetscCall(VecDestroy(&z1));
    PetscCall(VecDestroy(&z2));
    PetscCall(VecDestroy(&w));
    PetscCall(MatDestroy(&A));
  }
  PetscCall(PetscRandomDestroy(&rdm));
  PetscCall(PetscFinalize());
  return 0;
}

/*TEST

   build:
      requires: fftw complex

   test:
      output_file: output/ex121.out
      TODO: Example or FFTW interface is broken

TEST*/