xref: /petsc/src/ksp/pc/tutorials/ex3.c (revision 98d129c30f3ee9fdddc40fdbc5a989b7be64f888)
1 static char help[] = "Test PC redistribute on matrix with load imbalance. \n\
2                       Modified from src/ksp/ksp/tutorials/ex2.c.\n\
3 Input parameters include:\n\
4   -random_exact_sol : use a random exact solution vector\n\
5   -view_exact_sol   : write exact solution vector to stdout\n\
6   -n <mesh_y>       : number of mesh points\n\n";
7 /*
8 Example:
9   mpiexec -n 8 ./ex3 -n 10000 -ksp_type cg -pc_type bjacobi -sub_pc_type icc -ksp_rtol 1.e-8 -log_view
10   mpiexec -n 8 ./ex3 -n 10000 -ksp_type preonly -pc_type redistribute -redistribute_ksp_type cg -redistribute_pc_type bjacobi -redistribute_sub_pc_type icc -redistribute_ksp_rtol 1.e-8 -log_view
11 */
12 
13 #include <petscksp.h>
14 
15 int main(int argc, char **args)
16 {
17   Vec           x, b, u; /* approx solution, RHS, exact solution */
18   Mat           A;       /* linear system matrix */
19   KSP           ksp;     /* linear solver context */
20   PetscRandom   rctx;    /* random number generator context */
21   PetscReal     norm;    /* norm of solution error */
22   PetscInt      i, j, Ii, J, Istart, Iend, m, n = 7, its, nloc, matdistribute = 0;
23   PetscBool     flg = PETSC_FALSE;
24   PetscScalar   v;
25   PetscMPIInt   rank, size;
26   PetscLogStage stage;
27 
28   PetscFunctionBeginUser;
29   PetscCall(PetscInitialize(&argc, &args, (char *)0, help));
30   PetscCallMPI(MPI_Comm_size(PETSC_COMM_WORLD, &size));
31   PetscCallMPI(MPI_Comm_rank(PETSC_COMM_WORLD, &rank));
32   PetscCheck(size > 1, PETSC_COMM_WORLD, PETSC_ERR_WRONG_MPI_SIZE, "This example requires at least 2 MPI processes!");
33 
34   PetscCall(PetscOptionsGetInt(NULL, NULL, "-n", &n, NULL));
35   PetscCall(PetscOptionsGetInt(NULL, NULL, "-matdistribute", &matdistribute, NULL));
36   /* - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
37          Compute the matrix and right-hand-side vector that define
38          the linear system, Ax = b.
39      - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - */
40   switch (matdistribute) {
41   case 1: /* very imbalanced process load for matrix A */
42     m    = (1 + size) * size;
43     nloc = (rank + 1) * n;
44     if (rank == size - 1) { /* proc[size-1] stores all remaining rows */
45       nloc = m * n;
46       for (i = 0; i < size - 1; i++) nloc -= (i + 1) * n;
47     }
48     break;
49   default: /* proc[0] and proc[1] load much smaller row blocks, the rest processes have same loads */
50     if (rank == 0 || rank == 1) {
51       nloc = n;
52     } else {
53       nloc = 10 * n; /* 10x larger load */
54     }
55     m = 2 + (size - 2) * 10;
56     break;
57   }
58   PetscCall(MatCreate(PETSC_COMM_WORLD, &A));
59   PetscCall(MatSetSizes(A, nloc, nloc, PETSC_DECIDE, PETSC_DECIDE));
60   PetscCall(MatSetFromOptions(A));
61   PetscCall(MatMPIAIJSetPreallocation(A, 5, NULL, 5, NULL));
62   PetscCall(MatSeqAIJSetPreallocation(A, 5, NULL));
63   PetscCall(MatSetUp(A));
64 
65   PetscCall(MatGetOwnershipRange(A, &Istart, &Iend));
66   nloc = Iend - Istart;
67   PetscCall(PetscSynchronizedPrintf(PETSC_COMM_WORLD, "[%d] A Istart,Iend: %" PetscInt_FMT " %" PetscInt_FMT "; nloc %" PetscInt_FMT "\n", rank, Istart, Iend, nloc));
68   PetscCall(PetscSynchronizedFlush(PETSC_COMM_WORLD, PETSC_STDOUT));
69 
70   PetscCall(PetscLogStageRegister("Assembly", &stage));
71   PetscCall(PetscLogStagePush(stage));
72   for (Ii = Istart; Ii < Iend; Ii++) {
73     v = -1.0;
74     i = Ii / n;
75     j = Ii - i * n;
76     if (i > 0) {
77       J = Ii - n;
78       PetscCall(MatSetValues(A, 1, &Ii, 1, &J, &v, INSERT_VALUES));
79     }
80     if (i < m - 1) {
81       J = Ii + n;
82       PetscCall(MatSetValues(A, 1, &Ii, 1, &J, &v, INSERT_VALUES));
83     }
84     if (j > 0) {
85       J = Ii - 1;
86       PetscCall(MatSetValues(A, 1, &Ii, 1, &J, &v, INSERT_VALUES));
87     }
88     if (j < n - 1) {
89       J = Ii + 1;
90       PetscCall(MatSetValues(A, 1, &Ii, 1, &J, &v, INSERT_VALUES));
91     }
92     v = 4.0;
93     PetscCall(MatSetValues(A, 1, &Ii, 1, &Ii, &v, INSERT_VALUES));
94   }
95   PetscCall(MatAssemblyBegin(A, MAT_FINAL_ASSEMBLY));
96   PetscCall(MatAssemblyEnd(A, MAT_FINAL_ASSEMBLY));
97   PetscCall(PetscLogStagePop());
98 
99   /* A is symmetric. Set symmetric flag to enable ICC/Cholesky preconditioner */
100   PetscCall(MatSetOption(A, MAT_SYMMETRIC, PETSC_TRUE));
101 
102   /* Create parallel vectors. */
103   PetscCall(VecCreate(PETSC_COMM_WORLD, &u));
104   PetscCall(VecSetSizes(u, nloc, PETSC_DECIDE));
105   PetscCall(VecSetFromOptions(u));
106   PetscCall(VecDuplicate(u, &b));
107   PetscCall(VecDuplicate(b, &x));
108 
109   /* Set exact solution; then compute right-hand-side vector. */
110   PetscCall(PetscOptionsGetBool(NULL, NULL, "-random_exact_sol", &flg, NULL));
111   if (flg) {
112     PetscCall(PetscRandomCreate(PETSC_COMM_WORLD, &rctx));
113     PetscCall(PetscRandomSetFromOptions(rctx));
114     PetscCall(VecSetRandom(u, rctx));
115     PetscCall(PetscRandomDestroy(&rctx));
116   } else {
117     PetscCall(VecSet(u, 1.0));
118   }
119   PetscCall(MatMult(A, u, b));
120 
121   /* View the exact solution vector if desired */
122   flg = PETSC_FALSE;
123   PetscCall(PetscOptionsGetBool(NULL, NULL, "-view_exact_sol", &flg, NULL));
124   if (flg) PetscCall(VecView(u, PETSC_VIEWER_STDOUT_WORLD));
125 
126   /* - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
127                 Create the linear solver and set various options
128      - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - */
129   PetscCall(KSPCreate(PETSC_COMM_WORLD, &ksp));
130   PetscCall(KSPSetOperators(ksp, A, A));
131   PetscCall(KSPSetTolerances(ksp, 1.e-2 / ((m + 1) * (n + 1)), PETSC_DEFAULT, PETSC_DEFAULT, PETSC_DEFAULT));
132   PetscCall(KSPSetFromOptions(ksp));
133 
134   /* - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
135                       Solve the linear system
136      - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - */
137   PetscCall(KSPSolve(ksp, b, x));
138 
139   /* - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
140                       Check solution and clean up
141      - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - */
142   PetscCall(VecAXPY(x, -1.0, u));
143   PetscCall(VecNorm(x, NORM_2, &norm));
144   PetscCall(KSPGetIterationNumber(ksp, &its));
145   PetscCall(PetscPrintf(PETSC_COMM_WORLD, "Norm of error %g iterations %" PetscInt_FMT "\n", (double)norm, its));
146 
147   /* Free work space. */
148   PetscCall(KSPDestroy(&ksp));
149   PetscCall(VecDestroy(&u));
150   PetscCall(VecDestroy(&x));
151   PetscCall(VecDestroy(&b));
152   PetscCall(MatDestroy(&A));
153   PetscCall(PetscFinalize());
154   return 0;
155 }
156 
157 /*TEST
158 
159    test:
160       nsize: 8
161       args: -n 100 -ksp_type cg -pc_type bjacobi -sub_pc_type icc -ksp_rtol 1.e-8
162 
163    test:
164       suffix: 2
165       nsize: 8
166       args: -n 100 -ksp_type preonly -pc_type redistribute -redistribute_ksp_type cg -redistribute_pc_type bjacobi -redistribute_sub_pc_type icc -redistribute_ksp_rtol 1.e-8
167 
168 TEST*/
169