Actual source code: ex62f.F90

  1: !
  2: !   Solves a linear system in parallel with KSP.  Also indicates
  3: !   use of a user-provided preconditioner.  Input parameters include:
  4: !
  5: !

  7: !
  8: !  -------------------------------------------------------------------------
  9: #include <petsc/finclude/petscksp.h>
 10: module ex62fmodule
 11:   use petscksp
 12:   implicit none
 13:   PC jacobi, sor
 14:   Vec work

 16: contains
 17: !/***********************************************************************/
 18: !/*          Routines for a user-defined shell preconditioner           */
 19: !/***********************************************************************/

 21: !
 22: !   SampleShellPCSetUp - This routine sets up a user-defined
 23: !   preconditioner context.
 24: !
 25: !   Input Parameters:
 26: !   pc    - preconditioner object
 27: !   x     - vector
 28: !
 29: !   Output Parameter:
 30: !   ierr  - error code (nonzero if error has been detected)
 31: !
 32: !   Notes:
 33: !   In this example, we define the shell preconditioner to be Jacobi
 34: !   method.  Thus, here we create a work vector for storing the reciprocal
 35: !   of the diagonal of the matrix used to compute the preconditioner; this vector is then
 36: !   used within the routine SampleShellPCApply().
 37: !
 38:   subroutine SampleShellPCSetUp(pc, x, ierr)

 40:     PC pc
 41:     Vec x
 42:     Mat pmat
 43:     PetscErrorCode ierr

 45:     PetscCallA(PCGetOperators(pc, PETSC_NULL_MAT, pmat, ierr))
 46:     PetscCallA(PCCreate(PETSC_COMM_WORLD, jacobi, ierr))
 47:     PetscCallA(PCSetType(jacobi, PCJACOBI, ierr))
 48:     PetscCallA(PCSetOperators(jacobi, pmat, pmat, ierr))
 49:     PetscCallA(PCSetUp(jacobi, ierr))

 51:     PetscCallA(PCCreate(PETSC_COMM_WORLD, sor, ierr))
 52:     PetscCallA(PCSetType(sor, PCSOR, ierr))
 53:     PetscCallA(PCSetOperators(sor, pmat, pmat, ierr))
 54: !      PetscCallA(PCSORSetSymmetric(sor,SOR_LOCAL_SYMMETRIC_SWEEP,ierr))
 55:     PetscCallA(PCSetUp(sor, ierr))

 57:     PetscCallA(VecDuplicate(x, work, ierr))
 58:   end

 60: ! -------------------------------------------------------------------
 61: !
 62: !   SampleShellPCApply - This routine demonstrates the use of a
 63: !   user-provided preconditioner.
 64: !
 65: !   Input Parameters:
 66: !   pc - preconditioner object
 67: !   x - input vector
 68: !
 69: !   Output Parameters:
 70: !   y - preconditioned vector
 71: !   ierr  - error code (nonzero if error has been detected)
 72: !
 73: !   Notes:
 74: !   This code implements the Jacobi preconditioner plus the
 75: !   SOR preconditioner
 76: !
 77: ! YOU CAN GET THE EXACT SAME EFFECT WITH THE PCCOMPOSITE preconditioner using
 78: ! mpiexec -n 1 ex21f -ksp_monitor -pc_type composite -pc_composite_pcs jacobi,sor -pc_composite_type additive
 79: !
 80:   subroutine SampleShellPCApply(pc, x, y, ierr)

 82:     PC pc
 83:     Vec x, y
 84:     PetscErrorCode ierr
 85:     PetscScalar one

 87:     one = 1.0
 88:     PetscCallA(PCApply(jacobi, x, y, ierr))
 89:     PetscCallA(PCApply(sor, x, work, ierr))
 90:     PetscCallA(VecAXPY(y, one, work, ierr))
 91:   end

 93: end module

 95: program main
 96:   use ex62fmodule
 97:   implicit none

 99: ! - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
100: !                   Variable declarations
101: ! - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
102: !
103: !  Variables:
104: !     ksp     - linear solver context
105: !     ksp      - Krylov subspace method context
106: !     pc       - preconditioner context
107: !     x, b, u  - approx solution, right-hand side, exact solution vectors
108: !     A        - matrix that defines linear system
109: !     its      - iterations for convergence
110: !     norm     - norm of solution error

112:   Vec x, b, u
113:   Mat A
114:   PC pc
115:   KSP ksp
116:   PetscScalar v, one, neg_one
117:   PetscReal norm, tol
118:   PetscInt i, j, II, JJ, Istart
119:   PetscInt Iend, m, n, its, ione
120:   PetscMPIInt rank
121:   PetscBool flg
122:   PetscErrorCode ierr

124: ! - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
125: !                 Beginning of program
126: ! - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

128:   PetscCallA(PetscInitialize(ierr))
129:   one = 1.0
130:   neg_one = -1.0
131:   m = 8
132:   n = 7
133:   ione = 1
134:   PetscCallA(PetscOptionsGetInt(PETSC_NULL_OPTIONS, PETSC_NULL_CHARACTER, '-m', m, flg, ierr))
135:   PetscCallA(PetscOptionsGetInt(PETSC_NULL_OPTIONS, PETSC_NULL_CHARACTER, '-n', n, flg, ierr))
136:   PetscCallMPIA(MPI_Comm_rank(PETSC_COMM_WORLD, rank, ierr))

138: ! - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
139: !      Compute the matrix and right-hand-side vector that define
140: !      the linear system, Ax = b.
141: ! - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

143: !  Create parallel matrix, specifying only its global dimensions.
144: !  When using MatCreate(), the matrix format can be specified at
145: !  runtime. Also, the parallel partitioning of the matrix is
146: !  determined by PETSc at runtime.

148:   PetscCallA(MatCreate(PETSC_COMM_WORLD, A, ierr))
149:   PetscCallA(MatSetSizes(A, PETSC_DECIDE, PETSC_DECIDE, m*n, m*n, ierr))
150:   PetscCallA(MatSetFromOptions(A, ierr))
151:   PetscCallA(MatSetUp(A, ierr))

153: !  Currently, all PETSc parallel matrix formats are partitioned by
154: !  contiguous chunks of rows across the processors.  Determine which
155: !  rows of the matrix are locally owned.

157:   PetscCallA(MatGetOwnershipRange(A, Istart, Iend, ierr))

159: !  Set matrix elements for the 2-D, five-point stencil in parallel.
160: !   - Each processor needs to insert only elements that it owns
161: !     locally (but any non-local elements will be sent to the
162: !     appropriate processor during matrix assembly).
163: !   - Always specify global row and columns of matrix entries.
164: !   - Note that MatSetValues() uses 0-based row and column numbers
165: !     in Fortran as well as in C.

167:   do II = Istart, Iend - 1
168:     v = -1.0
169:     i = II/n
170:     j = II - i*n
171:     if (i > 0) then
172:       JJ = II - n
173:       PetscCallA(MatSetValues(A, ione, [II], ione, [JJ], [v], ADD_VALUES, ierr))
174:     end if
175:     if (i < m - 1) then
176:       JJ = II + n
177:       PetscCallA(MatSetValues(A, ione, [II], ione, [JJ], [v], ADD_VALUES, ierr))
178:     end if
179:     if (j > 0) then
180:       JJ = II - 1
181:       PetscCallA(MatSetValues(A, ione, [II], ione, [JJ], [v], ADD_VALUES, ierr))
182:     end if
183:     if (j < n - 1) then
184:       JJ = II + 1
185:       PetscCallA(MatSetValues(A, ione, [II], ione, [JJ], [v], ADD_VALUES, ierr))
186:     end if
187:     v = 4.0
188:     PetscCallA(MatSetValues(A, ione, [II], ione, [II], [v], ADD_VALUES, ierr))
189:   end do

191: !  Assemble matrix, using the 2-step process:
192: !       MatAssemblyBegin(), MatAssemblyEnd()
193: !  Computations can be done while messages are in transition,
194: !  by placing code between these two statements.

196:   PetscCallA(MatAssemblyBegin(A, MAT_FINAL_ASSEMBLY, ierr))
197:   PetscCallA(MatAssemblyEnd(A, MAT_FINAL_ASSEMBLY, ierr))

199: !  Create parallel vectors.
200: !   - Here, the parallel partitioning of the vector is determined by
201: !     PETSc at runtime.  We could also specify the local dimensions
202: !     if desired -- or use the more general routine VecCreate().
203: !   - When solving a linear system, the vectors and matrices MUST
204: !     be partitioned accordingly.  PETSc automatically generates
205: !     appropriately partitioned matrices and vectors when MatCreate()
206: !     and VecCreate() are used with the same communicator.
207: !   - Note: We form 1 vector from scratch and then duplicate as needed.

209:   PetscCallA(VecCreateFromOptions(PETSC_COMM_WORLD, PETSC_NULL_CHARACTER, ione, PETSC_DECIDE, m*n, u, ierr))
210:   PetscCallA(VecDuplicate(u, b, ierr))
211:   PetscCallA(VecDuplicate(b, x, ierr))

213: !  Set exact solution; then compute right-hand-side vector.

215:   PetscCallA(VecSet(u, one, ierr))
216:   PetscCallA(MatMult(A, u, b, ierr))

218: ! - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
219: !         Create the linear solver and set various options
220: ! - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

222: !  Create linear solver context

224:   PetscCallA(KSPCreate(PETSC_COMM_WORLD, ksp, ierr))

226: !  Set operators. Here the matrix that defines the linear system
227: !  also serves as the matrix from which the preconditioner is constructed.

229:   PetscCallA(KSPSetOperators(ksp, A, A, ierr))

231: !  Set linear solver defaults for this problem (optional).
232: !   - By extracting the KSP and PC contexts from the KSP context,
233: !     we can then directly call any KSP and PC routines
234: !     to set various options.

236:   PetscCallA(KSPGetPC(ksp, pc, ierr))
237:   tol = 1.e-7
238:   PetscCallA(KSPSetTolerances(ksp, tol, PETSC_CURRENT_REAL, PETSC_CURRENT_REAL, PETSC_CURRENT_INTEGER, ierr))

240: !
241: !  Set a user-defined shell preconditioner
242: !

244: !  (Required) Indicate to PETSc that we are using a shell preconditioner
245:   PetscCallA(PCSetType(pc, PCSHELL, ierr))

247: !  (Required) Set the user-defined routine for applying the preconditioner
248:   PetscCallA(PCShellSetApply(pc, SampleShellPCApply, ierr))

250: !  (Optional) Do any setup required for the preconditioner
251: !     Note: if you use PCShellSetSetUp, this will be done for your
252:   PetscCallA(SampleShellPCSetUp(pc, x, ierr))

254: !  Set runtime options, e.g.,
255: !      -ksp_type <type> -pc_type <type> -ksp_monitor -ksp_rtol <rtol>
256: !  These options will override those specified above as long as
257: !  KSPSetFromOptions() is called _after_ any other customization
258: !  routines.

260:   PetscCallA(KSPSetFromOptions(ksp, ierr))

262: ! - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
263: !                      Solve the linear system
264: ! - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

266:   PetscCallA(KSPSolve(ksp, b, x, ierr))

268: ! - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
269: !                     Check solution and clean up
270: ! - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

272: !  Check the error

274:   PetscCallA(VecAXPY(x, neg_one, u, ierr))
275:   PetscCallA(VecNorm(x, NORM_2, norm, ierr))
276:   PetscCallA(KSPGetIterationNumber(ksp, its, ierr))

278:   if (rank == 0) then
279:     if (norm > 1.e-12) then
280:       write (6, 100) norm, its
281:     else
282:       write (6, 110) its
283:     end if
284:   end if
285: 100 format('Norm of error ', 1pe11.4, ' iterations ', i5)
286: 110 format('Norm of error < 1.e-12,iterations ', i5)

288: !  Free work space.  All PETSc objects should be destroyed when they
289: !  are no longer needed.

291:   PetscCallA(KSPDestroy(ksp, ierr))
292:   PetscCallA(VecDestroy(u, ierr))
293:   PetscCallA(VecDestroy(x, ierr))
294:   PetscCallA(VecDestroy(b, ierr))
295:   PetscCallA(MatDestroy(A, ierr))

297: ! Free up PCShell data
298:   PetscCallA(PCDestroy(sor, ierr))
299:   PetscCallA(PCDestroy(jacobi, ierr))
300:   PetscCallA(VecDestroy(work, ierr))

302: !  Always call PetscFinalize() before exiting a program.

304:   PetscCallA(PetscFinalize(ierr))
305: end

307: !/*TEST
308: !
309: !   test:
310: !     requires: !single
311: !
312: !TEST*/