Actual source code: pbjacobi.c

petsc-master 2016-04-27
Report Typos and Errors
  2: /*
  3:    Include files needed for the PBJacobi preconditioner:
  4:      pcimpl.h - private include file intended for use by all preconditioners
  5: */

  7: #include <petsc/private/matimpl.h>
  8: #include <petsc/private/pcimpl.h>   /*I "petscpc.h" I*/

 10: /*
 11:    Private context (data structure) for the PBJacobi preconditioner.
 12: */
 13: typedef struct {
 14:   const MatScalar *diag;
 15:   PetscInt        bs,mbs;
 16: } PC_PBJacobi;


 21: static PetscErrorCode PCApply_PBJacobi_1(PC pc,Vec x,Vec y)
 22: {
 23:   PC_PBJacobi       *jac = (PC_PBJacobi*)pc->data;
 24:   PetscErrorCode    ierr;
 25:   PetscInt          i,m = jac->mbs;
 26:   const MatScalar   *diag = jac->diag;
 27:   const PetscScalar *xx;
 28:   PetscScalar       *yy;

 31:   VecGetArrayRead(x,&xx);
 32:   VecGetArray(y,&yy);
 33:   for (i=0; i<m; i++) yy[i] = diag[i]*xx[i];
 34:   VecRestoreArrayRead(x,&xx);
 35:   VecRestoreArray(y,&yy);
 36:   PetscLogFlops(2.0*m);
 37:   return(0);
 38: }

 42: static PetscErrorCode PCApply_PBJacobi_2(PC pc,Vec x,Vec y)
 43: {
 44:   PC_PBJacobi     *jac = (PC_PBJacobi*)pc->data;
 45:   PetscErrorCode  ierr;
 46:   PetscInt        i,m = jac->mbs;
 47:   const MatScalar *diag = jac->diag;
 48:   PetscScalar     x0,x1,*yy;
 49:   const PetscScalar *xx;

 52:   VecGetArrayRead(x,&xx);
 53:   VecGetArray(y,&yy);
 54:   for (i=0; i<m; i++) {
 55:     x0        = xx[2*i]; x1 = xx[2*i+1];
 56:     yy[2*i]   = diag[0]*x0 + diag[2]*x1;
 57:     yy[2*i+1] = diag[1]*x0 + diag[3]*x1;
 58:     diag     += 4;
 59:   }
 60:   VecRestoreArrayRead(x,&xx);
 61:   VecRestoreArray(y,&yy);
 62:   PetscLogFlops(6.0*m);
 63:   return(0);
 64: }
 67: static PetscErrorCode PCApply_PBJacobi_3(PC pc,Vec x,Vec y)
 68: {
 69:   PC_PBJacobi     *jac = (PC_PBJacobi*)pc->data;
 70:   PetscErrorCode  ierr;
 71:   PetscInt        i,m = jac->mbs;
 72:   const MatScalar *diag = jac->diag;
 73:   PetscScalar     x0,x1,x2,*yy;
 74:   const PetscScalar *xx;

 77:   VecGetArrayRead(x,&xx);
 78:   VecGetArray(y,&yy);
 79:   for (i=0; i<m; i++) {
 80:     x0 = xx[3*i]; x1 = xx[3*i+1]; x2 = xx[3*i+2];

 82:     yy[3*i]   = diag[0]*x0 + diag[3]*x1 + diag[6]*x2;
 83:     yy[3*i+1] = diag[1]*x0 + diag[4]*x1 + diag[7]*x2;
 84:     yy[3*i+2] = diag[2]*x0 + diag[5]*x1 + diag[8]*x2;
 85:     diag     += 9;
 86:   }
 87:   VecRestoreArrayRead(x,&xx);
 88:   VecRestoreArray(y,&yy);
 89:   PetscLogFlops(15.0*m);
 90:   return(0);
 91: }
 94: static PetscErrorCode PCApply_PBJacobi_4(PC pc,Vec x,Vec y)
 95: {
 96:   PC_PBJacobi     *jac = (PC_PBJacobi*)pc->data;
 97:   PetscErrorCode  ierr;
 98:   PetscInt        i,m = jac->mbs;
 99:   const MatScalar *diag = jac->diag;
100:   PetscScalar     x0,x1,x2,x3,*yy;
101:   const PetscScalar *xx;

104:   VecGetArrayRead(x,&xx);
105:   VecGetArray(y,&yy);
106:   for (i=0; i<m; i++) {
107:     x0 = xx[4*i]; x1 = xx[4*i+1]; x2 = xx[4*i+2]; x3 = xx[4*i+3];

109:     yy[4*i]   = diag[0]*x0 + diag[4]*x1 + diag[8]*x2  + diag[12]*x3;
110:     yy[4*i+1] = diag[1]*x0 + diag[5]*x1 + diag[9]*x2  + diag[13]*x3;
111:     yy[4*i+2] = diag[2]*x0 + diag[6]*x1 + diag[10]*x2 + diag[14]*x3;
112:     yy[4*i+3] = diag[3]*x0 + diag[7]*x1 + diag[11]*x2 + diag[15]*x3;
113:     diag     += 16;
114:   }
115:   VecRestoreArrayRead(x,&xx);
116:   VecRestoreArray(y,&yy);
117:   PetscLogFlops(28.0*m);
118:   return(0);
119: }
122: static PetscErrorCode PCApply_PBJacobi_5(PC pc,Vec x,Vec y)
123: {
124:   PC_PBJacobi     *jac = (PC_PBJacobi*)pc->data;
125:   PetscErrorCode  ierr;
126:   PetscInt        i,m = jac->mbs;
127:   const MatScalar *diag = jac->diag;
128:   PetscScalar     x0,x1,x2,x3,x4,*yy;
129:   const PetscScalar *xx;

132:   VecGetArrayRead(x,&xx);
133:   VecGetArray(y,&yy);
134:   for (i=0; i<m; i++) {
135:     x0 = xx[5*i]; x1 = xx[5*i+1]; x2 = xx[5*i+2]; x3 = xx[5*i+3]; x4 = xx[5*i+4];

137:     yy[5*i]   = diag[0]*x0 + diag[5]*x1 + diag[10]*x2  + diag[15]*x3 + diag[20]*x4;
138:     yy[5*i+1] = diag[1]*x0 + diag[6]*x1 + diag[11]*x2  + diag[16]*x3 + diag[21]*x4;
139:     yy[5*i+2] = diag[2]*x0 + diag[7]*x1 + diag[12]*x2 + diag[17]*x3 + diag[22]*x4;
140:     yy[5*i+3] = diag[3]*x0 + diag[8]*x1 + diag[13]*x2 + diag[18]*x3 + diag[23]*x4;
141:     yy[5*i+4] = diag[4]*x0 + diag[9]*x1 + diag[14]*x2 + diag[19]*x3 + diag[24]*x4;
142:     diag     += 25;
143:   }
144:   VecRestoreArrayRead(x,&xx);
145:   VecRestoreArray(y,&yy);
146:   PetscLogFlops(45.0*m);
147:   return(0);
148: }
151: static PetscErrorCode PCApply_PBJacobi_6(PC pc,Vec x,Vec y)
152: {
153:   PC_PBJacobi     *jac = (PC_PBJacobi*)pc->data;
154:   PetscErrorCode  ierr;
155:   PetscInt        i,m = jac->mbs;
156:   const MatScalar *diag = jac->diag;
157:   PetscScalar     x0,x1,x2,x3,x4,x5,*yy;
158:   const PetscScalar *xx;

161:   VecGetArrayRead(x,&xx);
162:   VecGetArray(y,&yy);
163:   for (i=0; i<m; i++) {
164:     x0 = xx[6*i]; x1 = xx[6*i+1]; x2 = xx[6*i+2]; x3 = xx[6*i+3]; x4 = xx[6*i+4]; x5 = xx[6*i+5];

166:     yy[6*i]   = diag[0]*x0 + diag[6]*x1  + diag[12]*x2  + diag[18]*x3 + diag[24]*x4 + diag[30]*x5;
167:     yy[6*i+1] = diag[1]*x0 + diag[7]*x1  + diag[13]*x2  + diag[19]*x3 + diag[25]*x4 + diag[31]*x5;
168:     yy[6*i+2] = diag[2]*x0 + diag[8]*x1  + diag[14]*x2  + diag[20]*x3 + diag[26]*x4 + diag[32]*x5;
169:     yy[6*i+3] = diag[3]*x0 + diag[9]*x1  + diag[15]*x2  + diag[21]*x3 + diag[27]*x4 + diag[33]*x5;
170:     yy[6*i+4] = diag[4]*x0 + diag[10]*x1 + diag[16]*x2  + diag[22]*x3 + diag[28]*x4 + diag[34]*x5;
171:     yy[6*i+5] = diag[5]*x0 + diag[11]*x1 + diag[17]*x2  + diag[23]*x3 + diag[29]*x4 + diag[35]*x5;
172:     diag     += 36;
173:   }
174:   VecRestoreArrayRead(x,&xx);
175:   VecRestoreArray(y,&yy);
176:   PetscLogFlops(66.0*m);
177:   return(0);
178: }
181: static PetscErrorCode PCApply_PBJacobi_7(PC pc,Vec x,Vec y)
182: {
183:   PC_PBJacobi     *jac = (PC_PBJacobi*)pc->data;
184:   PetscErrorCode  ierr;
185:   PetscInt        i,m = jac->mbs;
186:   const MatScalar *diag = jac->diag;
187:   PetscScalar     x0,x1,x2,x3,x4,x5,x6,*yy;
188:   const PetscScalar *xx;

191:   VecGetArrayRead(x,&xx);
192:   VecGetArray(y,&yy);
193:   for (i=0; i<m; i++) {
194:     x0 = xx[7*i]; x1 = xx[7*i+1]; x2 = xx[7*i+2]; x3 = xx[7*i+3]; x4 = xx[7*i+4]; x5 = xx[7*i+5]; x6 = xx[7*i+6];

196:     yy[7*i]   = diag[0]*x0 + diag[7]*x1  + diag[14]*x2  + diag[21]*x3 + diag[28]*x4 + diag[35]*x5 + diag[42]*x6;
197:     yy[7*i+1] = diag[1]*x0 + diag[8]*x1  + diag[15]*x2  + diag[22]*x3 + diag[29]*x4 + diag[36]*x5 + diag[43]*x6;
198:     yy[7*i+2] = diag[2]*x0 + diag[9]*x1  + diag[16]*x2  + diag[23]*x3 + diag[30]*x4 + diag[37]*x5 + diag[44]*x6;
199:     yy[7*i+3] = diag[3]*x0 + diag[10]*x1 + diag[17]*x2  + diag[24]*x3 + diag[31]*x4 + diag[38]*x5 + diag[45]*x6;
200:     yy[7*i+4] = diag[4]*x0 + diag[11]*x1 + diag[18]*x2  + diag[25]*x3 + diag[32]*x4 + diag[39]*x5 + diag[46]*x6;
201:     yy[7*i+5] = diag[5]*x0 + diag[12]*x1 + diag[19]*x2  + diag[26]*x3 + diag[33]*x4 + diag[40]*x5 + diag[47]*x6;
202:     yy[7*i+6] = diag[6]*x0 + diag[13]*x1 + diag[20]*x2  + diag[27]*x3 + diag[34]*x4 + diag[41]*x5 + diag[48]*x6;
203:     diag     += 49;
204:   }
205:   VecRestoreArrayRead(x,&xx);
206:   VecRestoreArray(y,&yy);
207:   PetscLogFlops(80.0*m);
208:   return(0);
209: }
210: /* -------------------------------------------------------------------------- */
213: static PetscErrorCode PCSetUp_PBJacobi(PC pc)
214: {
215:   PC_PBJacobi    *jac = (PC_PBJacobi*)pc->data;
217:   Mat            A = pc->pmat;

220:   if (A->rmap->n != A->cmap->n) SETERRQ(PETSC_COMM_SELF,PETSC_ERR_SUP,"Supported only for square matrices and square storage");

222:   MatInvertBlockDiagonal(A,&jac->diag);
223:   if (A->errortype) pc->failedreason = (PCFailedReason)A->errortype;
224: 
225:   MatGetBlockSize(A,&jac->bs);
226:   jac->mbs = A->rmap->n/jac->bs;
227:   switch (jac->bs) {
228:   case 1:
229:     pc->ops->apply = PCApply_PBJacobi_1;
230:     break;
231:   case 2:
232:     pc->ops->apply = PCApply_PBJacobi_2;
233:     break;
234:   case 3:
235:     pc->ops->apply = PCApply_PBJacobi_3;
236:     break;
237:   case 4:
238:     pc->ops->apply = PCApply_PBJacobi_4;
239:     break;
240:   case 5:
241:     pc->ops->apply = PCApply_PBJacobi_5;
242:     break;
243:   case 6:
244:     pc->ops->apply = PCApply_PBJacobi_6;
245:     break;
246:   case 7:
247:     pc->ops->apply = PCApply_PBJacobi_7;
248:     break;
249:   default:
250:     SETERRQ1(PetscObjectComm((PetscObject)pc),PETSC_ERR_SUP,"not supported for block size %D",jac->bs);
251:   }
252:   return(0);
253: }
254: /* -------------------------------------------------------------------------- */
257: static PetscErrorCode PCDestroy_PBJacobi(PC pc)
258: {

262:   /*
263:       Free the private data structure that was hanging off the PC
264:   */
265:   PetscFree(pc->data);
266:   return(0);
267: }

271: static PetscErrorCode PCView_PBJacobi(PC pc,PetscViewer viewer)
272: {
274:   PC_PBJacobi    *jac = (PC_PBJacobi*)pc->data;
275:   PetscBool      iascii;

278:   PetscObjectTypeCompare((PetscObject)viewer,PETSCVIEWERASCII,&iascii);
279:   if (iascii) {
280:     PetscViewerASCIIPrintf(viewer,"  point-block Jacobi: block size %D\n",jac->bs);
281:   }
282:   return(0);
283: }

285: /* -------------------------------------------------------------------------- */
286: /*MC
287:      PCPBJACOBI - Point block Jacobi preconditioner


290:    Notes: See PCJACOBI for point Jacobi preconditioning

292:    This works for AIJ and BAIJ matrices and uses the blocksize provided to the matrix

294:    Uses dense LU factorization with partial pivoting to invert the blocks; if a zero pivot
295:    is detected a PETSc error is generated.

297:    Developer Notes: This should support the PCSetErrorIfFailure() flag set to PETSC_TRUE to allow
298:    the factorization to continue even after a zero pivot is found resulting in a Nan and hence
299:    terminating KSP with a KSP_DIVERGED_NANORIF allowing
300:    a nonlinear solver/ODE integrator to recover without stopping the program as currently happens.

302:    Developer Note: Perhaps should provide an option that allows generation of a valid preconditioner
303:    even if a block is singular as the PCJACOBI does.

305:    Level: beginner

307:   Concepts: point block Jacobi


310: .seealso:  PCCreate(), PCSetType(), PCType (for list of available types), PC, PCJACOBI

312: M*/

316: PETSC_EXTERN PetscErrorCode PCCreate_PBJacobi(PC pc)
317: {
318:   PC_PBJacobi    *jac;

322:   /*
323:      Creates the private data structure for this preconditioner and
324:      attach it to the PC object.
325:   */
326:   PetscNewLog(pc,&jac);
327:   pc->data = (void*)jac;

329:   /*
330:      Initialize the pointers to vectors to ZERO; these will be used to store
331:      diagonal entries of the matrix for fast preconditioner application.
332:   */
333:   jac->diag = 0;

335:   /*
336:       Set the pointers for the functions that are provided above.
337:       Now when the user-level routines (such as PCApply(), PCDestroy(), etc.)
338:       are called, they will automatically call these functions.  Note we
339:       choose not to provide a couple of these functions since they are
340:       not needed.
341:   */
342:   pc->ops->apply               = 0; /*set depending on the block size */
343:   pc->ops->applytranspose      = 0;
344:   pc->ops->setup               = PCSetUp_PBJacobi;
345:   pc->ops->destroy             = PCDestroy_PBJacobi;
346:   pc->ops->setfromoptions      = 0;
347:   pc->ops->view                = PCView_PBJacobi;
348:   pc->ops->applyrichardson     = 0;
349:   pc->ops->applysymmetricleft  = 0;
350:   pc->ops->applysymmetricright = 0;
351:   return(0);
352: }