Actual source code: redistribute.c

petsc-3.4.5 2014-06-29
  2: /*
  3:   This file defines a "solve the problem redistributely on each subgroup of processor" preconditioner.
  4: */
  5: #include <petsc-private/pcimpl.h>     /*I "petscksp.h" I*/
  6: #include <petscksp.h>

  8: typedef struct {
  9:   KSP         ksp;
 10:   Vec         x,b;
 11:   VecScatter  scatter;
 12:   IS          is;
 13:   PetscInt    dcnt,*drows;    /* these are the local rows that have only diagonal entry */
 14:   PetscScalar *diag;
 15:   Vec         work;
 16: } PC_Redistribute;

 20: static PetscErrorCode PCView_Redistribute(PC pc,PetscViewer viewer)
 21: {
 22:   PC_Redistribute *red = (PC_Redistribute*)pc->data;
 23:   PetscErrorCode  ierr;
 24:   PetscBool       iascii,isstring;
 25:   PetscInt        ncnt,N;

 28:   PetscObjectTypeCompare((PetscObject)viewer,PETSCVIEWERASCII,&iascii);
 29:   PetscObjectTypeCompare((PetscObject)viewer,PETSCVIEWERSTRING,&isstring);
 30:   if (iascii) {
 31:     MPI_Allreduce(&red->dcnt,&ncnt,1,MPIU_INT,MPI_SUM,PetscObjectComm((PetscObject)pc));
 32:     MatGetSize(pc->pmat,&N,NULL);
 33:     PetscViewerASCIIPrintf(viewer,"    Number rows eliminated %D Percentage rows eliminated %g\n",ncnt,100.0*((PetscReal)ncnt)/((PetscReal)N));
 34:     PetscViewerASCIIPrintf(viewer,"  Redistribute preconditioner: \n");
 35:     KSPView(red->ksp,viewer);
 36:   } else if (isstring) {
 37:     PetscViewerStringSPrintf(viewer," Redistribute preconditioner");
 38:     KSPView(red->ksp,viewer);
 39:   }
 40:   return(0);
 41: }

 45: static PetscErrorCode PCSetUp_Redistribute(PC pc)
 46: {
 47:   PC_Redistribute   *red = (PC_Redistribute*)pc->data;
 48:   PetscErrorCode    ierr;
 49:   MPI_Comm          comm;
 50:   PetscInt          rstart,rend,i,nz,cnt,*rows,ncnt,dcnt,*drows;
 51:   PetscLayout       map,nmap;
 52:   PetscMPIInt       size,imdex,tag,n;
 53:   PetscInt          *source = NULL;
 54:   PetscMPIInt       *nprocs = NULL,nrecvs;
 55:   PetscInt          j,nsends;
 56:   PetscInt          *owner = NULL,*starts = NULL,count,slen;
 57:   PetscInt          *rvalues,*svalues,recvtotal;
 58:   PetscMPIInt       *onodes1,*olengths1;
 59:   MPI_Request       *send_waits = NULL,*recv_waits = NULL;
 60:   MPI_Status        recv_status,*send_status;
 61:   Vec               tvec,diag;
 62:   Mat               tmat;
 63:   const PetscScalar *d;

 66:   if (pc->setupcalled) {
 67:     KSPGetOperators(red->ksp,NULL,&tmat,NULL);
 68:     MatGetSubMatrix(pc->pmat,red->is,red->is,MAT_REUSE_MATRIX,&tmat);
 69:     KSPSetOperators(red->ksp,tmat,tmat,SAME_NONZERO_PATTERN);
 70:   } else {
 71:     PetscInt NN;

 73:     PetscObjectGetComm((PetscObject)pc,&comm);
 74:     MPI_Comm_size(comm,&size);
 75:     PetscObjectGetNewTag((PetscObject)pc,&tag);

 77:     /* count non-diagonal rows on process */
 78:     MatGetOwnershipRange(pc->mat,&rstart,&rend);
 79:     cnt  = 0;
 80:     for (i=rstart; i<rend; i++) {
 81:       MatGetRow(pc->mat,i,&nz,NULL,NULL);
 82:       if (nz > 1) cnt++;
 83:       MatRestoreRow(pc->mat,i,&nz,NULL,NULL);
 84:     }
 85:     PetscMalloc(cnt*sizeof(PetscInt),&rows);
 86:     PetscMalloc((rend - rstart - cnt)*sizeof(PetscInt),&drows);

 88:     /* list non-diagonal rows on process */
 89:     cnt = 0; dcnt = 0;
 90:     for (i=rstart; i<rend; i++) {
 91:       MatGetRow(pc->mat,i,&nz,NULL,NULL);
 92:       if (nz > 1) rows[cnt++] = i;
 93:       else drows[dcnt++] = i - rstart;
 94:       MatRestoreRow(pc->mat,i,&nz,NULL,NULL);
 95:     }

 97:     /* create PetscLayout for non-diagonal rows on each process */
 98:     PetscLayoutCreate(comm,&map);
 99:     PetscLayoutSetLocalSize(map,cnt);
100:     PetscLayoutSetBlockSize(map,1);
101:     PetscLayoutSetUp(map);
102:     rstart = map->rstart;
103:     rend   = map->rend;

105:     /* create PetscLayout for load-balanced non-diagonal rows on each process */
106:     PetscLayoutCreate(comm,&nmap);
107:     MPI_Allreduce(&cnt,&ncnt,1,MPIU_INT,MPI_SUM,comm);
108:     PetscLayoutSetSize(nmap,ncnt);
109:     PetscLayoutSetBlockSize(nmap,1);
110:     PetscLayoutSetUp(nmap);

112:     MatGetSize(pc->pmat,&NN,NULL);
113:     PetscInfo2(pc,"Number of diagonal rows eliminated %d, percentage eliminated %g\n",NN-ncnt,((PetscReal)(NN-ncnt))/((PetscReal)(NN)));
114:     /*
115:         this code is taken from VecScatterCreate_PtoS()
116:         Determines what rows need to be moved where to
117:         load balance the non-diagonal rows
118:     */
119:     /*  count number of contributors to each processor */
120:     PetscMalloc2(size,PetscMPIInt,&nprocs,cnt,PetscInt,&owner);
121:     PetscMemzero(nprocs,size*sizeof(PetscMPIInt));
122:     j      = 0;
123:     nsends = 0;
124:     for (i=rstart; i<rend; i++) {
125:       if (i < nmap->range[j]) j = 0;
126:       for (; j<size; j++) {
127:         if (i < nmap->range[j+1]) {
128:           if (!nprocs[j]++) nsends++;
129:           owner[i-rstart] = j;
130:           break;
131:         }
132:       }
133:     }
134:     /* inform other processors of number of messages and max length*/
135:     PetscGatherNumberOfMessages(comm,NULL,nprocs,&nrecvs);
136:     PetscGatherMessageLengths(comm,nsends,nrecvs,nprocs,&onodes1,&olengths1);
137:     PetscSortMPIIntWithArray(nrecvs,onodes1,olengths1);
138:     recvtotal = 0; for (i=0; i<nrecvs; i++) recvtotal += olengths1[i];

140:     /* post receives:  rvalues - rows I will own; count - nu */
141:     PetscMalloc3(recvtotal,PetscInt,&rvalues,nrecvs,PetscInt,&source,nrecvs,MPI_Request,&recv_waits);
142:     count = 0;
143:     for (i=0; i<nrecvs; i++) {
144:       MPI_Irecv((rvalues+count),olengths1[i],MPIU_INT,onodes1[i],tag,comm,recv_waits+i);
145:       count += olengths1[i];
146:     }

148:     /* do sends:
149:        1) starts[i] gives the starting index in svalues for stuff going to
150:        the ith processor
151:     */
152:     PetscMalloc3(cnt,PetscInt,&svalues,nsends,MPI_Request,&send_waits,size,PetscInt,&starts);
153:     starts[0] = 0;
154:     for (i=1; i<size; i++) starts[i] = starts[i-1] + nprocs[i-1];
155:     for (i=0; i<cnt; i++)  svalues[starts[owner[i]]++] = rows[i];
156:     for (i=0; i<cnt; i++)  rows[i] = rows[i] - rstart;
157:     red->drows = drows;
158:     red->dcnt  = dcnt;
159:     PetscFree(rows);

161:     starts[0] = 0;
162:     for (i=1; i<size; i++) starts[i] = starts[i-1] + nprocs[i-1];
163:     count = 0;
164:     for (i=0; i<size; i++) {
165:       if (nprocs[i]) {
166:         MPI_Isend(svalues+starts[i],nprocs[i],MPIU_INT,i,tag,comm,send_waits+count++);
167:       }
168:     }

170:     /*  wait on receives */
171:     count = nrecvs;
172:     slen  = 0;
173:     while (count) {
174:       MPI_Waitany(nrecvs,recv_waits,&imdex,&recv_status);
175:       /* unpack receives into our local space */
176:       MPI_Get_count(&recv_status,MPIU_INT,&n);
177:       slen += n;
178:       count--;
179:     }
180:     if (slen != recvtotal) SETERRQ2(PETSC_COMM_SELF,PETSC_ERR_PLIB,"Total message lengths %D not expected %D",slen,recvtotal);

182:     ISCreateGeneral(comm,slen,rvalues,PETSC_COPY_VALUES,&red->is);

184:     /* free up all work space */
185:     PetscFree(olengths1);
186:     PetscFree(onodes1);
187:     PetscFree3(rvalues,source,recv_waits);
188:     PetscFree2(nprocs,owner);
189:     if (nsends) {   /* wait on sends */
190:       PetscMalloc(nsends*sizeof(MPI_Status),&send_status);
191:       MPI_Waitall(nsends,send_waits,send_status);
192:       PetscFree(send_status);
193:     }
194:     PetscFree3(svalues,send_waits,starts);
195:     PetscLayoutDestroy(&map);
196:     PetscLayoutDestroy(&nmap);

198:     VecCreateMPI(comm,slen,PETSC_DETERMINE,&red->b);
199:     VecDuplicate(red->b,&red->x);
200:     MatGetVecs(pc->pmat,&tvec,NULL);
201:     VecScatterCreate(tvec,red->is,red->b,NULL,&red->scatter);
202:     VecDestroy(&tvec);
203:     MatGetSubMatrix(pc->pmat,red->is,red->is,MAT_INITIAL_MATRIX,&tmat);
204:     KSPSetOperators(red->ksp,tmat,tmat,SAME_NONZERO_PATTERN);
205:     MatDestroy(&tmat);
206:   }

208:   /* get diagonal portion of matrix */
209:   PetscMalloc(red->dcnt*sizeof(PetscScalar),&red->diag);
210:   MatGetVecs(pc->pmat,&diag,NULL);
211:   MatGetDiagonal(pc->pmat,diag);
212:   VecGetArrayRead(diag,&d);
213:   for (i=0; i<red->dcnt; i++) red->diag[i] = 1.0/d[red->drows[i]];
214:   VecRestoreArrayRead(diag,&d);
215:   VecDestroy(&diag);
216:   KSPSetUp(red->ksp);
217:   return(0);
218: }

222: static PetscErrorCode PCApply_Redistribute(PC pc,Vec b,Vec x)
223: {
224:   PC_Redistribute   *red = (PC_Redistribute*)pc->data;
225:   PetscErrorCode    ierr;
226:   PetscInt          dcnt   = red->dcnt,i;
227:   const PetscInt    *drows = red->drows;
228:   PetscScalar       *xwork;
229:   const PetscScalar *bwork,*diag = red->diag;

232:   if (!red->work) {
233:     VecDuplicate(b,&red->work);
234:   }
235:   /* compute the rows of solution that have diagonal entries only */
236:   VecSet(x,0.0);         /* x = diag(A)^{-1} b */
237:   VecGetArray(x,&xwork);
238:   VecGetArrayRead(b,&bwork);
239:   for (i=0; i<dcnt; i++) xwork[drows[i]] = diag[i]*bwork[drows[i]];
240:   PetscLogFlops(dcnt);
241:   VecRestoreArray(red->work,&xwork);
242:   VecRestoreArrayRead(b,&bwork);
243:   /* update the right hand side for the reduced system with diagonal rows (and corresponding columns) removed */
244:   MatMult(pc->pmat,x,red->work);
245:   VecAYPX(red->work,-1.0,b);   /* red->work = b - A x */

247:   VecScatterBegin(red->scatter,red->work,red->b,INSERT_VALUES,SCATTER_FORWARD);
248:   VecScatterEnd(red->scatter,red->work,red->b,INSERT_VALUES,SCATTER_FORWARD);
249:   KSPSolve(red->ksp,red->b,red->x);
250:   VecScatterBegin(red->scatter,red->x,x,INSERT_VALUES,SCATTER_REVERSE);
251:   VecScatterEnd(red->scatter,red->x,x,INSERT_VALUES,SCATTER_REVERSE);
252:   return(0);
253: }

257: static PetscErrorCode PCDestroy_Redistribute(PC pc)
258: {
259:   PC_Redistribute *red = (PC_Redistribute*)pc->data;
260:   PetscErrorCode  ierr;

263:   VecScatterDestroy(&red->scatter);
264:   ISDestroy(&red->is);
265:   VecDestroy(&red->b);
266:   VecDestroy(&red->x);
267:   KSPDestroy(&red->ksp);
268:   VecDestroy(&red->work);
269:   PetscFree(red->drows);
270:   PetscFree(red->diag);
271:   PetscFree(pc->data);
272:   return(0);
273: }

277: static PetscErrorCode PCSetFromOptions_Redistribute(PC pc)
278: {
279:   PetscErrorCode  ierr;
280:   PC_Redistribute *red = (PC_Redistribute*)pc->data;

283:   KSPSetFromOptions(red->ksp);
284:   return(0);
285: }

289: /*@
290:    PCRedistributeGetKSP - Gets the KSP created by the PCREDISTRIBUTE

292:    Not Collective

294:    Input Parameter:
295: .  pc - the preconditioner context

297:    Output Parameter:
298: .  innerksp - the inner KSP

300:    Level: advanced

302: .keywords: PC, redistribute solve
303: @*/
304: PetscErrorCode  PCRedistributeGetKSP(PC pc,KSP *innerksp)
305: {
306:   PC_Redistribute *red = (PC_Redistribute*)pc->data;

311:   *innerksp = red->ksp;
312:   return(0);
313: }

315: /* -------------------------------------------------------------------------------------*/
316: /*MC
317:      PCREDISTRIBUTE - Redistributes a matrix for load balancing, removing the rows that only have a diagonal entry and then applys a KSP to that new matrix

319:      Options for the redistribute preconditioners can be set with -redistribute_ksp_xxx <values> and -redistribute_pc_xxx <values>

321:      Notes:  Usually run this with -ksp_type preonly

323:      If you have used MatZeroRows() to eliminate (for example, Dirichlet) boundary conditions for a symmetric problem then you can use, for example, -ksp_type preonly
324:      -pc_type redistribute -redistribute_ksp_type cg -redistribute_pc_type bjacobi -redistribute_sub_pc_type icc to take advantage of the symmetry.

326:      This does NOT call a partitioner to reorder rows to lower communication; the ordering of the rows in the original matrix and redistributed matrix is the same.

328:      Developer Notes: Should add an option to this preconditioner to use a partitioner to redistribute the rows to lower communication.

330:    Level: intermediate

332: .seealso:  PCCreate(), PCSetType(), PCType (for list of available types), PCRedistributeGetKSP()
333: M*/

337: PETSC_EXTERN PetscErrorCode PCCreate_Redistribute(PC pc)
338: {
339:   PetscErrorCode  ierr;
340:   PC_Redistribute *red;
341:   const char      *prefix;

344:   PetscNewLog(pc,PC_Redistribute,&red);
345:   pc->data = (void*)red;

347:   pc->ops->apply          = PCApply_Redistribute;
348:   pc->ops->applytranspose = 0;
349:   pc->ops->setup          = PCSetUp_Redistribute;
350:   pc->ops->destroy        = PCDestroy_Redistribute;
351:   pc->ops->setfromoptions = PCSetFromOptions_Redistribute;
352:   pc->ops->view           = PCView_Redistribute;

354:   KSPCreate(PetscObjectComm((PetscObject)pc),&red->ksp);
355:   PetscObjectIncrementTabLevel((PetscObject)red->ksp,(PetscObject)pc,1);
356:   PetscLogObjectParent(pc,red->ksp);
357:   PCGetOptionsPrefix(pc,&prefix);
358:   KSPSetOptionsPrefix(red->ksp,prefix);
359:   KSPAppendOptionsPrefix(red->ksp,"redistribute_");
360:   return(0);
361: }