Dune Core Modules (2.9.1)

fastamg.hh
Go to the documentation of this file.
1// SPDX-FileCopyrightText: Copyright (C) DUNE Project contributors, see file LICENSE.md in module root
2// SPDX-License-Identifier: LicenseRef-GPL-2.0-only-with-DUNE-exception
3// -*- tab-width: 4; indent-tabs-mode: nil; c-basic-offset: 2 -*-
4// vi: set et ts=4 sw=2 sts=2:
5#ifndef DUNE_ISTL_FASTAMG_HH
6#define DUNE_ISTL_FASTAMG_HH
7
8#include <memory>
14#include <dune/istl/solvers.hh>
16#include <dune/istl/superlu.hh>
17#include <dune/istl/umfpack.hh>
19#include <dune/istl/io.hh>
21
22#include "fastamgsmoother.hh"
23
32namespace Dune
33{
34 namespace Amg
35 {
58 template<class M, class X, class PI=SequentialInformation, class A=std::allocator<X> >
59 class FastAMG : public Preconditioner<X,X>
60 {
61 public:
63 typedef M Operator;
75
77 typedef X Domain;
79 typedef X Range;
82
90 FastAMG(OperatorHierarchy& matrices, CoarseSolver& coarseSolver,
91 const Parameters& parms,
92 bool symmetric=true);
93
105 template<class C>
106 FastAMG(const Operator& fineOperator, const C& criterion,
107 const Parameters& parms=Parameters(),
108 bool symmetric=true,
110
114 FastAMG(const FastAMG& amg);
115
117 void pre(Domain& x, Range& b);
118
120 void apply(Domain& v, const Range& d);
121
124 {
126 }
127
129 void post(Domain& x);
130
135 template<class A1>
136 void getCoarsestAggregateNumbers(std::vector<std::size_t,A1>& cont);
137
138 std::size_t levels();
139
140 std::size_t maxlevels();
141
151 {
152 matrices_->recalculateGalerkin(NegateSet<typename PI::OwnerSet>());
153 }
154
159 bool usesDirectCoarseLevelSolver() const;
160
161 private:
168 template<class C>
169 void createHierarchies(C& criterion,
170 const std::shared_ptr<const Operator>& matrixptr,
171 const PI& pinfo);
172
179 struct LevelContext
180 {
192 typename OperatorHierarchy::RedistributeInfoList::const_iterator redist;
196 typename OperatorHierarchy::AggregatesMapList::const_iterator aggregates;
212 std::size_t level;
213 };
214
216 void mgc(LevelContext& levelContext, Domain& x, const Range& b);
217
224 void presmooth(LevelContext& levelContext, Domain& x, const Range& b);
225
232 void postsmooth(LevelContext& levelContext, Domain& x, const Range& b);
233
240 void moveToFineLevel(LevelContext& levelContext, bool processedFineLevel,
241 Domain& fineX);
242
247 bool moveToCoarseLevel(LevelContext& levelContext);
248
253 void initIteratorsWithFineLevel(LevelContext& levelContext);
254
256 std::shared_ptr<OperatorHierarchy> matrices_;
258 std::shared_ptr<CoarseSolver> solver_;
260 std::shared_ptr<Hierarchy<Range,A>> rhs_;
262 std::shared_ptr<Hierarchy<Domain,A>> lhs_;
264 std::shared_ptr<Hierarchy<Domain,A>> residual_;
265
269 std::shared_ptr<ScalarProduct> scalarProduct_;
271 std::size_t gamma_;
273 std::size_t preSteps_;
275 std::size_t postSteps_;
276 std::size_t level;
277 bool buildHierarchy_;
278 bool symmetric;
279 bool coarsesolverconverged;
281 typedef std::shared_ptr<Smoother> SmootherPointer;
282 SmootherPointer coarseSmoother_;
284 std::size_t verbosity_;
285 };
286
287 template<class M, class X, class PI, class A>
289 : matrices_(amg.matrices_), solver_(amg.solver_),
290 rhs_(), lhs_(), residual_(), scalarProduct_(amg.scalarProduct_),
291 gamma_(amg.gamma_), preSteps_(amg.preSteps_), postSteps_(amg.postSteps_),
292 symmetric(amg.symmetric), coarsesolverconverged(amg.coarsesolverconverged),
293 coarseSmoother_(amg.coarseSmoother_), verbosity_(amg.verbosity_)
294 {}
295
296 template<class M, class X, class PI, class A>
298 const Parameters& parms, bool symmetric_)
299 : matrices_(stackobject_to_shared_ptr(matrices)), solver_(&coarseSolver),
300 rhs_(), lhs_(), residual_(), scalarProduct_(),
301 gamma_(parms.getGamma()), preSteps_(parms.getNoPreSmoothSteps()),
302 postSteps_(parms.getNoPostSmoothSteps()), buildHierarchy_(false),
303 symmetric(symmetric_), coarsesolverconverged(true),
304 coarseSmoother_(), verbosity_(parms.debugLevel())
305 {
306 if(preSteps_>1||postSteps_>1)
307 {
308 std::cerr<<"WARNING only one step of smoothing is supported!"<<std::endl;
309 preSteps_=postSteps_=0;
310 }
311 assert(matrices_->isBuilt());
312 static_assert(std::is_same<PI,SequentialInformation>::value,
313 "Currently only sequential runs are supported");
314 }
315 template<class M, class X, class PI, class A>
316 template<class C>
318 const C& criterion,
319 const Parameters& parms,
320 bool symmetric_,
321 const PI& pinfo)
322 : solver_(), rhs_(), lhs_(), residual_(), scalarProduct_(), gamma_(parms.getGamma()),
323 preSteps_(parms.getNoPreSmoothSteps()), postSteps_(parms.getNoPostSmoothSteps()),
324 buildHierarchy_(true),
325 symmetric(symmetric_), coarsesolverconverged(true),
326 coarseSmoother_(), verbosity_(criterion.debugLevel())
327 {
328 if(preSteps_>1||postSteps_>1)
329 {
330 std::cerr<<"WARNING only one step of smoothing is supported!"<<std::endl;
331 preSteps_=postSteps_=1;
332 }
333 static_assert(std::is_same<PI,SequentialInformation>::value,
334 "Currently only sequential runs are supported");
335 // TODO: reestablish compile time checks.
336 //static_assert(static_cast<int>(PI::category)==static_cast<int>(S::category),
337 // "Matrix and Solver must match in terms of category!");
338 auto matrixptr = stackobject_to_shared_ptr(matrix);
339 createHierarchies(criterion, matrixptr, pinfo);
340 }
341
342 template<class M, class X, class PI, class A>
343 template<class C>
344 void FastAMG<M,X,PI,A>::createHierarchies(C& criterion,
345 const std::shared_ptr<const Operator>& matrixptr,
346 const PI& pinfo)
347 {
348 Timer watch;
349 matrices_ = std::make_shared<OperatorHierarchy>(
350 std::const_pointer_cast<Operator>(matrixptr),
351 stackobject_to_shared_ptr(const_cast<PI&>(pinfo)));
352
353 matrices_->template build<NegateSet<typename PI::OwnerSet> >(criterion);
354
355 if(verbosity_>0 && matrices_->parallelInformation().finest()->communicator().rank()==0)
356 std::cout<<"Building Hierarchy of "<<matrices_->maxlevels()<<" levels took "<<watch.elapsed()<<" seconds."<<std::endl;
357
358 if(buildHierarchy_ && matrices_->levels()==matrices_->maxlevels()) {
359 // We have the carsest level. Create the coarse Solver
360 typedef typename SmootherTraits<Smoother>::Arguments SmootherArgs;
361 SmootherArgs sargs;
362 sargs.iterations = 1;
363
365 cargs.setArgs(sargs);
366 if(matrices_->redistributeInformation().back().isSetup()) {
367 // Solve on the redistributed partitioning
368 cargs.setMatrix(matrices_->matrices().coarsest().getRedistributed().getmat());
369 cargs.setComm(matrices_->parallelInformation().coarsest().getRedistributed());
370 }else{
371 cargs.setMatrix(matrices_->matrices().coarsest()->getmat());
372 cargs.setComm(*matrices_->parallelInformation().coarsest());
373 }
374
375 coarseSmoother_ = ConstructionTraits<Smoother>::construct(cargs);
376 scalarProduct_ = createScalarProduct<X>(cargs.getComm(),category());
377
378#if HAVE_SUPERLU|| HAVE_SUITESPARSE_UMFPACK
379#if HAVE_SUITESPARSE_UMFPACK
380#define DIRECTSOLVER UMFPack
381#else
382#define DIRECTSOLVER SuperLU
383#endif
384 // Use superlu if we are purely sequential or with only one processor on the coarsest level.
385 if(std::is_same<ParallelInformation,SequentialInformation>::value // sequential mode
386 || matrices_->parallelInformation().coarsest()->communicator().size()==1 //parallel mode and only one processor
387 || (matrices_->parallelInformation().coarsest().isRedistributed()
388 && matrices_->parallelInformation().coarsest().getRedistributed().communicator().size()==1
389 && matrices_->parallelInformation().coarsest().getRedistributed().communicator().size()>0)) { // redistribute and 1 proc
390 if(verbosity_>0 && matrices_->parallelInformation().coarsest()->communicator().rank()==0)
391 std::cout<<"Using superlu"<<std::endl;
392 if(matrices_->parallelInformation().coarsest().isRedistributed())
393 {
394 if(matrices_->matrices().coarsest().getRedistributed().getmat().N()>0)
395 // We are still participating on this level
396 solver_.reset(new DIRECTSOLVER<typename M::matrix_type>(matrices_->matrices().coarsest().getRedistributed().getmat(), false, false));
397 else
398 solver_.reset();
399 }else
400 solver_.reset(new DIRECTSOLVER<typename M::matrix_type>(matrices_->matrices().coarsest()->getmat(), false, false));
401 }else
402#undef DIRECTSOLVER
403#endif // HAVE_SUPERLU|| HAVE_SUITESPARSE_UMFPACK
404 {
405 if(matrices_->parallelInformation().coarsest().isRedistributed())
406 {
407 if(matrices_->matrices().coarsest().getRedistributed().getmat().N()>0)
408 // We are still participating on this level
409 solver_.reset(new BiCGSTABSolver<X>(const_cast<M&>(matrices_->matrices().coarsest().getRedistributed()),
410 *scalarProduct_,
411 *coarseSmoother_, 1E-2, 1000, 0));
412 else
413 solver_.reset();
414 }else
415 solver_.reset(new BiCGSTABSolver<X>(const_cast<M&>(*matrices_->matrices().coarsest()),
416 *scalarProduct_,
417 *coarseSmoother_, 1E-2, 1000, 0));
418 }
419 }
420
421 if(verbosity_>0 && matrices_->parallelInformation().finest()->communicator().rank()==0)
422 std::cout<<"Building Hierarchy of "<<matrices_->maxlevels()<<" levels took "<<watch.elapsed()<<" seconds."<<std::endl;
423 }
424
425
426 template<class M, class X, class PI, class A>
428 {
429 Timer watch, watch1;
430 // Detect Matrix rows where all offdiagonal entries are
431 // zero and set x such that A_dd*x_d=b_d
432 // Thus users can be more careless when setting up their linear
433 // systems.
434 typedef typename M::matrix_type Matrix;
435 typedef typename Matrix::ConstRowIterator RowIter;
436 typedef typename Matrix::ConstColIterator ColIter;
437 typedef typename Matrix::block_type Block;
438 Block zero;
439 zero=typename Matrix::field_type();
440
441 const Matrix& mat=matrices_->matrices().finest()->getmat();
442 for(RowIter row=mat.begin(); row!=mat.end(); ++row) {
443 bool isDirichlet = true;
444 bool hasDiagonal = false;
445 ColIter diag;
446 for(ColIter col=row->begin(); col!=row->end(); ++col) {
447 if(row.index()==col.index()) {
448 diag = col;
449 hasDiagonal = (*col != zero);
450 }else{
451 if(*col!=zero)
452 isDirichlet = false;
453 }
454 }
455 if(isDirichlet && hasDiagonal)
456 diag->solve(x[row.index()], b[row.index()]);
457 }
458 if (verbosity_>0)
459 std::cout<<" Preprocessing Dirichlet took "<<watch1.elapsed()<<std::endl;
460 watch1.reset();
461 // No smoother to make x consistent! Do it by hand
462 matrices_->parallelInformation().coarsest()->copyOwnerToAll(x,x);
463 rhs_ = std::make_shared<Hierarchy<Range,A>>(std::make_shared<Range>(b));
464 lhs_ = std::make_shared<Hierarchy<Domain,A>>(std::make_shared<Domain>(x));
465 residual_ = std::make_shared<Hierarchy<Domain,A>>(std::make_shared<Domain>(x));
466 matrices_->coarsenVector(*rhs_);
467 matrices_->coarsenVector(*lhs_);
468 matrices_->coarsenVector(*residual_);
469
470 // The preconditioner might change x and b. So we have to
471 // copy the changes to the original vectors.
472 x = *lhs_->finest();
473 b = *rhs_->finest();
474 }
475 template<class M, class X, class PI, class A>
476 std::size_t FastAMG<M,X,PI,A>::levels()
477 {
478 return matrices_->levels();
479 }
480 template<class M, class X, class PI, class A>
481 std::size_t FastAMG<M,X,PI,A>::maxlevels()
482 {
483 return matrices_->maxlevels();
484 }
485
487 template<class M, class X, class PI, class A>
489 {
490 LevelContext levelContext;
491 // Init all iterators for the current level
492 initIteratorsWithFineLevel(levelContext);
493
494 assert(v.two_norm()==0);
495
496 level=0;
497 if(matrices_->maxlevels()==1){
498 // The coarse solver might modify the d!
499 Range b(d);
500 mgc(levelContext, v, b);
501 }else
502 mgc(levelContext, v, d);
503 if(postSteps_==0||matrices_->maxlevels()==1)
504 levelContext.pinfo->copyOwnerToAll(v, v);
505 }
506
507 template<class M, class X, class PI, class A>
508 void FastAMG<M,X,PI,A>::initIteratorsWithFineLevel(LevelContext& levelContext)
509 {
510 levelContext.matrix = matrices_->matrices().finest();
511 levelContext.pinfo = matrices_->parallelInformation().finest();
512 levelContext.redist =
513 matrices_->redistributeInformation().begin();
514 levelContext.aggregates = matrices_->aggregatesMaps().begin();
515 levelContext.lhs = lhs_->finest();
516 levelContext.residual = residual_->finest();
517 levelContext.rhs = rhs_->finest();
518 levelContext.level=0;
519 }
520
521 template<class M, class X, class PI, class A>
522 bool FastAMG<M,X,PI,A>
523 ::moveToCoarseLevel(LevelContext& levelContext)
524 {
525 bool processNextLevel=true;
526
527 if(levelContext.redist->isSetup()) {
528 throw "bla";
529 levelContext.redist->redistribute(static_cast<const Range&>(*levelContext.residual),
530 levelContext.residual.getRedistributed());
531 processNextLevel = levelContext.residual.getRedistributed().size()>0;
532 if(processNextLevel) {
533 //restrict defect to coarse level right hand side.
534 ++levelContext.pinfo;
535 Transfer<typename OperatorHierarchy::AggregatesMap::AggregateDescriptor,Range,ParallelInformation>
536 ::restrictVector(*(*levelContext.aggregates), *levelContext.rhs,
537 static_cast<const Range&>(levelContext.residual.getRedistributed()),
538 *levelContext.pinfo);
539 }
540 }else{
541 //restrict defect to coarse level right hand side.
542 ++levelContext.rhs;
543 ++levelContext.pinfo;
544 Transfer<typename OperatorHierarchy::AggregatesMap::AggregateDescriptor,Range,ParallelInformation>
545 ::restrictVector(*(*levelContext.aggregates), *levelContext.rhs,
546 static_cast<const Range&>(*levelContext.residual), *levelContext.pinfo);
547 }
548
549 if(processNextLevel) {
550 // prepare coarse system
551 ++levelContext.residual;
552 ++levelContext.lhs;
553 ++levelContext.matrix;
554 ++levelContext.level;
555 ++levelContext.redist;
556
557 if(levelContext.matrix != matrices_->matrices().coarsest() || matrices_->levels()<matrices_->maxlevels()) {
558 // next level is not the globally coarsest one
559 ++levelContext.aggregates;
560 }
561 // prepare the lhs on the next level
562 *levelContext.lhs=0;
563 *levelContext.residual=0;
564 }
565 return processNextLevel;
566 }
567
568 template<class M, class X, class PI, class A>
569 void FastAMG<M,X,PI,A>
570 ::moveToFineLevel(LevelContext& levelContext, bool processNextLevel, Domain& x)
571 {
572 if(processNextLevel) {
573 if(levelContext.matrix != matrices_->matrices().coarsest() || matrices_->levels()<matrices_->maxlevels()) {
574 // previous level is not the globally coarsest one
575 --levelContext.aggregates;
576 }
577 --levelContext.redist;
578 --levelContext.level;
579 //prolongate and add the correction (update is in coarse left hand side)
580 --levelContext.matrix;
581 --levelContext.residual;
582
583 }
584
585 typename Hierarchy<Domain,A>::Iterator coarseLhs = levelContext.lhs--;
586 if(levelContext.redist->isSetup()) {
587
588 // Need to redistribute during prolongate
589 Transfer<typename OperatorHierarchy::AggregatesMap::AggregateDescriptor,Range,ParallelInformation>
590 ::prolongateVector(*(*levelContext.aggregates), *coarseLhs, x,
591 levelContext.lhs.getRedistributed(),
592 matrices_->getProlongationDampingFactor(),
593 *levelContext.pinfo, *levelContext.redist);
594 }else{
595 Transfer<typename OperatorHierarchy::AggregatesMap::AggregateDescriptor,Range,ParallelInformation>
596 ::prolongateVector(*(*levelContext.aggregates), *coarseLhs, x,
597 matrices_->getProlongationDampingFactor(), *levelContext.pinfo);
598
599 // printvector(std::cout, *lhs, "prolongated coarse grid correction", "lhs", 10, 10, 10);
600 }
601
602
603 if(processNextLevel) {
604 --levelContext.rhs;
605 }
606
607 }
608
609
610 template<class M, class X, class PI, class A>
611 void FastAMG<M,X,PI,A>
612 ::presmooth(LevelContext& levelContext, Domain& x, const Range& b)
613 {
614 constexpr auto bl = blockLevel<typename M::matrix_type>();
615 GaussSeidelPresmoothDefect<bl>::apply(levelContext.matrix->getmat(),
616 x,
617 *levelContext.residual,
618 b);
619 }
620
621 template<class M, class X, class PI, class A>
622 void FastAMG<M,X,PI,A>
623 ::postsmooth(LevelContext& levelContext, Domain& x, const Range& b)
624 {
625 constexpr auto bl = blockLevel<typename M::matrix_type>();
626 GaussSeidelPostsmoothDefect<bl>
627 ::apply(levelContext.matrix->getmat(), x, *levelContext.residual, b);
628 }
629
630
631 template<class M, class X, class PI, class A>
633 {
634 return IsDirectSolver< CoarseSolver>::value;
635 }
636
637 template<class M, class X, class PI, class A>
638 void FastAMG<M,X,PI,A>::mgc(LevelContext& levelContext, Domain& v, const Range& b){
639
640 if(levelContext.matrix == matrices_->matrices().coarsest() && levels()==maxlevels()) {
641 // Solve directly
643 res.converged=true; // If we do not compute this flag will not get updated
644 if(levelContext.redist->isSetup()) {
645 levelContext.redist->redistribute(b, levelContext.rhs.getRedistributed());
646 if(levelContext.rhs.getRedistributed().size()>0) {
647 // We are still participating in the computation
648 levelContext.pinfo.getRedistributed().copyOwnerToAll(levelContext.rhs.getRedistributed(),
649 levelContext.rhs.getRedistributed());
650 solver_->apply(levelContext.lhs.getRedistributed(), levelContext.rhs.getRedistributed(), res);
651 }
652 levelContext.redist->redistributeBackward(v, levelContext.lhs.getRedistributed());
653 levelContext.pinfo->copyOwnerToAll(v, v);
654 }else{
655 levelContext.pinfo->copyOwnerToAll(b, b);
656 solver_->apply(v, const_cast<Range&>(b), res);
657 }
658
659 // printvector(std::cout, *lhs, "coarse level update", "u", 10, 10, 10);
660 // printvector(std::cout, *rhs, "coarse level rhs", "rhs", 10, 10, 10);
661 if (!res.converged)
662 coarsesolverconverged = false;
663 }else{
664 // presmoothing
665 presmooth(levelContext, v, b);
666 // printvector(std::cout, *lhs, "update", "u", 10, 10, 10);
667 // printvector(std::cout, *residual, "post presmooth residual", "r", 10);
668#ifndef DUNE_AMG_NO_COARSEGRIDCORRECTION
669 bool processNextLevel = moveToCoarseLevel(levelContext);
670
671 if(processNextLevel) {
672 // next level
673 for(std::size_t i=0; i<gamma_; i++)
674 mgc(levelContext, *levelContext.lhs, *levelContext.rhs);
675 }
676
677 moveToFineLevel(levelContext, processNextLevel, v);
678#else
679 *lhs=0;
680#endif
681
682 if(levelContext.matrix == matrices_->matrices().finest()) {
683 coarsesolverconverged = matrices_->parallelInformation().finest()->communicator().prod(coarsesolverconverged);
684 if(!coarsesolverconverged)
685 DUNE_THROW(MathError, "Coarse solver did not converge");
686 }
687
688 postsmooth(levelContext, v, b);
689 }
690 }
691
692
694 template<class M, class X, class PI, class A>
695 void FastAMG<M,X,PI,A>::post([[maybe_unused]] Domain& x)
696 {
697 lhs_=nullptr;
698 rhs_=nullptr;
699 residual_=nullptr;
700 }
701
702 template<class M, class X, class PI, class A>
703 template<class A1>
704 void FastAMG<M,X,PI,A>::getCoarsestAggregateNumbers(std::vector<std::size_t,A1>& cont)
705 {
706 matrices_->getCoarsestAggregatesOnFinest(cont);
707 }
708
709 } // end namespace Amg
710} // end namespace Dune
711
712#endif
A fast (sequential) algebraic multigrid based on agglomeration that saves memory bandwidth.
Definition: fastamg.hh:60
LevelIterator< Hierarchy< ParallelInformation, Allocator >, ParallelInformation > Iterator
Type of the mutable iterator.
Definition: hierarchy.hh:216
LevelIterator< const Hierarchy< MatrixOperator, Allocator >, const MatrixOperator > ConstIterator
Type of the const iterator.
Definition: hierarchy.hh:219
The hierarchies build by the coarsening process.
Definition: matrixhierarchy.hh:61
All parameters for AMG.
Definition: parameters.hh:393
ConstIterator class for sequential access.
Definition: matrix.hh:404
A generic dynamic dense matrix.
Definition: matrix.hh:561
typename Imp::BlockTraits< T >::field_type field_type
Export the type representing the underlying field.
Definition: matrix.hh:565
row_type::const_iterator ConstColIterator
Const iterator for the entries of each row.
Definition: matrix.hh:589
T block_type
Export the type representing the components.
Definition: matrix.hh:568
The negation of a set. An item is contained in the set if and only if it is not contained in the nega...
Definition: enumset.hh:96
Base class for matrix free definition of preconditioners.
Definition: preconditioner.hh:32
Base class for scalar product and norm computation.
Definition: scalarproducts.hh:52
Sequential SSOR preconditioner.
Definition: preconditioners.hh:141
A simple stop watch.
Definition: timer.hh:43
void reset() noexcept
Reset timer while keeping the running/stopped state.
Definition: timer.hh:57
double elapsed() const noexcept
Get elapsed user-time from last reset until now/last stop in seconds.
Definition: timer.hh:77
A few common exception classes.
Some generic functions for pretty printing vectors and matrices.
#define DUNE_THROW(E, m)
Definition: exceptions.hh:218
OperatorHierarchy::ParallelMatrixHierarchy::ConstIterator matrix
The iterator over the matrices.
Definition: fastamg.hh:184
void getCoarsestAggregateNumbers(std::vector< std::size_t, A1 > &cont)
Get the aggregate number of each unknown on the coarsest level.
Definition: fastamg.hh:704
ParallelInformationHierarchy::Iterator pinfo
The iterator over the parallel information.
Definition: fastamg.hh:188
void recalculateHierarchy()
Recalculate the matrix hierarchy.
Definition: fastamg.hh:150
void post(Domain &x)
Clean up.
Definition: fastamg.hh:695
X Domain
The domain type.
Definition: fastamg.hh:77
Hierarchy< Domain, A >::Iterator residual
The iterator over the residuals.
Definition: fastamg.hh:204
virtual SolverCategory::Category category() const
Category of the preconditioner (see SolverCategory::Category)
Definition: fastamg.hh:123
MatrixHierarchy< M, ParallelInformation, A > OperatorHierarchy
The operator hierarchy type.
Definition: fastamg.hh:72
OperatorHierarchy::RedistributeInfoList::const_iterator redist
The iterator over the redistribution information.
Definition: fastamg.hh:192
X Range
The range type.
Definition: fastamg.hh:79
PI ParallelInformation
The type of the parallel information. Either OwnerOverlapCommunication or another type describing the...
Definition: fastamg.hh:70
M Operator
The matrix operator type.
Definition: fastamg.hh:63
InverseOperator< X, X > CoarseSolver
the type of the coarse solver.
Definition: fastamg.hh:81
bool usesDirectCoarseLevelSolver() const
Check whether the coarse solver used is a direct solver.
Definition: fastamg.hh:632
Hierarchy< Domain, A >::Iterator lhs
The iterator over the left hand side.
Definition: fastamg.hh:200
Hierarchy< Range, A >::Iterator rhs
The iterator over the right hand sided.
Definition: fastamg.hh:208
std::size_t level
The level index.
Definition: fastamg.hh:212
void apply(Domain &v, const Range &d)
Apply one step of the preconditioner to the system A(v)=d.
Definition: fastamg.hh:488
OperatorHierarchy::ParallelInformationHierarchy ParallelInformationHierarchy
The parallal data distribution hierarchy type.
Definition: fastamg.hh:74
void pre(Domain &x, Range &b)
Prepare the preconditioner.
Definition: fastamg.hh:427
FastAMG(OperatorHierarchy &matrices, CoarseSolver &coarseSolver, const Parameters &parms, bool symmetric=true)
Construct a new amg with a specific coarse solver.
Definition: fastamg.hh:297
OperatorHierarchy::AggregatesMapList::const_iterator aggregates
The iterator over the aggregates maps.
Definition: fastamg.hh:196
void presmooth(LevelContext &levelContext, size_t steps)
Apply pre smoothing on the current level.
Definition: smoother.hh:406
const void * Arguments
A type holding all the arguments needed to call the constructor.
Definition: construction.hh:44
static std::shared_ptr< T > construct(Arguments &args)
Construct an object with the specified arguments.
Definition: construction.hh:52
void postsmooth(LevelContext &levelContext, size_t steps)
Apply post smoothing on the current level.
Definition: smoother.hh:428
Provides a classes representing the hierarchies in AMG.
Dune namespace.
Definition: alignedallocator.hh:13
std::shared_ptr< T > stackobject_to_shared_ptr(T &t)
Create a shared_ptr for a stack-allocated object.
Definition: shared_ptr.hh:72
Define general preconditioner interface.
Define base class for scalar product and norm.
Classes for the generic construction and application of the smoothers.
Implementations of the inverse operator interface.
Templates characterizing the type of a solver.
Traits class for getting the attribute class of a smoother.
Definition: smoother.hh:66
Statistics about the application of an inverse operator.
Definition: solver.hh:48
bool converged
True if convergence criterion has been met.
Definition: solver.hh:73
Category
Definition: solvercategory.hh:23
@ sequential
Category for sequential solvers.
Definition: solvercategory.hh:25
Classes for using SuperLU with ISTL matrices.
Prolongation and restriction for amg.
Traits for type conversions and type information.
Classes for using UMFPack with ISTL matrices.
Creative Commons License   |  Legal Statements / Impressum  |  Hosted by TU Dresden  |  generated with Hugo v0.111.3 (Nov 21, 23:30, 2024)