DUNE PDELab (2.8)

smoother.hh
Go to the documentation of this file.
1// -*- tab-width: 4; indent-tabs-mode: nil; c-basic-offset: 2 -*-
2// vi: set et ts=4 sw=2 sts=2:
3#ifndef DUNE_AMGSMOOTHER_HH
4#define DUNE_AMGSMOOTHER_HH
5
9#include <dune/istl/schwarz.hh>
10#include <dune/istl/novlpschwarz.hh>
11#include <dune/common/propertymap.hh>
13
14namespace Dune
15{
16 namespace Amg
17 {
18
34 template<class T>
36 {
40 typedef typename FieldTraits<T>::real_type RelaxationFactor;
41
50
56 {}
57 };
58
62 template<class T>
64 {
66
67 };
68
69 template<class X, class Y>
70 struct SmootherTraits<Richardson<X,Y>>
71 {
73
74 };
75
76 template<class X, class Y, class C, class T>
77 struct SmootherTraits<BlockPreconditioner<X,Y,C,T> >
78 : public SmootherTraits<T>
79 {};
80
81 template<class C, class T>
82 struct SmootherTraits<NonoverlappingBlockPreconditioner<C,T> >
83 : public SmootherTraits<T>
84 {};
85
89 template<class T>
91 {
92 typedef typename T::matrix_type Matrix;
93
95
97
98 public:
100 {}
101
102 void setMatrix(const Matrix& matrix)
103 {
104 matrix_=&matrix;
105 }
106 virtual void setMatrix(const Matrix& matrix, [[maybe_unused]] const AggregatesMap& amap)
107 {
108 setMatrix(matrix);
109 }
110
111
112 const Matrix& getMatrix() const
113 {
114 return *matrix_;
115 }
116
117 void setArgs(const SmootherArgs& args)
118 {
119 args_=&args;
120 }
121
122 template<class T1>
123 void setComm([[maybe_unused]] T1& comm)
124 {}
125
126 const SequentialInformation& getComm()
127 {
128 return comm_;
129 }
130
131 const SmootherArgs getArgs() const
132 {
133 return *args_;
134 }
135
136 protected:
137 const Matrix* matrix_;
138 private:
139 const SmootherArgs* args_;
140 SequentialInformation comm_;
141 };
142
143 template<class T>
144 struct ConstructionArgs
145 : public DefaultConstructionArgs<T>
146 {};
147
148 template<class T, class C=SequentialInformation>
149 class DefaultParallelConstructionArgs
150 : public ConstructionArgs<T>
151 {
152 public:
153 virtual ~DefaultParallelConstructionArgs()
154 {}
155
156 void setComm(const C& comm)
157 {
158 comm_ = &comm;
159 }
160
161 const C& getComm() const
162 {
163 return *comm_;
164 }
165 private:
166 const C* comm_;
167 };
168
169
170 template<class X, class Y>
171 class DefaultConstructionArgs<Richardson<X,Y>>
172 {
173 typedef Richardson<X,Y> T;
174
175 typedef typename SmootherTraits<T>::Arguments SmootherArgs;
176
177 public:
178 virtual ~DefaultConstructionArgs()
179 {}
180
181 template <class... Args>
182 void setMatrix(const Args&...)
183 {}
184
185 void setArgs(const SmootherArgs& args)
186 {
187 args_=&args;
188 }
189
190 template<class T1>
191 void setComm([[maybe_unused]] T1& comm)
192 {}
193
194 const SequentialInformation& getComm()
195 {
196 return comm_;
197 }
198
199 const SmootherArgs getArgs() const
200 {
201 return *args_;
202 }
203
204 private:
205 const SmootherArgs* args_;
206 SequentialInformation comm_;
207 };
208
209
210
211 template<class T>
212 struct ConstructionTraits;
213
217 template<class M, class X, class Y, int l>
218 struct ConstructionTraits<SeqSSOR<M,X,Y,l> >
219 {
221
222 static inline std::shared_ptr<SeqSSOR<M,X,Y,l>> construct(Arguments& args)
223 {
224 return std::make_shared<SeqSSOR<M,X,Y,l>>
225 (args.getMatrix(), args.getArgs().iterations, args.getArgs().relaxationFactor);
226 }
227 };
228
229
233 template<class M, class X, class Y, int l>
234 struct ConstructionTraits<SeqSOR<M,X,Y,l> >
235 {
237
238 static inline std::shared_ptr<SeqSOR<M,X,Y,l>> construct(Arguments& args)
239 {
240 return std::make_shared<SeqSOR<M,X,Y,l>>
241 (args.getMatrix(), args.getArgs().iterations, args.getArgs().relaxationFactor);
242 }
243 };
244
245
249 template<class M, class X, class Y, int l>
250 struct ConstructionTraits<SeqJac<M,X,Y,l> >
251 {
253
254 static inline std::shared_ptr<SeqJac<M,X,Y,l>> construct(Arguments& args)
255 {
256 return std::make_shared<SeqJac<M,X,Y,l>>
257 (args.getMatrix(), args.getArgs().iterations, args.getArgs().relaxationFactor);
258 }
259 };
260
264 template<class X, class Y>
266 {
268
269 static inline std::shared_ptr<Richardson<X,Y>> construct(Arguments& args)
270 {
271 return std::make_shared<Richardson<X,Y>>
272 (args.getArgs().relaxationFactor);
273 }
274 };
275
276
277 template<class M, class X, class Y>
278 class ConstructionArgs<SeqILU<M,X,Y> >
279 : public DefaultConstructionArgs<SeqILU<M,X,Y> >
280 {
281 public:
282 ConstructionArgs(int n=0)
283 : n_(n)
284 {}
285
286 void setN(int n)
287 {
288 n_ = n;
289 }
290
291 int getN()
292 {
293 return n_;
294 }
295
296 private:
297 int n_;
298 };
299
300
304 template<class M, class X, class Y>
306 {
307 typedef ConstructionArgs<SeqILU<M,X,Y> > Arguments;
308
309 static inline std::shared_ptr<SeqILU<M,X,Y>> construct(Arguments& args)
310 {
311 return std::make_shared<SeqILU<M,X,Y>>
312 (args.getMatrix(), args.getN(), args.getArgs().relaxationFactor);
313 }
314 };
315
319 template<class M, class X, class Y, class C>
320 struct ConstructionTraits<ParSSOR<M,X,Y,C> >
321 {
322 typedef DefaultParallelConstructionArgs<M,C> Arguments;
323
324 static inline std::shared_ptr<ParSSOR<M,X,Y,C>> construct(Arguments& args)
325 {
326 return std::make_shared<ParSSOR<M,X,Y,C>>
327 (args.getMatrix(), args.getArgs().iterations,
328 args.getArgs().relaxationFactor, args.getComm());
329 }
330 };
331
332 template<class X, class Y, class C, class T>
334 {
335 typedef DefaultParallelConstructionArgs<T,C> Arguments;
336 typedef ConstructionTraits<T> SeqConstructionTraits;
337 static inline std::shared_ptr<BlockPreconditioner<X,Y,C,T>> construct(Arguments& args)
338 {
339 auto seqPrec = SeqConstructionTraits::construct(args);
340 return std::make_shared<BlockPreconditioner<X,Y,C,T>> (seqPrec, args.getComm());
341 }
342 };
343
344 template<class C, class T>
345 struct ConstructionTraits<NonoverlappingBlockPreconditioner<C,T> >
346 {
347 typedef DefaultParallelConstructionArgs<T,C> Arguments;
348 typedef ConstructionTraits<T> SeqConstructionTraits;
349 static inline std::shared_ptr<NonoverlappingBlockPreconditioner<C,T>> construct(Arguments& args)
350 {
351 auto seqPrec = SeqConstructionTraits::construct(args);
352 return std::make_shared<NonoverlappingBlockPreconditioner<C,T>> (seqPrec, args.getComm());
353 }
354 };
355
366 template<class T>
368 {
369 typedef T Smoother;
370 typedef typename Smoother::range_type Range;
371 typedef typename Smoother::domain_type Domain;
372
380 static void preSmooth(Smoother& smoother, Domain& v, const Range& d)
381 {
382 smoother.apply(v,d);
383 }
384
392 static void postSmooth(Smoother& smoother, Domain& v, const Range& d)
393 {
394 smoother.apply(v,d);
395 }
396 };
397
403 template<typename LevelContext>
404 void presmooth(LevelContext& levelContext, size_t steps)
405 {
406 for(std::size_t i=0; i < steps; ++i) {
407 *levelContext.lhs=0;
409 ::preSmooth(*levelContext.smoother, *levelContext.lhs,
410 *levelContext.rhs);
411 // Accumulate update
412 *levelContext.update += *levelContext.lhs;
413
414 // update defect
415 levelContext.matrix->applyscaleadd(-1, *levelContext.lhs, *levelContext.rhs);
416 levelContext.pinfo->project(*levelContext.rhs);
417 }
418 }
419
425 template<typename LevelContext>
426 void postsmooth(LevelContext& levelContext, size_t steps)
427 {
428 for(std::size_t i=0; i < steps; ++i) {
429 // update defect
430 levelContext.matrix->applyscaleadd(-1, *levelContext.lhs,
431 *levelContext.rhs);
432 *levelContext.lhs=0;
433 levelContext.pinfo->project(*levelContext.rhs);
435 ::postSmooth(*levelContext.smoother, *levelContext.lhs, *levelContext.rhs);
436 // Accumulate update
437 *levelContext.update += *levelContext.lhs;
438 }
439 }
440
441 template<class M, class X, class Y, int l>
442 struct SmootherApplier<SeqSOR<M,X,Y,l> >
443 {
444 typedef SeqSOR<M,X,Y,l> Smoother;
445 typedef typename Smoother::range_type Range;
446 typedef typename Smoother::domain_type Domain;
447
448 static void preSmooth(Smoother& smoother, Domain& v, Range& d)
449 {
450 smoother.template apply<true>(v,d);
451 }
452
453
454 static void postSmooth(Smoother& smoother, Domain& v, Range& d)
455 {
456 smoother.template apply<false>(v,d);
457 }
458 };
459
460 template<class M, class X, class Y, class C, int l>
461 struct SmootherApplier<BlockPreconditioner<X,Y,C,SeqSOR<M,X,Y,l> > >
462 {
463 typedef BlockPreconditioner<X,Y,C,SeqSOR<M,X,Y,l> > Smoother;
464 typedef typename Smoother::range_type Range;
465 typedef typename Smoother::domain_type Domain;
466
467 static void preSmooth(Smoother& smoother, Domain& v, Range& d)
468 {
469 smoother.template apply<true>(v,d);
470 }
471
472
473 static void postSmooth(Smoother& smoother, Domain& v, Range& d)
474 {
475 smoother.template apply<false>(v,d);
476 }
477 };
478
479 template<class M, class X, class Y, class C, int l>
480 struct SmootherApplier<NonoverlappingBlockPreconditioner<C,SeqSOR<M,X,Y,l> > >
481 {
482 typedef NonoverlappingBlockPreconditioner<C,SeqSOR<M,X,Y,l> > Smoother;
483 typedef typename Smoother::range_type Range;
484 typedef typename Smoother::domain_type Domain;
485
486 static void preSmooth(Smoother& smoother, Domain& v, Range& d)
487 {
488 smoother.template apply<true>(v,d);
489 }
490
491
492 static void postSmooth(Smoother& smoother, Domain& v, Range& d)
493 {
494 smoother.template apply<false>(v,d);
495 }
496 };
497
498 } // end namespace Amg
499
500 // forward declarations
501 template<class M, class X, class MO, class MS, class A>
502 class SeqOverlappingSchwarz;
503
504 struct MultiplicativeSchwarzMode;
505
506 namespace Amg
507 {
508 template<class M, class X, class MS, class TA>
509 struct SmootherApplier<SeqOverlappingSchwarz<M,X,MultiplicativeSchwarzMode,
510 MS,TA> >
511 {
512 typedef SeqOverlappingSchwarz<M,X,MultiplicativeSchwarzMode,MS,TA> Smoother;
513 typedef typename Smoother::range_type Range;
514 typedef typename Smoother::domain_type Domain;
515
516 static void preSmooth(Smoother& smoother, Domain& v, const Range& d)
517 {
518 smoother.template apply<true>(v,d);
519 }
520
521
522 static void postSmooth(Smoother& smoother, Domain& v, const Range& d)
523 {
524 smoother.template apply<false>(v,d);
525
526 }
527 };
528
529 // template<class M, class X, class TM, class TA>
530 // class SeqOverlappingSchwarz;
531
532 template<class T>
533 struct SeqOverlappingSchwarzSmootherArgs
534 : public DefaultSmootherArgs<T>
535 {
536 enum Overlap {vertex, aggregate, pairwise, none};
537
538 Overlap overlap;
539 bool onthefly;
540
541 SeqOverlappingSchwarzSmootherArgs(Overlap overlap_=vertex,
542 bool onthefly_=false)
543 : overlap(overlap_), onthefly(onthefly_)
544 {}
545 };
546
547 template<class M, class X, class TM, class TS, class TA>
548 struct SmootherTraits<SeqOverlappingSchwarz<M,X,TM,TS,TA> >
549 {
550 typedef SeqOverlappingSchwarzSmootherArgs<typename M::field_type> Arguments;
551 };
552
553 template<class M, class X, class TM, class TS, class TA>
554 class ConstructionArgs<SeqOverlappingSchwarz<M,X,TM,TS,TA> >
555 : public DefaultConstructionArgs<SeqOverlappingSchwarz<M,X,TM,TS,TA> >
556 {
557 typedef DefaultConstructionArgs<SeqOverlappingSchwarz<M,X,TM,TS,TA> > Father;
558
559 public:
560 typedef typename MatrixGraph<M>::VertexDescriptor VertexDescriptor;
561 typedef Dune::Amg::AggregatesMap<VertexDescriptor> AggregatesMap;
562 typedef typename AggregatesMap::AggregateDescriptor AggregateDescriptor;
564 typedef typename Vector::value_type Subdomain;
565
566 virtual void setMatrix(const M& matrix, const AggregatesMap& amap)
567 {
568 Father::setMatrix(matrix);
569
570 std::vector<bool> visited(amap.noVertices(), false);
571 typedef IteratorPropertyMap<std::vector<bool>::iterator,IdentityMap> VisitedMapType;
572 VisitedMapType visitedMap(visited.begin());
573
574 MatrixGraph<const M> graph(matrix);
575
576 typedef SeqOverlappingSchwarzSmootherArgs<typename M::field_type> SmootherArgs;
577
578 switch(Father::getArgs().overlap) {
580 {
581 VertexAdder visitor(subdomains, amap);
582 createSubdomains(matrix, graph, amap, visitor, visitedMap);
583 }
584 break;
585 case SmootherArgs::pairwise :
586 {
587 createPairDomains(graph);
588 }
589 break;
590 case SmootherArgs::aggregate :
591 {
592 AggregateAdder<VisitedMapType> visitor(subdomains, amap, graph, visitedMap);
593 createSubdomains(matrix, graph, amap, visitor, visitedMap);
594 }
595 break;
596 case SmootherArgs::none :
597 NoneAdder visitor;
598 createSubdomains(matrix, graph, amap, visitor, visitedMap);
599 break;
600 default :
601 DUNE_THROW(NotImplemented, "This overlapping scheme is not supported!");
602 }
603 }
604 void setMatrix(const M& matrix)
605 {
606 Father::setMatrix(matrix);
607
608 /* Create aggregates map where each aggregate is just one vertex. */
609 AggregatesMap amap(matrix.N());
610 VertexDescriptor v=0;
611 for(typename AggregatesMap::iterator iter=amap.begin();
612 iter!=amap.end(); ++iter)
613 *iter=v++;
614
615 std::vector<bool> visited(amap.noVertices(), false);
616 typedef IteratorPropertyMap<std::vector<bool>::iterator,IdentityMap> VisitedMapType;
617 VisitedMapType visitedMap(visited.begin());
618
619 MatrixGraph<const M> graph(matrix);
620
621 typedef SeqOverlappingSchwarzSmootherArgs<typename M::field_type> SmootherArgs;
622
623 switch(Father::getArgs().overlap) {
625 {
626 VertexAdder visitor(subdomains, amap);
627 createSubdomains(matrix, graph, amap, visitor, visitedMap);
628 }
629 break;
630 case SmootherArgs::aggregate :
631 {
632 DUNE_THROW(NotImplemented, "Aggregate overlap is not supported yet");
633 /*
634 AggregateAdder<VisitedMapType> visitor(subdomains, amap, graph, visitedMap);
635 createSubdomains(matrix, graph, amap, visitor, visitedMap);
636 */
637 }
638 break;
639 case SmootherArgs::pairwise :
640 {
641 createPairDomains(graph);
642 }
643 break;
644 case SmootherArgs::none :
645 NoneAdder visitor;
646 createSubdomains(matrix, graph, amap, visitor, visitedMap);
647
648 }
649 }
650
651 const Vector& getSubDomains()
652 {
653 return subdomains;
654 }
655
656 private:
657 struct VertexAdder
658 {
659 VertexAdder(Vector& subdomains_, const AggregatesMap& aggregates_)
660 : subdomains(subdomains_), max(-1), subdomain(-1), aggregates(aggregates_)
661 {}
662 template<class T>
663 void operator()(const T& edge)
664 {
665 if(aggregates[edge.target()]!=AggregatesMap::ISOLATED)
666 subdomains[subdomain].insert(edge.target());
667 }
668 int setAggregate(const AggregateDescriptor& aggregate_)
669 {
670 subdomain=aggregate_;
671 max = std::max(subdomain, aggregate_);
672 return subdomain;
673 }
674 int noSubdomains() const
675 {
676 return max+1;
677 }
678 private:
679 Vector& subdomains;
680 AggregateDescriptor max;
681 AggregateDescriptor subdomain;
682 const AggregatesMap& aggregates;
683 };
684 struct NoneAdder
685 {
686 template<class T>
687 void operator()(const T& edge)
688 {}
689 int setAggregate(const AggregateDescriptor& aggregate_)
690 {
691 return -1;
692 }
693 int noSubdomains() const
694 {
695 return -1;
696 }
697 };
698
699 template<class VM>
700 struct AggregateAdder
701 {
702 AggregateAdder(Vector& subdomains_, const AggregatesMap& aggregates_,
703 const MatrixGraph<const M>& graph_, VM& visitedMap_)
704 : subdomains(subdomains_), subdomain(-1), aggregates(aggregates_),
705 adder(subdomains_, aggregates_), graph(graph_), visitedMap(visitedMap_)
706 {}
707 template<class T>
708 void operator()(const T& edge)
709 {
710 subdomains[subdomain].insert(edge.target());
711 // If we (the neighbouring vertex of the aggregate)
712 // are not isolated, add the aggregate we belong to
713 // to the same subdomain using the OneOverlapAdder
714 if(aggregates[edge.target()]!=AggregatesMap::ISOLATED) {
715 assert(aggregates[edge.target()]!=aggregate);
716 typename AggregatesMap::VertexList vlist;
717 aggregates.template breadthFirstSearch<true,false>(edge.target(), aggregate,
718 graph, vlist, adder, adder,
719 visitedMap);
720 }
721 }
722
723 int setAggregate(const AggregateDescriptor& aggregate_)
724 {
725 adder.setAggregate(aggregate_);
726 aggregate=aggregate_;
727 return ++subdomain;
728 }
729 int noSubdomains() const
730 {
731 return subdomain+1;
732 }
733
734 private:
735 AggregateDescriptor aggregate;
736 Vector& subdomains;
737 int subdomain;
738 const AggregatesMap& aggregates;
739 VertexAdder adder;
740 const MatrixGraph<const M>& graph;
741 VM& visitedMap;
742 };
743
744 void createPairDomains(const MatrixGraph<const M>& graph)
745 {
746 typedef typename MatrixGraph<const M>::ConstVertexIterator VIter;
747 typedef typename MatrixGraph<const M>::ConstEdgeIterator EIter;
748 typedef typename M::size_type size_type;
749
750 std::set<std::pair<size_type,size_type> > pairs;
751 int total=0;
752 for(VIter v=graph.begin(), ve=graph.end(); ve != v; ++v)
753 for(EIter e = v.begin(), ee=v.end(); ee!=e; ++e)
754 {
755 ++total;
756 if(e.source()<e.target())
757 pairs.insert(std::make_pair(e.source(),e.target()));
758 else
759 pairs.insert(std::make_pair(e.target(),e.source()));
760 }
761
762
763 subdomains.resize(pairs.size());
764 Dune::dinfo <<std::endl<< "Created "<<pairs.size()<<" ("<<total<<") pair domains"<<std::endl<<std::endl;
765 typedef typename std::set<std::pair<size_type,size_type> >::const_iterator SIter;
766 typename Vector::iterator subdomain=subdomains.begin();
767
768 for(SIter s=pairs.begin(), se =pairs.end(); se!=s; ++s)
769 {
770 subdomain->insert(s->first);
771 subdomain->insert(s->second);
772 ++subdomain;
773 }
774 std::size_t minsize=10000;
775 std::size_t maxsize=0;
776 int sum=0;
777 for(typename Vector::size_type i=0; i < subdomains.size(); ++i) {
778 sum+=subdomains[i].size();
779 minsize=std::min(minsize, subdomains[i].size());
780 maxsize=std::max(maxsize, subdomains[i].size());
781 }
782 Dune::dinfo<<"Subdomain size: min="<<minsize<<" max="<<maxsize<<" avg="<<(sum/subdomains.size())
783 <<" no="<<subdomains.size()<<std::endl;
784 }
785
786 template<class Visitor>
787 void createSubdomains(const M& matrix, const MatrixGraph<const M>& graph,
788 const AggregatesMap& amap, Visitor& overlapVisitor,
789 IteratorPropertyMap<std::vector<bool>::iterator,IdentityMap>& visitedMap )
790 {
791 // count number ag aggregates. We assume that the
792 // aggregates are numbered consecutively from 0 except
793 // for the isolated ones. All isolated vertices form
794 // one aggregate, here.
795 int isolated=0;
796 AggregateDescriptor maxAggregate=0;
797
798 for(std::size_t i=0; i < amap.noVertices(); ++i)
799 if(amap[i]==AggregatesMap::ISOLATED)
800 isolated++;
801 else
802 maxAggregate = std::max(maxAggregate, amap[i]);
803
804 subdomains.resize(maxAggregate+1+isolated);
805
806 // reset the subdomains
807 for(typename Vector::size_type i=0; i < subdomains.size(); ++i)
808 subdomains[i].clear();
809
810 // Create the subdomains from the aggregates mapping.
811 // For each aggregate we mark all entries and the
812 // neighbouring vertices as belonging to the same subdomain
813 VertexAdder aggregateVisitor(subdomains, amap);
814
815 for(VertexDescriptor i=0; i < amap.noVertices(); ++i)
816 if(!get(visitedMap, i)) {
817 AggregateDescriptor aggregate=amap[i];
818
819 if(amap[i]==AggregatesMap::ISOLATED) {
820 // isolated vertex gets its own aggregate
821 subdomains.push_back(Subdomain());
822 aggregate=subdomains.size()-1;
823 }
824 overlapVisitor.setAggregate(aggregate);
825 aggregateVisitor.setAggregate(aggregate);
826 subdomains[aggregate].insert(i);
827 typename AggregatesMap::VertexList vlist;
828 amap.template breadthFirstSearch<false,false>(i, aggregate, graph, vlist, aggregateVisitor,
829 overlapVisitor, visitedMap);
830 }
831
832 std::size_t minsize=10000;
833 std::size_t maxsize=0;
834 int sum=0;
835 for(typename Vector::size_type i=0; i < subdomains.size(); ++i) {
836 sum+=subdomains[i].size();
837 minsize=std::min(minsize, subdomains[i].size());
838 maxsize=std::max(maxsize, subdomains[i].size());
839 }
840 Dune::dinfo<<"Subdomain size: min="<<minsize<<" max="<<maxsize<<" avg="<<(sum/subdomains.size())
841 <<" no="<<subdomains.size()<<" isolated="<<isolated<<std::endl;
842
843
844
845 }
846 Vector subdomains;
847 };
848
849
850 template<class M, class X, class TM, class TS, class TA>
851 struct ConstructionTraits<SeqOverlappingSchwarz<M,X,TM,TS,TA> >
852 {
853 typedef ConstructionArgs<SeqOverlappingSchwarz<M,X,TM,TS,TA> > Arguments;
854
855 static inline std::shared_ptr<SeqOverlappingSchwarz<M,X,TM,TS,TA>> construct(Arguments& args)
856 {
857 return std::make_shared<SeqOverlappingSchwarz<M,X,TM,TS,TA>>
858 (args.getMatrix(),
859 args.getSubDomains(),
860 args.getArgs().relaxationFactor,
861 args.getArgs().onthefly);
862 }
863 };
864
865
866 } // namespace Amg
867} // namespace Dune
868
869
870
871#endif
Provides classes for the Coloring process of AMG.
Class providing information about the mapping of the vertices onto aggregates.
Definition: aggregates.hh:558
Construction Arguments for the default smoothers.
Definition: smoother.hh:91
VertexIteratorT< const MatrixGraph< Matrix > > ConstVertexIterator
The constant vertex iterator type.
Definition: graph.hh:306
M::size_type VertexDescriptor
The vertex descriptor.
Definition: graph.hh:71
EdgeIteratorT< const MatrixGraph< Matrix > > ConstEdgeIterator
The constant edge iterator type.
Definition: graph.hh:296
Block parallel preconditioner.
Definition: schwarz.hh:279
A parallel SSOR preconditioner.
Definition: schwarz.hh:176
Richardson preconditioner.
Definition: preconditioners.hh:711
Sequential ILU preconditioner.
Definition: preconditioners.hh:530
The sequential jacobian preconditioner.
Definition: preconditioners.hh:410
std::vector< subdomain_type, typename std::allocator_traits< TA >::template rebind_alloc< subdomain_type > > subdomain_vector
The vector type containing the subdomain to row index mapping.
Definition: overlappingschwarz.hh:795
Sequential SOR preconditioner.
Definition: preconditioners.hh:259
Sequential SSOR preconditioner.
Definition: preconditioners.hh:139
Helper classes for the construction of classes without empty constructor.
Type traits to determine the type of reals (when working with complex numbers)
#define DUNE_THROW(E, m)
Definition: exceptions.hh:216
constexpr GeometryType none(unsigned int dim)
Returns a GeometryType representing a singular of dimension dim.
Definition: type.hh:479
constexpr GeometryType vertex
GeometryType representing a vertex.
Definition: type.hh:504
void presmooth(LevelContext &levelContext, size_t steps)
Apply pre smoothing on the current level.
Definition: smoother.hh:404
V AggregateDescriptor
The aggregate descriptor type.
Definition: aggregates.hh:578
static const V ISOLATED
Identifier of isolated vertices.
Definition: aggregates.hh:569
DefaultSmootherArgs()
Default constructor.
Definition: smoother.hh:54
const void * Arguments
A type holding all the arguments needed to call the constructor.
Definition: construction.hh:42
static void postSmooth(Smoother &smoother, Domain &v, const Range &d)
apply post smoothing in forward direction
Definition: smoother.hh:392
static std::shared_ptr< T > construct(Arguments &args)
Construct an object with the specified arguments.
Definition: construction.hh:50
FieldTraits< T >::real_type RelaxationFactor
The type of the relaxation factor.
Definition: smoother.hh:40
SLList< VertexDescriptor, Allocator > VertexList
The type of a single linked list of vertex descriptors.
Definition: aggregates.hh:590
void postsmooth(LevelContext &levelContext, size_t steps)
Apply post smoothing on the current level.
Definition: smoother.hh:426
RelaxationFactor relaxationFactor
The relaxation factor to use.
Definition: smoother.hh:49
static void preSmooth(Smoother &smoother, Domain &v, const Range &d)
apply pre smoothing in forward direction
Definition: smoother.hh:380
int iterations
The numbe of iterations to perform.
Definition: smoother.hh:45
auto min(ADLTag< 0 >, const V &v1, const V &v2)
implements binary Simd::min()
Definition: defaults.hh:87
auto max(ADLTag< 0 >, const V &v1, const V &v2)
implements binary Simd::max()
Definition: defaults.hh:79
DInfoType dinfo(std::cout)
Stream for informative output.
Definition: stdstreams.hh:138
PartitionSet<... > Overlap
Type of PartitionSet for the overlap partition.
Definition: partitionset.hh:248
Dune namespace.
Definition: alignedallocator.hh:11
Define general preconditioner interface.
Traits class for generically constructing non default constructable types.
Definition: construction.hh:37
The default class for the smoother arguments.
Definition: smoother.hh:36
Helper class for applying the smoothers.
Definition: smoother.hh:368
Traits class for getting the attribute class of a smoother.
Definition: smoother.hh:64
Creative Commons License   |  Legal Statements / Impressum  |  Hosted by TU Dresden  |  generated with Hugo v0.111.3 (Dec 22, 23:30, 2024)