forked from CEED/Laghos
-
Notifications
You must be signed in to change notification settings - Fork 2
/
Copy pathlaghost_remhos.cpp
1567 lines (1378 loc) · 49.7 KB
/
laghost_remhos.cpp
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
#include "mfem.hpp"
#include <fstream>
#include <iostream>
#include "remhos_ho.hpp"
#include "remhos_lo.hpp"
#include "remhos_fct.hpp"
#include "remhos_mono.hpp"
#include "remhos_tools.hpp"
#include "remhos_sync.hpp"
// #include <unistd.h>
using namespace std;
using namespace mfem;
using std::cout;
using std::endl;
enum class HOSolverType {None, Neumann, CG, LocalInverse};
enum class FCTSolverType {None, FluxBased, ClipScale,
NonlinearPenalty, FCTProject};
enum class LOSolverType {None, DiscrUpwind, DiscrUpwindPrec,
ResDist, ResDistSubcell, MassBased};
enum class MonolithicSolverType {None, ResDistMono, ResDistMonoSubcell};
enum class TimeStepControl {FixedTimeStep, LOBoundsError};
// Choice for the problem setup. The fluid velocity, initial condition and
// inflow boundary condition are chosen based on this parameter.
int problem_num;
// 0 is standard transport.
// 1 is standard remap (mesh moves, solution is fixed).
int exec_mode;
// Velocity coefficient
void velocity_function(const Vector &x, Vector &v);
// Initial condition
double u0_function(const Vector &x);
double s0_function(const Vector &x);
// Inflow boundary condition
double inflow_function(const Vector &x);
// Mesh bounding box
Vector bb_min, bb_max;
// void attach_debugger(bool condition) {
// if (!condition) return;
// bool attached = false;
// // also write PID to a file
// ofstream os("/tmp/mpi_debug.pid");
// os << getpid() << endl;
// os.close();
// cout << "Waiting for debugger to be attached, PID: "
// << getpid() << endl;
// while (!attached) sleep(1);
// }
class AdvectionOperator : public TimeDependentOperator
{
private:
BilinearForm &Mbf, &ml;
ParBilinearForm &Kbf;
ParBilinearForm &M_HO, &K_HO;
Vector &lumpedM;
Vector start_mesh_pos, start_submesh_pos;
GridFunction &mesh_pos, *submesh_pos, &mesh_vel, &submesh_vel;
mutable ParGridFunction x_gf;
double dt;
TimeStepControl dt_control;
mutable double dt_est;
Assembly &asmbl;
LowOrderMethod &lom;
DofInfo &dofs;
HOSolver *ho_solver;
LOSolver *lo_solver;
FCTSolver *fct_solver;
MonolithicSolver *mono_solver;
void UpdateTimeStepEstimate(const Vector &x, const Vector &dx,
const Vector &x_min, const Vector &x_max) const;
public:
AdvectionOperator(int size, BilinearForm &Mbf_, BilinearForm &_ml,
Vector &_lumpedM,
ParBilinearForm &Kbf_,
ParBilinearForm &M_HO_, ParBilinearForm &K_HO_,
GridFunction &pos, GridFunction *sub_pos,
GridFunction &vel, GridFunction &sub_vel,
Assembly &_asmbl, LowOrderMethod &_lom, DofInfo &_dofs,
HOSolver *hos, LOSolver *los, FCTSolver *fct,
MonolithicSolver *mos);
virtual void Mult(const Vector &x, Vector &y) const;
void SetTimeStepControl(TimeStepControl tsc)
{
if (tsc == TimeStepControl::LOBoundsError)
{
MFEM_VERIFY(lo_solver,
"The selected time step control requires a LO solver.");
}
dt_control = tsc;
}
void SetDt(double _dt) { dt = _dt; dt_est = dt; }
double GetTimeStepEstimate() { return dt_est; }
void SetRemapStartPos(const Vector &m_pos, const Vector &sm_pos)
{
start_mesh_pos = m_pos;
start_submesh_pos = sm_pos;
}
virtual ~AdvectionOperator() { }
};
#ifdef MFEM_USE_MPI
void Remapping(ParMesh *pmesh, ParGridFunction &x, ParGridFunction &x_mod, ParGridFunction &u_gf, int &mesh_order, int &order, bool &pa, bool &ncmesh)
{
MPI_Comm comm = pmesh->GetComm();
int num_procs, myid;
MPI_Comm_size(comm, &num_procs);
MPI_Comm_rank(comm, &myid);
int ode_solver_type = 3;
HOSolverType ho_type = HOSolverType::LocalInverse;
LOSolverType lo_type = LOSolverType::DiscrUpwind;
FCTSolverType fct_type = FCTSolverType::FluxBased;
MonolithicSolverType mono_type = MonolithicSolverType::None;
// HOSolverType ho_type = HOSolverType::LocalInverse;
// LOSolverType lo_type = LOSolverType::MassBased;
// FCTSolverType fct_type = FCTSolverType::FluxBased;
// MonolithicSolverType mono_type = MonolithicSolverType::None;
// enum class HOSolverType {None, Neumann, CG, LocalInverse};
// enum class FCTSolverType {None, FluxBased, ClipScale,
// NonlinearPenalty, FCTProject};
// enum class LOSolverType {None, DiscrUpwind, DiscrUpwindPrec,
// ResDist, ResDistSubcell, MassBased};
// enum class MonolithicSolverType {None, ResDistMono, ResDistMonoSubcell};
// enum class TimeStepControl {FixedTimeStep, LOBoundsError};
if(ncmesh)
{
if(myid==0){std::cout << "ncmesh" << std::endl;}
ho_type = HOSolverType::LocalInverse;
lo_type = LOSolverType::None;
fct_type = FCTSolverType::None;
// lo_type = LOSolverType::MassBased;
// fct_type = FCTSolverType::FluxBased;
}
int bounds_type = 0;
// bool pa = false;
pa = false;
bool next_gen_full = false;
int smth_ind_type = 0;
double t_final = 1.0;
TimeStepControl dt_control = TimeStepControl::FixedTimeStep;
// double dt = 0.01;
double dt = 0.005;
bool visualization = false;
bool visit = false;
bool verify_bounds = false;
bool product_sync = false;
int vis_steps = 100;
// const char *device_config = "cpu";
int precision = 16;
int dim = pmesh->Dimension();
// cout.precision(precision);
exec_mode = 1; // Always standard remap
problem_num = 10;
// if (myid == 0) {std::cout << "standard remap (mesh moves, solution is fixed)" <<std::endl;}
// Only standard assembly in 1D (some mfem functions just abort in 1D).
if ((pa || next_gen_full) && dim == 1)
{
MFEM_WARNING("Disabling PA / FA for 1D.");
pa = false;
next_gen_full = false;
}
const int NE = pmesh->GetNE();
// Define the ODE solver used for time integration. Several explicit
// Runge-Kutta methods are available.
ODESolver *ode_solver = NULL;
switch (ode_solver_type)
{
case 1: ode_solver = new ForwardEulerSolver; break;
case 2: ode_solver = new RK2Solver(1.0); break;
case 3: ode_solver = new RK3SSPSolver; break;
case 4:
if (myid == 0) { MFEM_WARNING("RK4 may violate the bounds."); }
ode_solver = new RK4Solver; break;
case 6:
if (myid == 0) { MFEM_WARNING("RK6 may violate the bounds."); }
ode_solver = new RK6Solver; break;
default:
cout << "Unknown ODE solver type for remapping: " << ode_solver_type << '\n';
delete pmesh; MPI_Finalize();
}
// Check if the input mesh is periodic.
const bool periodic = pmesh->GetNodes() != NULL &&
dynamic_cast<const L2_FECollection *>
(pmesh->GetNodes()->FESpace()->FEColl()) != NULL;
if(ncmesh){pmesh->EnsureNCMesh(true);}
pmesh->SetCurvature(mesh_order, periodic);
FiniteElementCollection *mesh_fec;
if (periodic)
{
if (myid == 0) {std::cout << "input mesh is periodic" << std::endl;}
mesh_fec = new L2_FECollection(mesh_order, dim, BasisType::GaussLobatto);
}
else
{
if (myid == 0) {std::cout << "input mesh is not periodic" << std::endl;}
mesh_fec = new H1_FECollection(mesh_order, dim, BasisType::GaussLobatto);
}
// Current mesh positions.
ParFiniteElementSpace mesh_pfes(pmesh, mesh_fec, dim);
// pmesh->GetBoundingBox(bb_min, bb_max, max(order, 1));
// std::cout << bb_min.Size() << "," << bb_min[0] << "," << bb_min[1] << std::endl;
// std::cout << bb_max.Size() << "," << bb_max[0] << "," << bb_max[1] << std::endl;
bb_min.SetSize(dim);
bb_max.SetSize(dim);
bb_min[0] =0.0; bb_min[1] =0.0;
bb_max[0] =1e3; bb_max[1] =1e3;
// Store initial mesh positions.
Vector x0(x.Size());
x0 = x;
// Velocity for the problem. Depending on the execution mode, this is the
// advective velocity (transport) or mesh velocity (remap).
VectorFunctionCoefficient velocity(dim, velocity_function);
// // Mesh velocity.
// // If remap is on, obtain the mesh velocity by moving the mesh to the final
// // mesh positions, and taking the displacement vector.
// // The mesh motion resembles a time-dependent deformation, e.g., similar to
// // a deformation that is obtained by a Lagrangian simulation.
// GridFunction v_gf(x.FESpace());
// v_gf = 0.0;
// VectorGridFunctionCoefficient v_mesh_coeff(&v_gf);
// if (exec_mode == 1)
// {
// ParGridFunction v(&mesh_pfes);
// VectorFunctionCoefficient vcoeff(dim, velocity_function);
// v.ProjectCoefficient(vcoeff);
// double t = 0.0;
// while (t < t_final)
// {
// t += dt;
// // Move the mesh nodes.
// x.Add(std::min(dt, t_final-t), v);
// // Update the node velocities.
// v.ProjectCoefficient(vcoeff);
// }
// // Pseudotime velocity.
// add(x, -1.0, x0, v_gf);
// // Return the mesh to the initial configuration.
// x = x0;
// }
// Mesh velocity.
// If remap is on, obtain the mesh velocity by moving the mesh to the final
// mesh positions, and taking the displacement vector.
// The mesh motion resembles a time-dependent deformation, e.g., similar to
// a deformation that is obtained by a Lagrangian simulation.
GridFunction v_gf(x.FESpace());
VectorGridFunctionCoefficient v_mesh_coeff(&v_gf);
v_gf = x; v_gf -= x_mod;
v_gf.Neg();
// Define the discontinuous DG finite element space of the given
// polynomial order on the refined mesh.
const int btype = BasisType::Positive;
DG_FECollection fec(order, dim, btype);
ParFiniteElementSpace pfes(pmesh, &fec);
// Check for meaningful combinations of parameters.
const bool forced_bounds = lo_type != LOSolverType::None ||
mono_type != MonolithicSolverType::None;
if (forced_bounds)
{
MFEM_VERIFY(btype == 2,
"Monotonicity treatment requires Bernstein basis.");
if (order == 0)
{
// Disable monotonicity treatment for piecewise constants.
if (myid == 0)
{ mfem_warning("For -o 0, monotonicity treatment is disabled."); }
lo_type = LOSolverType::None;
fct_type = FCTSolverType::None;
mono_type = MonolithicSolverType::None;
}
}
const bool use_subcell_RD =
( lo_type == LOSolverType::ResDistSubcell ||
mono_type == MonolithicSolverType::ResDistMonoSubcell );
if (use_subcell_RD)
{
MFEM_VERIFY(order > 1, "Subcell schemes require FE order > 2.");
}
if (dt_control == TimeStepControl::LOBoundsError)
{
MFEM_VERIFY(bounds_type == 1, "Error: -dtc 1 requires -bt 1.");
}
const int prob_size = pfes.GlobalTrueVSize();
if (myid == 0) { cout << "Number of unknowns: " << prob_size << endl; }
// Fields related to inflow BC.
FunctionCoefficient inflow(inflow_function);
ParGridFunction inflow_gf(&pfes);
inflow_gf =0.0;
// inflow_gf.ProjectCoefficient(inflow);
// Set up the bilinear and linear forms corresponding to the DG
// discretization.
ParBilinearForm m(&pfes);
m.AddDomainIntegrator(new MassIntegrator);
ParBilinearForm M_HO(&pfes);
M_HO.AddDomainIntegrator(new MassIntegrator);
ParBilinearForm k(&pfes);
ParBilinearForm K_HO(&pfes);
if (exec_mode == 0)
{
k.AddDomainIntegrator(new ConvectionIntegrator(velocity, -1.0));
K_HO.AddDomainIntegrator(new ConvectionIntegrator(velocity, -1.0));
}
else if (exec_mode == 1)
{
k.AddDomainIntegrator(new ConvectionIntegrator(v_mesh_coeff)); // mesh velocity
K_HO.AddDomainIntegrator(new ConvectionIntegrator(v_mesh_coeff)); // mesh velocity
}
if (ho_type == HOSolverType::CG ||
ho_type == HOSolverType::LocalInverse ||
fct_type == FCTSolverType::FluxBased)
{
if (exec_mode == 0)
{
DGTraceIntegrator *dgt_i = new DGTraceIntegrator(velocity, 1.0, -0.5);
DGTraceIntegrator *dgt_b = new DGTraceIntegrator(velocity, 1.0, -0.5);
K_HO.AddInteriorFaceIntegrator(new TransposeIntegrator(dgt_i));
K_HO.AddBdrFaceIntegrator(new TransposeIntegrator(dgt_b));
}
else if (exec_mode == 1)
{
auto dgt_i = new DGTraceIntegrator(v_mesh_coeff, -1.0, -0.5);
auto dgt_b = new DGTraceIntegrator(v_mesh_coeff, -1.0, -0.5);
K_HO.AddInteriorFaceIntegrator(new TransposeIntegrator(dgt_i));
K_HO.AddBdrFaceIntegrator(new TransposeIntegrator(dgt_b));
}
K_HO.KeepNbrBlock(true);
}
if (pa)
{
M_HO.SetAssemblyLevel(AssemblyLevel::PARTIAL);
K_HO.SetAssemblyLevel(AssemblyLevel::PARTIAL);
}
if (next_gen_full)
{
K_HO.SetAssemblyLevel(AssemblyLevel::FULL);
}
M_HO.Assemble();
// std::cout <<"K_HO.Assemble(0);" << std::endl;
K_HO.Assemble(0);
// std::cout <<"K_HO.Assemble(1);" << std::endl;
if (pa == false)
{
M_HO.Finalize();
K_HO.Finalize(0);
}
// Compute the lumped mass matrix.
Vector lumpedM;
ParBilinearForm ml(&pfes);
ml.AddDomainIntegrator(new LumpedIntegrator(new MassIntegrator));
ml.Assemble();
ml.Finalize();
ml.SpMat().GetDiag(lumpedM);
m.Assemble();
m.Finalize();
int skip_zeros = 0;
k.Assemble(skip_zeros);
k.Finalize(skip_zeros);
// Store topological dof data.
DofInfo dofs(pfes, bounds_type);
// Precompute data required for high and low order schemes. This could be put
// into a separate routine. I am using a struct now because the various
// schemes require quite different information.
LowOrderMethod lom;
lom.subcell_scheme = use_subcell_RD;
lom.pk = NULL;
if (lo_type == LOSolverType::DiscrUpwind)
{
lom.smap = SparseMatrix_Build_smap(k.SpMat());
lom.D = k.SpMat();
if (exec_mode == 0)
{
ComputeDiscreteUpwindingMatrix(k.SpMat(), lom.smap, lom.D);
}
}
else if (lo_type == LOSolverType::DiscrUpwindPrec)
{
lom.pk = new ParBilinearForm(&pfes);
if (exec_mode == 0)
{
lom.pk->AddDomainIntegrator(
new PrecondConvectionIntegrator(velocity, -1.0) );
}
else if (exec_mode == 1)
{
lom.pk->AddDomainIntegrator(
new PrecondConvectionIntegrator(v_mesh_coeff) );
}
lom.pk->Assemble(skip_zeros);
lom.pk->Finalize(skip_zeros);
lom.smap = SparseMatrix_Build_smap(lom.pk->SpMat());
lom.D = lom.pk->SpMat();
if (exec_mode == 0)
{
ComputeDiscreteUpwindingMatrix(lom.pk->SpMat(), lom.smap, lom.D);
}
}
if (exec_mode == 1) { lom.coef = &v_mesh_coeff; }
else { lom.coef = &velocity; }
// Face integration rule.
const FaceElementTransformations *ft = pmesh->GetFaceElementTransformations(0);
const int el_order = pfes.GetFE(0)->GetOrder();
int ft_order = ft->Elem1->OrderW() + 2 * el_order;
if (pfes.GetFE(0)->Space() == FunctionSpace::Pk) { ft_order++; }
lom.irF = &IntRules.Get(ft->FaceGeom, ft_order);
DG_FECollection fec0(0, dim, btype);
DG_FECollection fec1(1, dim, btype);
ParMesh *subcell_mesh = NULL;
lom.SubFes0 = NULL;
lom.SubFes1 = NULL;
FiniteElementCollection *fec_sub = NULL;
ParFiniteElementSpace *pfes_sub = NULL;;
ParGridFunction *xsub = NULL;
ParGridFunction v_sub_gf;
VectorGridFunctionCoefficient v_sub_coef;
Vector x0_sub;
if (use_subcell_RD)
{
// Get a uniformly refined mesh.
const int btype = BasisType::ClosedUniform;
subcell_mesh = new ParMesh(ParMesh::MakeRefined(*pmesh, order, btype));
// Check if the mesh is periodic.
const L2_FECollection *L2_coll = dynamic_cast<const L2_FECollection *>
(pmesh->GetNodes()->FESpace()->FEColl());
if (L2_coll == NULL)
{
// Standard non-periodic mesh.
// Note that the fine mesh is always linear.
fec_sub = new H1_FECollection(1, dim, BasisType::ClosedUniform);
pfes_sub = new ParFiniteElementSpace(subcell_mesh, fec_sub, dim);
xsub = new ParGridFunction(pfes_sub);
subcell_mesh->SetCurvature(1);
subcell_mesh->SetNodalGridFunction(xsub);
}
else
{
// Periodic mesh - the node positions must be corrected after the call
// to the above Mesh constructor. Note that the fine mesh is always
// linear.
const bool disc_nodes = true;
subcell_mesh->SetCurvature(1, disc_nodes);
fec_sub = new L2_FECollection(1, dim, BasisType::ClosedUniform);
pfes_sub = new ParFiniteElementSpace(subcell_mesh, fec_sub, dim);
xsub = new ParGridFunction(pfes_sub);
subcell_mesh->SetNodalGridFunction(xsub);
GridFunction *coarse = pmesh->GetNodes();
InterpolationGridTransfer transf(*coarse->FESpace(), *pfes_sub);
transf.ForwardOperator().Mult(*coarse, *xsub);
}
lom.SubFes0 = new FiniteElementSpace(subcell_mesh, &fec0);
lom.SubFes1 = new FiniteElementSpace(subcell_mesh, &fec1);
// Submesh velocity.
v_sub_gf.SetSpace(pfes_sub);
v_sub_gf.ProjectCoefficient(velocity);
// v_sub_gf.ProjectCoefficient(v_mesh_coeff);
// Zero it out on boundaries (not moving boundaries).
Array<int> ess_bdr, ess_vdofs;
if (subcell_mesh->bdr_attributes.Size() > 0)
{
ess_bdr.SetSize(subcell_mesh->bdr_attributes.Max());
}
ess_bdr = 1;
xsub->ParFESpace()->GetEssentialVDofs(ess_bdr, ess_vdofs);
for (int i = 0; i < ess_vdofs.Size(); i++)
{
if (ess_vdofs[i] == -1) { v_sub_gf(i) = 0.0; }
}
v_sub_coef.SetGridFunction(&v_sub_gf);
// Store initial submesh positions.
x0_sub = *xsub;
// Integrator on the submesh.
if (exec_mode == 0)
{
lom.subcellCoeff = &velocity;
lom.VolumeTerms = new MixedConvectionIntegrator(velocity, -1.0);
}
else if (exec_mode == 1)
{
lom.subcellCoeff = &v_sub_coef;
lom.VolumeTerms = new MixedConvectionIntegrator(v_sub_coef);
}
}
else { subcell_mesh = pmesh; }
Assembly asmbl(dofs, lom, inflow_gf, pfes, subcell_mesh, exec_mode);
// Setup the initial conditions.
const int vsize = pfes.GetVSize();
Array<int> offset((product_sync) ? 3 : 2);
for (int i = 0; i < offset.Size(); i++) { offset[i] = i*vsize; }
BlockVector S(offset, Device::GetMemoryType());
// Primary scalar field is u.
ParGridFunction u(&pfes);
u.MakeRef(&pfes, S, offset[0]);
// FunctionCoefficient u0(u0_function);
// u.ProjectCoefficient(u0);
u = u_gf;
u.SyncAliasMemory(S);
// For the case of product remap, we also solve for s and u_s.
ParGridFunction s, us;
Array<bool> u_bool_el, u_bool_dofs;
if (product_sync)
{
s.SetSpace(&pfes);
ComputeBoolIndicators(pmesh->GetNE(), u, u_bool_el, u_bool_dofs);
BoolFunctionCoefficient sc(s0_function, u_bool_el);
s.ProjectCoefficient(sc);
us.MakeRef(&pfes, S, offset[1]);
double *h_us = us.HostWrite();
const double *h_u = u.HostRead();
const double *h_s = s.HostRead();
// Simple - we don't target conservation at initialization.
for (int i = 0; i < s.Size(); i++) { h_us[i] = h_u[i] * h_s[i]; }
us.SyncAliasMemory(S);
}
// Smoothness indicator.
SmoothnessIndicator *smth_indicator = NULL;
if (smth_ind_type)
{
smth_indicator = new SmoothnessIndicator(smth_ind_type, *subcell_mesh,
pfes, u, dofs);
}
// Setup of the high-order solver (if any).
HOSolver *ho_solver = NULL;
if (ho_type == HOSolverType::Neumann)
{
ho_solver = new NeumannHOSolver(pfes, m, k, lumpedM, asmbl);
}
else if (ho_type == HOSolverType::CG)
{
ho_solver = new CGHOSolver(pfes, M_HO, K_HO);
}
else if (ho_type == HOSolverType::LocalInverse)
{
ho_solver = new LocalInverseHOSolver(pfes, M_HO, K_HO);
}
// Setup the low order solver (if any).
LOSolver *lo_solver = NULL;
Array<int> lo_smap;
const bool time_dep = (exec_mode == 0) ? false : true;
if (lo_type == LOSolverType::DiscrUpwind)
{
lo_smap = SparseMatrix_Build_smap(k.SpMat());
lo_solver = new DiscreteUpwind(pfes, k.SpMat(), lo_smap,
lumpedM, asmbl, time_dep);
}
else if (lo_type == LOSolverType::DiscrUpwindPrec)
{
lo_smap = SparseMatrix_Build_smap(lom.pk->SpMat());
lo_solver = new DiscreteUpwind(pfes, lom.pk->SpMat(), lo_smap,
lumpedM, asmbl, time_dep);
}
else if (lo_type == LOSolverType::ResDist)
{
const bool subcell_scheme = false;
if (pa)
{
lo_solver = new PAResidualDistribution(pfes, k, asmbl, lumpedM,
subcell_scheme, time_dep);
if (exec_mode == 0)
{
const PAResidualDistribution *RD_ptr =
dynamic_cast<const PAResidualDistribution*>(lo_solver);
RD_ptr->SampleVelocity(FaceType::Interior);
RD_ptr->SampleVelocity(FaceType::Boundary);
RD_ptr->SetupPA(FaceType::Interior);
RD_ptr->SetupPA(FaceType::Boundary);
}
}
else
{
lo_solver = new ResidualDistribution(pfes, k, asmbl, lumpedM,
subcell_scheme, time_dep);
}
}
else if (lo_type == LOSolverType::ResDistSubcell)
{
const bool subcell_scheme = true;
if (pa)
{
lo_solver = new PAResidualDistributionSubcell(pfes, k, asmbl, lumpedM,
subcell_scheme, time_dep);
if (exec_mode == 0)
{
const PAResidualDistributionSubcell *RD_ptr =
dynamic_cast<const PAResidualDistributionSubcell*>(lo_solver);
RD_ptr->SampleVelocity(FaceType::Interior);
RD_ptr->SampleVelocity(FaceType::Boundary);
RD_ptr->SetupPA(FaceType::Interior);
RD_ptr->SetupPA(FaceType::Boundary);
}
}
else
{
lo_solver = new ResidualDistribution(pfes, k, asmbl, lumpedM,
subcell_scheme, time_dep);
}
}
else if (lo_type == LOSolverType::MassBased)
{
MFEM_VERIFY(ho_solver != nullptr,
"Mass-Based LO solver requires a choice of a HO solver.");
lo_solver = new MassBasedAvg(pfes, *ho_solver,
(exec_mode == 1) ? &v_gf : nullptr);
}
// Setup of the monolithic solver (if any).
MonolithicSolver *mono_solver = NULL;
bool mass_lim = (problem_num != 6 && problem_num != 7) ? true : false;
if (mono_type == MonolithicSolverType::ResDistMono)
{
const bool subcell_scheme = false;
mono_solver = new MonoRDSolver(pfes, k.SpMat(), m.SpMat(), lumpedM,
asmbl, smth_indicator, velocity,
subcell_scheme, time_dep, mass_lim);
}
else if (mono_type == MonolithicSolverType::ResDistMonoSubcell)
{
const bool subcell_scheme = true;
mono_solver = new MonoRDSolver(pfes, k.SpMat(), m.SpMat(), lumpedM,
asmbl, smth_indicator, velocity,
subcell_scheme, time_dep, mass_lim);
}
// Print the starting meshes and initial condition.
if(visualization)
{
ofstream meshHO("meshHO_init.mesh");
meshHO.precision(precision);
pmesh->PrintAsOne(meshHO);
if (subcell_mesh)
{
ofstream meshLO("meshLO_init.mesh");
meshLO.precision(precision);
subcell_mesh->PrintAsOne(meshLO);
}
ofstream sltn("sltn_init.gf");
sltn.precision(precision);
u.SaveAsOne(sltn);
}
// Create data collection for solution output: either VisItDataCollection for
// ASCII data files, or SidreDataCollection for binary data files.
DataCollection *dc = NULL;
if (visit)
{
dc = new VisItDataCollection("Remhos", pmesh);
dc->SetPrecision(precision);
dc->RegisterField("solution", &u);
dc->SetCycle(0);
dc->SetTime(0.0);
dc->Save();
}
socketstream sout, vis_s, vis_us;
char vishost[] = "localhost";
int visport = 19916;
if (visualization)
{
// Make sure all MPI ranks have sent their 'v' solution before initiating
// another set of GLVis connections (one from each rank):
MPI_Barrier(pmesh->GetComm());
sout.precision(8);
vis_s.precision(8);
vis_us.precision(8);
int Wx = 0, Wy = 0; // window position
const int Ww = 400, Wh = 400; // window size
u.HostRead();
s.HostRead();
VisualizeField(sout, vishost, visport, u, "Solution u", Wx, Wy, Ww, Wh);
if (product_sync)
{
VisualizeField(vis_s, vishost, visport, s, "Solution s",
Wx + Ww, Wy, Ww, Wh);
VisualizeField(vis_us, vishost, visport, us, "Solution u_s",
Wx + 2*Ww, Wy, Ww, Wh);
}
}
// Record the initial mass.
Vector masses(lumpedM);
const double mass0_u_loc = lumpedM * u;
double mass0_u, mass0_us;
MPI_Allreduce(&mass0_u_loc, &mass0_u, 1, MPI_DOUBLE, MPI_SUM, comm);
if (product_sync)
{
const double mass0_us_loc = lumpedM * us;
MPI_Allreduce(&mass0_us_loc, &mass0_us, 1, MPI_DOUBLE, MPI_SUM, comm);
}
// Setup of the FCT solver (if any).
Array<int> K_HO_smap;
FCTSolver *fct_solver = NULL;
if (fct_type == FCTSolverType::FluxBased)
{
MFEM_VERIFY(pa == false, "Flux-based FCT and PA are incompatible.");
K_HO.SpMat().HostReadI();
K_HO.SpMat().HostReadJ();
K_HO.SpMat().HostReadData();
K_HO_smap = SparseMatrix_Build_smap(K_HO.SpMat());
const int fct_iterations = 1;
fct_solver = new FluxBasedFCT(pfes, smth_indicator, dt, K_HO.SpMat(),
K_HO_smap, M_HO.SpMat(), fct_iterations);
}
else if (fct_type == FCTSolverType::ClipScale)
{
fct_solver = new ClipScaleSolver(pfes, smth_indicator, dt);
}
else if (fct_type == FCTSolverType::NonlinearPenalty)
{
fct_solver = new NonlinearPenaltySolver(pfes, smth_indicator, dt);
}
else if (fct_type == FCTSolverType::FCTProject)
{
fct_solver = new ElementFCTProjection(pfes, dt);
}
AdvectionOperator adv(S.Size(), m, ml, lumpedM, k, M_HO, K_HO,
x, xsub, v_gf, v_sub_gf, asmbl, lom, dofs,
ho_solver, lo_solver, fct_solver, mono_solver);
double t = 0.0;
adv.SetTime(t);
adv.SetTimeStepControl(dt_control);
ode_solver->Init(adv);
double umin, umax;
GetMinMax(u, umin, umax);
if (exec_mode == 1)
{
adv.SetRemapStartPos(x0, x0_sub);
// For remap, the pseudo-time always evolves from 0 to 1.
t_final = 1.0;
}
ParGridFunction res = u;
double residual = 0.0;
double s_min_glob = numeric_limits<double>::infinity(),
s_max_glob = -numeric_limits<double>::infinity();
// Time-integration (loop over the time iterations, ti, with a time-step dt).
bool done = false;
BlockVector Sold(S);
int ti_total = 0, ti = 0;
while (done == false)
{
double dt_real = min(dt, t_final - t);
// This also resets the time step estimate when automatic dt is on.
adv.SetDt(dt_real);
if (lo_solver) { lo_solver->UpdateTimeStep(dt_real); }
if (fct_solver) { fct_solver->UpdateTimeStep(dt_real); }
if (product_sync)
{
ComputeMinMaxS(pmesh->GetNE(), us, u, s_min_glob, s_max_glob);
#ifdef REMHOS_FCT_PRODUCT_DEBUG
if (myid == 0)
{
std::cout << " --- Full time step" << std::endl;
std::cout << " in: ";
std::cout << std::scientific << std::setprecision(5);
std::cout << "min_s: " << s_min_glob
<< "; max_s: " << s_max_glob << std::endl;
}
#endif
}
Sold = S;
ode_solver->Step(S, t, dt_real);
ti++;
ti_total++;
if (dt_control != TimeStepControl::FixedTimeStep)
{
double dt_est = adv.GetTimeStepEstimate();
if (dt_est < dt_real)
{
// Repeat with the proper time step.
if (myid == 0)
{
cout << "Repeat / decrease dt: "
<< dt_real << " --> " << 0.85 * dt << endl;
}
ti--;
t -= dt_real;
S = Sold;
dt = 0.85 * dt;
if (dt < 1e-12) { MFEM_ABORT("The time step crashed!"); }
continue;
}
else if (dt_est > 1.25 * dt_real) { dt *= 1.02; }
}
// S has been modified, update the alias
u.SyncMemory(S);
if (product_sync)
{
us.SyncMemory(S);
// It is known that RK time integrators with more than 1 stage may
// cause violation of the lower bounds for us.
// The lower bound is corrected, causing small conservation error.
// Correction can also be done with localized bounds for s, but for
// now we have implemented only the minimum global bound.
u.HostRead();
us.HostReadWrite();
const int s = u.Size();
Array<bool> active_elem, active_dofs;
ComputeBoolIndicators(NE, u, active_elem, active_dofs);
for (int i = 0; i < s; i++)
{
if (active_dofs[i] == false) { continue; }
double us_min = u(i) * s_min_glob;
if (us(i) < us_min) { us(i) = us_min; }
}
#ifdef REMHOS_FCT_PRODUCT_DEBUG
ComputeMinMaxS(NE, us, u, s_min_glob, s_max_glob);
if (myid == 0)
{
std::cout << " out: ";
std::cout << std::scientific << std::setprecision(5);
std::cout << "min_s: " << s_min_glob
<< "; max_s: " << s_max_glob << std::endl;
}
#endif
}
// Monotonicity check for debug purposes mainly.
if (verify_bounds && forced_bounds && smth_indicator == NULL)
{
double umin_new, umax_new;
GetMinMax(u, umin_new, umax_new);
if (problem_num % 10 != 6 && problem_num % 10 != 7)
{
if (myid == 0)
{
MFEM_VERIFY(umin_new > umin - 1e-12,
"Undershoot of " << umin - umin_new);
MFEM_VERIFY(umax_new < umax + 1e-12,
"Overshoot of " << umax_new - umax);
}
umin = umin_new;
umax = umax_new;
}
else
{
if (myid == 0)
{
MFEM_VERIFY(umin_new > 0.0 - 1e-12,
"Undershoot of " << 0.0 - umin_new);
MFEM_VERIFY(umax_new < 1.0 + 1e-12,
"Overshoot of " << umax_new - 1.0);
}
}
}
if (exec_mode == 1)
{
x0.HostReadWrite(); v_sub_gf.HostReadWrite();
x.HostReadWrite();
add(x0, t, v_gf, x);
if (use_subcell_RD)
{
x0_sub.HostReadWrite(); v_sub_gf.HostReadWrite();
MFEM_VERIFY(xsub != NULL, "Subcell mesh not defined!");
xsub->HostReadWrite();
add(x0_sub, t, v_sub_gf, *xsub);
}
}
if (problem_num != 6 && problem_num != 7 && problem_num != 8)
{
done = (t >= t_final - 1.e-8*dt);
}
else
{
// Steady state problems - stop at convergence.
double res_loc = 0.;
lumpedM.HostReadWrite(); u.HostReadWrite(); res.HostReadWrite();
for (int i = 0; i < res.Size(); i++)
{
res_loc += pow( (lumpedM(i) * u(i) / dt) - (lumpedM(i) * res(i) / dt), 2. );
}
MPI_Allreduce(&res_loc, &residual, 1, MPI_DOUBLE, MPI_SUM, comm);
residual = sqrt(residual);
if (residual < 1.e-12 && t >= 1.) { done = true; u = res; }
else { res = u; }
}
// if (done || ti % vis_steps == 0)
if (vis_steps == 0)
{
if (myid == 0)
{
cout << "time step: " << ti << ", time: " << t
<< ", dt: " << dt << ", residual: " << residual << endl;
}
if (visualization)
{
int Wx = 0, Wy = 0; // window position