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
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084 | // *****************************************************************************
/*!
\file src/Inciter/Discretization.cpp
\copyright 2012-2015 J. Bakosi,
2016-2018 Los Alamos National Security, LLC.,
2019-2021 Triad National Security, LLC.,
2022-2024 J. Bakosi
All rights reserved. See the LICENSE file for details.
\details Data and functionality common to all discretization schemes
\see Discretization.h and Discretization.C for more info.
*/
// *****************************************************************************
#include <iomanip>
#include "Reorder.hpp"
#include "Vector.hpp"
#include "DerivedData.hpp"
#include "Discretization.hpp"
#include "MeshWriter.hpp"
#include "DiagWriter.hpp"
#include "InciterConfig.hpp"
#include "Print.hpp"
#include "Around.hpp"
#include "XystBuildConfig.hpp"
#include "Box.hpp"
namespace inciter {
static CkReduction::reducerType PDFMerger;
extern ctr::Config g_cfg;
} // inciter::
using inciter::Discretization;
Discretization::Discretization(
std::size_t meshid,
const CProxy_Transporter& transporter,
const tk::CProxy_MeshWriter& meshwriter,
const tk::UnsMesh::CoordMap& coordmap,
const tk::UnsMesh::Chunk& el,
const std::map< int, std::unordered_set< std::size_t > >& nodeCommMap,
int nc ) :
m_meshid( meshid ),
m_nchare( nc ),
m_it( 0 ),
m_itr( 0 ),
m_itf( 0 ),
m_initial( 1 ),
m_t( g_cfg.get< tag::t0 >() ),
m_lastDumpTime( -std::numeric_limits< tk::real >::max() ),
m_physFieldFloor( 0.0 ),
m_physHistFloor( 0.0 ),
m_physIntegFloor( 0.0 ),
m_rangeFieldFloor( g_cfg.get< tag::fieldout_range >().size(), 0.0 ),
m_rangeHistFloor( g_cfg.get< tag::histout_range >().size(), 0.0 ),
m_rangeIntegFloor( g_cfg.get< tag::integout_range >().size(), 0.0 ),
m_dt( g_cfg.get< tag::dt >() ),
m_dtn( m_dt ),
m_nvol( 0 ),
m_transporter( transporter ),
m_meshwriter( meshwriter ),
m_el( el ), // fills m_inpoel, m_gid, m_lid
m_coord( setCoord( coordmap ) ),
m_meshvol( 0.0 ),
m_v( m_gid.size(), 0.0 ),
m_vol( m_gid.size(), 0.0 ),
m_refined( 0 ),
m_prevstatus( std::chrono::high_resolution_clock::now() ),
m_nrestart( 0 ),
m_res( 0.0 ),
m_res0( 0.0 ),
m_res1( 0.0 ),
m_dea( 0 ),
m_deastarted( 0 ),
m_pit( 0 ),
m_mit( 0 )
// *****************************************************************************
// Constructor
//! \param[in] meshid Mesh ID
//! \param[in] transporter Host (Transporter) proxy
//! \param[in] meshwriter Mesh writer proxy
//! \param[in] coordmap Coordinates of mesh nodes and their global IDs
//! \param[in] el Elements of the mesh chunk we operate on
//! \param[in] nodeCommMap Node lists associated to chare IDs bordering the
//! mesh chunk we operate on
//! \param[in] nc Total number of Discretization chares
// *****************************************************************************
{
Assert( !m_inpoel.empty(), "No elements assigned to Discretization chare" );
Assert( tk::positiveJacobians( m_inpoel, m_coord ),
"Jacobian in input mesh to Discretization non-positive" );
#if not defined(__INTEL_COMPILER) || defined(NDEBUG)
// The above ifdef skips running the conformity test with the intel compiler
// in debug mode only. This is necessary because in tk::conforming(), filling
// up the map can fail with some meshes (only in parallel), e.g., tube.exo,
// used by some regression tests, due to the intel compiler generating some
// garbage incorrect code - only in debug, only in parallel, only with that
// mesh.
Assert( tk::conforming( m_inpoel, m_coord ),
"Input mesh to Discretization not conforming" );
#endif
// Store node communication map
for (const auto& [c,map] : nodeCommMap) m_nodeCommMap[c] = map;
// Get ready for computing/communicating nodal volumes
startvol();
// Find host elements of user-specified points where time histories are
// saved, and save the shape functions evaluated at the point locations
const auto& pt = g_cfg.get< tag::histout >();
for (std::size_t p=0; p<pt.size(); ++p) {
std::array< tk::real, 4 > N;
const auto& l = pt[p];
for (std::size_t e=0; e<m_inpoel.size()/4; ++e) {
if (tk::intet( m_coord, m_inpoel, l, e, N )) {
m_histdata.push_back( HistData{{ "p"+std::to_string(p+1), e, N }} );
break;
}
}
}
// Compute number of mesh points owned
std::size_t npoin = m_gid.size();<--- Shadow variable
for (auto g : m_gid) if (tk::slave( m_nodeCommMap, g, thisIndex ) ) --npoin;
// Tell the RTS that the Discretization chares have been created and compute
// the total number of mesh points across the distributed mesh
std::vector< std::size_t > meshdata{ m_meshid, npoin };
contribute( meshdata, CkReduction::sum_ulong,
CkCallback( CkReductionTarget(Transporter,disccreated), m_transporter ) );
}
void
Discretization::resizePostAMR(
const tk::UnsMesh::Chunk& chunk,
const tk::UnsMesh::Coords& coord,
const std::unordered_map< int, std::unordered_set< std::size_t > >&
nodeCommMap,
const std::set< std::size_t >& /*removedNodes*/ )
// *****************************************************************************
// Resize mesh data structures after mesh refinement
//! \param[in] chunk New mesh chunk (connectivity and global<->local id maps)
//! \param[in] coord New mesh node coordinates
//! \param[in] nodeCommMap New node communication map
//! \param[in] removedNodes Newly removed mesh node local ids
// *****************************************************************************
{
m_el = chunk; // updates m_inpoel, m_gid, m_lid
m_nodeCommMap = nodeCommMap;
// Update mesh volume container size
m_vol.resize( m_gid.size(), 0.0 );
// update mesh node coordinates
m_coord = coord;
}
void
Discretization::startvol()
// *****************************************************************************
// Get ready for (re-)computing/communicating nodal volumes
// *****************************************************************************
{
m_nvol = 0;
thisProxy[ thisIndex ].wait4vol();
// Zero out mesh volume container
std::fill( begin(m_vol), end(m_vol), 0.0 );
// Clear receive buffer that will be used for collecting nodal volumes
m_volc.clear();
}
void
Discretization::registerReducers()
// *****************************************************************************
// Configure Charm++ reduction types
//! \details Since this is a [initnode] routine, see the .ci file, the
//! Charm++ runtime system executes the routine exactly once on every
//! logical node early on in the Charm++ init sequence. Must be static as
//! it is called without an object. See also: Section "Initializations at
//! Program Startup" at in the Charm++ manual
//! http://charm.cs.illinois.edu/manuals/html/charm++/manual.html.
// *****************************************************************************
{
PDFMerger = CkReduction::addReducer( tk::mergeUniPDFs );
}
tk::UnsMesh::Coords
Discretization::setCoord( const tk::UnsMesh::CoordMap& coordmap )
// *****************************************************************************
// Set mesh coordinates based on coordinates map
// *****************************************************************************
{
Assert( coordmap.size() == m_gid.size(), "Size mismatch" );
Assert( coordmap.size() == m_lid.size(), "Size mismatch" );
tk::UnsMesh::Coords coord;
coord[0].resize( coordmap.size() );
coord[1].resize( coordmap.size() );
coord[2].resize( coordmap.size() );
for (const auto& [ gid, coords ] : coordmap) {
auto i = tk::cref_find( m_lid, gid );
coord[0][i] = coords[0];
coord[1][i] = coords[1];
coord[2][i] = coords[2];
}
return coord;
}
void
Discretization::remap(
const std::unordered_map< std::size_t, std::size_t >& map )
// *****************************************************************************
// Remap mesh data based on new local ids
//! \param[in] map Mapping of old->new local ids
// *****************************************************************************
{
// Remap connectivity containing local IDs
for (auto& l : m_inpoel) l = tk::cref_find(map,l);
// Remap global->local id map
for (auto& [g,l] : m_lid) l = tk::cref_find(map,l);
// Remap global->local id map
auto maxid = std::numeric_limits< std::size_t >::max();
std::vector< std::size_t > newgid( m_gid.size(), maxid );
for (const auto& [o,n] : map) newgid[n] = m_gid[o];
m_gid = std::move( newgid );
Assert( std::all_of( m_gid.cbegin(), m_gid.cend(),
[=](std::size_t i){ return i < maxid; } ),
"Not all gid have been remapped" );
// Remap nodal volumes (with contributions along chare-boundaries)
std::vector< tk::real > newvol( m_vol.size(), 0.0 );
for (const auto& [o,n] : map) newvol[n] = m_vol[o];
m_vol = std::move( newvol );
// Remap nodal volumes (without contributions along chare-boundaries)
std::vector< tk::real > newv( m_v.size(), 0.0 );
for (const auto& [o,n] : map) newv[n] = m_v[o];
m_v = std::move( newv );
// Remap locations of node coordinates
tk::UnsMesh::Coords newcoord;
auto npoin = m_coord[0].size();<--- Shadow variable
newcoord[0].resize( npoin );
newcoord[1].resize( npoin );
newcoord[2].resize( npoin );
for (const auto& [o,n] : map) {
newcoord[0][n] = m_coord[0][o];
newcoord[1][n] = m_coord[1][o];
newcoord[2][n] = m_coord[2][o];
}
m_coord = std::move( newcoord );
}
void
Discretization::setRefiner( const CProxy_Refiner& ref )
// *****************************************************************************
// Set Refiner Charm++ proxy
//! \param[in] ref Incoming refiner proxy to store
// *****************************************************************************
{
m_refiner = ref;
}
void
Discretization::vol()
// *****************************************************************************
// Sum mesh volumes to nodes, start communicating them on chare-boundaries
// *****************************************************************************
{
const auto& x = m_coord[0];
const auto& y = m_coord[1];
const auto& z = m_coord[2];
// Compute nodal volumes on our chunk of the mesh
for (std::size_t e=0; e<m_inpoel.size()/4; ++e) {
const auto N = m_inpoel.data() + e*4;<--- Variable 'N' can be declared as pointer to const
const std::array< tk::real, 3 >
ba{{ x[N[1]]-x[N[0]], y[N[1]]-y[N[0]], z[N[1]]-z[N[0]] }},
ca{{ x[N[2]]-x[N[0]], y[N[2]]-y[N[0]], z[N[2]]-z[N[0]] }},
da{{ x[N[3]]-x[N[0]], y[N[3]]-y[N[0]], z[N[3]]-z[N[0]] }};
const auto J = tk::triple( ba, ca, da ) / 24.0;
ErrChk( J > 0, "Element Jacobian non-positive: PE:" +
std::to_string(CkMyPe()) + ", node IDs: " +
std::to_string(m_gid[N[0]]) + ',' +
std::to_string(m_gid[N[1]]) + ',' +
std::to_string(m_gid[N[2]]) + ',' +
std::to_string(m_gid[N[3]]) + ", coords: (" +
std::to_string(x[N[0]]) + ", " +
std::to_string(y[N[0]]) + ", " +
std::to_string(z[N[0]]) + "), (" +
std::to_string(x[N[1]]) + ", " +
std::to_string(y[N[1]]) + ", " +
std::to_string(z[N[1]]) + "), (" +
std::to_string(x[N[2]]) + ", " +
std::to_string(y[N[2]]) + ", " +
std::to_string(z[N[2]]) + "), (" +
std::to_string(x[N[3]]) + ", " +
std::to_string(y[N[3]]) + ", " +
std::to_string(z[N[3]]) + ')' );
// scatter add V/4 to nodes
for (std::size_t j=0; j<4; ++j) m_vol[N[j]] += J;
}
// Store nodal volumes without contributions from other chares on
// chare-boundaries
m_v = m_vol;
// Send our nodal volume contributions to neighbor chares
if (m_nodeCommMap.empty()) {
comvol_complete();
} else {
for (const auto& [c,n] : m_nodeCommMap) {
std::vector< tk::real > v( n.size() );
std::size_t j = 0;
for (auto i : n) v[ j++ ] = m_vol[ tk::cref_find(m_lid,i) ];
thisProxy[c].comvol( thisIndex,
std::vector<std::size_t>(begin(n), end(n)), v );
}
}
ownvol_complete();
}
void
Discretization::comvol( int c,
const std::vector< std::size_t >& gid,
const std::vector< tk::real >& nodevol )
// *****************************************************************************
// Receive nodal volumes on chare-boundaries
//! \param[in] c Sender chare id
//! \param[in] gid Global mesh node IDs at which we receive volume contributions
//! \param[in] nodevol Partial sums of nodal volume contributions to
//! chare-boundary nodes
// *****************************************************************************
{
Assert( nodevol.size() == gid.size(), "Size mismatch" );
auto& cvolc = m_cvolc[c];
for (std::size_t i=0; i<gid.size(); ++i) {
m_volc[ gid[i] ] += nodevol[i];
cvolc[ tk::cref_find(m_lid,gid[i]) ] = nodevol[i];
}
if (++m_nvol == m_nodeCommMap.size()) {
m_nvol = 0;
comvol_complete();
}
}
void
Discretization::totalvol()
// *****************************************************************************
// Sum mesh volumes and contribute own mesh volume to total volume
// *****************************************************************************
{
// Add received contributions to nodal volumes
for (const auto& [gid, vol] : m_volc)<--- Shadow variable
m_vol[ tk::cref_find(m_lid,gid) ] += vol;
// Clear receive buffer
tk::destroy(m_volc);
// Sum mesh volume to host
std::vector< tk::real > tvol{ 0.0,
static_cast<tk::real>(m_initial),
static_cast<tk::real>(m_meshid) };
for (auto v : m_v) tvol[0] += v;
contribute( tvol, CkReduction::sum_double,
CkCallback(CkReductionTarget(Transporter,totalvol), m_transporter) );
}
void
Discretization::stat( tk::real mesh_volume )
// *****************************************************************************
// Compute mesh cell statistics
//! \param[in] mesh_volume Total mesh volume
// *****************************************************************************
{
// Store total mesh volume
m_meshvol = mesh_volume;
const auto& x = m_coord[0];
const auto& y = m_coord[1];
const auto& z = m_coord[2];
auto MIN = -std::numeric_limits< tk::real >::max();
auto MAX = std::numeric_limits< tk::real >::max();
std::vector< tk::real > min( 6, MAX );<--- Variable 'min' is assigned a value that is never used.
std::vector< tk::real > max( 6, MIN );<--- Variable 'max' is assigned a value that is never used.
std::vector< tk::real > sum( 9, 0.0 );<--- Variable 'sum' is assigned a value that is never used.
tk::UniPDF edgePDF( 1e-4 );
tk::UniPDF volPDF( 1e-4 );
tk::UniPDF ntetPDF( 1e-4 );
// Compute points surrounding points
auto psup = tk::genPsup( m_inpoel, 4, tk::genEsup(m_inpoel,4) );
Assert( psup.second.size()-1 == m_gid.size(),
"Number of mesh points and number of global IDs unequal" );
// Compute edge length statistics
// Note that while the min and max edge lengths are independent of the number
// of CPUs (by the time they are aggregated across all chares), the sum of
// the edge lengths and the edge length PDF are not. This is because the
// edges on the chare-boundary are counted multiple times and we
// conscientiously do not make an effort to precisely compute this, because
// that would require communication and more complex logic. Since these
// statistics are intended as simple average diagnostics, we ignore these
// small differences. For reproducible average edge lengths and edge length
// PDFs, run the mesh in serial.
tk::UnsMesh::EdgeSet edges;
for (std::size_t p=0; p<m_gid.size(); ++p) {
for (auto i : tk::Around(psup,p)) {
const auto dx = x[ i ] - x[ p ];
const auto dy = y[ i ] - y[ p ];
const auto dz = z[ i ] - z[ p ];
const auto length = std::sqrt( dx*dx + dy*dy + dz*dz );
if (length < min[0]) min[0] = length;
if (length > max[0]) max[0] = length;
sum[0] += 1.0;
sum[1] += length;
edgePDF.add( length );
edges.insert( { m_gid[i], m_gid[p] } );
}
}
// Compute mesh cell volume statistics
for (std::size_t e=0; e<m_inpoel.size()/4; ++e) {
const std::array< std::size_t, 4 > N{{ m_inpoel[e*4+0], m_inpoel[e*4+1],
m_inpoel[e*4+2], m_inpoel[e*4+3] }};
const std::array< tk::real, 3 >
ba{{ x[N[1]]-x[N[0]], y[N[1]]-y[N[0]], z[N[1]]-z[N[0]] }},
ca{{ x[N[2]]-x[N[0]], y[N[2]]-y[N[0]], z[N[2]]-z[N[0]] }},
da{{ x[N[3]]-x[N[0]], y[N[3]]-y[N[0]], z[N[3]]-z[N[0]] }};
const auto L = std::cbrt( tk::triple( ba, ca, da ) / 6.0 );
if (L < min[1]) min[1] = L;
if (L > max[1]) max[1] = L;
sum[2] += 1.0;
sum[3] += L;
volPDF.add( L );
}
// Contribute statistics
sum[4] = 1.0;
min[2] = max[2] = sum[5] = static_cast< tk::real >( m_inpoel.size() / 4 );
min[3] = max[3] = sum[6] = static_cast< tk::real >( m_gid.size() );
min[4] = max[4] = sum[7] = static_cast< tk::real >( edges.size() );
min[5] = max[5] = sum[8] =
static_cast< tk::real >( tk::sumvalsize(m_nodeCommMap) ) /
static_cast< tk::real >( m_gid.size() );
ntetPDF.add( min[2] );
min.push_back( static_cast<tk::real>(m_meshid) );
max.push_back( static_cast<tk::real>(m_meshid) );
sum.push_back( static_cast<tk::real>(m_meshid) );
// Contribute to mesh statistics across all Discretization chares
contribute( min, CkReduction::min_double,
CkCallback(CkReductionTarget(Transporter,minstat), m_transporter) );
contribute( max, CkReduction::max_double,
CkCallback(CkReductionTarget(Transporter,maxstat), m_transporter) );
contribute( sum, CkReduction::sum_double,
CkCallback(CkReductionTarget(Transporter,sumstat), m_transporter) );
// Serialize PDFs to raw stream
auto stream = tk::serialize( m_meshid, { edgePDF, volPDF, ntetPDF } );
// Create Charm++ callback function for reduction of PDFs with
// Transporter::pdfstat() as the final target where the results will appear.
CkCallback cb( CkIndex_Transporter::pdfstat(nullptr), m_transporter );
// Contribute serialized PDF of partial sums to host via Charm++ reduction
contribute( stream.first, stream.second.get(), PDFMerger, cb );
}
void
Discretization::boxvol()
// *****************************************************************************
// Compute total box IC volume
// *****************************************************************************
{
// Determine which nodes reside in user-defined IC box(es) if any
m_boxnodes = problems::boxnodes( m_coord );
// Compute partial box IC volume (just add up all boxes)
tk::real boxvol = 0.0;
// cppcheck-suppress useStlAlgorithm
for (const auto& b : m_boxnodes) for (auto i : b) boxvol += m_v[i];
// Sum up box IC volume across all chares
std::vector< tk::real > meshdata{ boxvol, static_cast<tk::real>(m_meshid) };
contribute( meshdata, CkReduction::sum_double,
CkCallback(CkReductionTarget(Transporter,boxvol), m_transporter) );
}
void
Discretization::write(
const std::vector< std::size_t >& inpoel,
const tk::UnsMesh::Coords& coord,
const std::map< int, std::vector< std::size_t > >& bface,
const std::map< int, std::vector< std::size_t > >& bnode,
const std::vector< std::size_t >& triinpoel,
const std::vector< std::string>& elemfieldnames,
const std::vector< std::string>& nodefieldnames,
const std::vector< std::string>& elemsurfnames,
const std::vector< std::string>& nodesurfnames,
const std::vector< std::vector< tk::real > >& elemfields,
const std::vector< std::vector< tk::real > >& nodefields,
const std::vector< std::vector< tk::real > >& elemsurfs,
const std::vector< std::vector< tk::real > >& nodesurfs,
CkCallback c )
// *****************************************************************************
// Output mesh and fields data (solution dump) to file(s)
//! \param[in] inpoel Mesh connectivity for the mesh chunk to be written
//! \param[in] coord Node coordinates of the mesh chunk to be written
//! \param[in] bface Map of boundary-face lists mapped to corresponding side set
//! ids for this mesh chunk
//! \param[in] bnode Map of boundary-node lists mapped to corresponding side set
//! ids for this mesh chunk
//! \param[in] triinpoel Interconnectivity of points and boundary-face in this
//! mesh chunk
//! \param[in] elemfieldnames Names of element fields to be output to file
//! \param[in] nodefieldnames Names of node fields to be output to file
//! \param[in] elemsurfnames Names of elem surface fields to be output to file
//! \param[in] nodesurfnames Names of node surface fields to be output to file
//! \param[in] elemfields Field data in mesh elements to output to file
//! \param[in] nodefields Field data in mesh nodes to output to file
//! \param[in] elemsurfs Surface field data in mesh elements to output to file
//! \param[in] nodesurfs Surface field data in mesh nodes to output to file
//! \param[in] c Function to continue with after the write
//! \details Since m_meshwriter is a Charm++ chare group, it never migrates and
//! an instance is guaranteed on every PE. We index the first PE on every
//! logical compute node. In Charm++'s non-SMP mode, a node is the same as a
//! PE, so the index is the same as CkMyPe(). In SMP mode the index is the
//! first PE on every logical node. In non-SMP mode this yields one or more
//! output files per PE with zero or non-zero virtualization, respectively. If
//! there are multiple chares on a PE, the writes are serialized per PE, since
//! only a single entry method call can be executed at any given time. In SMP
//! mode, still the same number of files are output (one per chare), but the
//! output is serialized through the first PE of each compute node. In SMP
//! mode, channeling multiple files via a single PE on each node is required
//! by NetCDF and HDF5, as well as ExodusII, since none of these libraries are
//! thread-safe.
// *****************************************************************************
{
// If the previous iteration refined (or moved) the mesh or this is called
// before the first time step, we also output the mesh.
bool meshoutput = m_itf == 0 ? true : false;
auto eps = std::numeric_limits< tk::real >::epsilon();
bool fieldoutput = false;
// Output field data only if there is no dump at this physical time yet
if (std::abs(m_lastDumpTime - m_t) > eps ) {
m_lastDumpTime = m_t;
++m_itf;
fieldoutput = true;
}
const auto& f = g_cfg.get< tag::fieldout >();
std::set< int > outsets( begin(f), end(f) );
m_meshwriter[ CkNodeFirst( CkMyNode() ) ].
write( m_meshid, meshoutput, fieldoutput, m_itr, m_itf, m_t, thisIndex,
g_cfg.get< tag::output >(),
inpoel, coord, bface, bnode, triinpoel, elemfieldnames,
nodefieldnames, elemsurfnames, nodesurfnames, elemfields, nodefields,
elemsurfs, nodesurfs, outsets, c );
}
void
Discretization::setdt( tk::real newdt )
// *****************************************************************************
// Set time step size
//! \param[in] newdt Size of the new time step
// *****************************************************************************
{
m_dtn = m_dt;
m_dt = newdt;
// Truncate the size of last time step
const auto term = g_cfg.get< tag::term >();
if (m_t+m_dt > term) m_dt = term - m_t;
}
void
Discretization::next()
// *****************************************************************************
// Prepare for next step
// *****************************************************************************
{
// Update floor of physics time divided by output interval times
const auto eps = std::numeric_limits< tk::real >::epsilon();
const auto ft = g_cfg.get< tag::fieldout_time >();
if (ft > eps) m_physFieldFloor = std::floor( m_t / ft );
const auto ht = g_cfg.get< tag::histout_time >();
if (ht > eps) m_physHistFloor = std::floor( m_t / ht );
const auto it = g_cfg.get< tag::integout_time >();
if (it > eps) m_physIntegFloor = std::floor( m_t / it );
// Update floors of physics time divided by output interval times for ranges
const auto& rf = g_cfg.get< tag::fieldout_range >();
for (std::size_t i=0; i<rf.size(); ++i) {
if (m_t > rf[i][0] and m_t < rf[i][1]) {
m_rangeFieldFloor[i] = std::floor( m_t / rf[i][2] );
}
}
const auto& rh = g_cfg.get< tag::histout_range >();
for (std::size_t i=0; i<rh.size(); ++i) {
if (m_t > rh[i][0] and m_t < rh[i][1]) {
m_rangeHistFloor[i] = std::floor( m_t / rh[i][2] );
}
}
const auto& ri = g_cfg.get< tag::integout_range >();
for (std::size_t i=0; i<ri.size(); ++i) {
if (m_t > ri[i][0] and m_t < ri[i][1]) {
m_rangeIntegFloor[i] = std::floor( m_t / ri[i][2] );
}
}
++m_it;
m_t += m_dt;
}
void
Discretization::grindZero()
// *****************************************************************************
// Zero grind-time
// *****************************************************************************
{
m_prevstatus = std::chrono::high_resolution_clock::now();
if (thisIndex == 0 && m_meshid == 0) {
tk::Print() << "Starting time stepping ...\n";
}
}
bool
Discretization::restarted( int nrestart )
// *****************************************************************************
// Detect if just returned from a checkpoint and if so, zero timers
//! \param[in] nrestart Number of times restarted
//! \return True if restart detected
// *****************************************************************************
{
// Detect if just restarted from checkpoint:
// nrestart == -1 if there was no checkpoint this step
// m_nrestart == nrestart if there was a checkpoint this step
// if both false, just restarted from a checkpoint
bool restarted = nrestart != -1 and m_nrestart != nrestart;
// If just restarted from checkpoint
if (restarted) {
// Update number of restarts
m_nrestart = nrestart;
// Start timer measuring time stepping wall clock time
m_timer.zero();
// Zero grind-timer
grindZero();
}
return restarted;
}
std::string
Discretization::histfilename( const std::string& id, std::streamsize precision )
// *****************************************************************************
// Construct history output filename
//! \param[in] id History point id
//! \param[in] precision Floating point precision to use for output
//! \return History file name
// *****************************************************************************
{
auto of = g_cfg.get< tag::output >();
std::stringstream ss;
ss << std::setprecision(static_cast<int>(precision)) << of << ".hist." << id;
return ss.str();
}
void
Discretization::histheader( std::vector< std::string >&& names )
// *****************************************************************************
// Output headers for time history files (one for each point)
//! \param[in] names History output variable names
// *****************************************************************************
{
for (const auto& h : m_histdata) {
auto prec = g_cfg.get< tag::histout_precision >();
tk::DiagWriter hw( histfilename( h.get< tag::id >(), prec ),
g_cfg.get< tag::histout_format >(),
prec );
hw.header( names );
}
}
void
Discretization::history( std::vector< std::vector< tk::real > >&& data )
// *****************************************************************************
// Output time history for a time step
//! \param[in] data Time history data for all variables and equations integrated
// *****************************************************************************
{
Assert( data.size() == m_histdata.size(), "Size mismatch" );
std::size_t i = 0;
for (const auto& h : m_histdata) {
auto prec = g_cfg.get< tag::histout_precision >();
tk::DiagWriter hw( histfilename( h.get< tag::id >(), prec ),
g_cfg.get< tag::histout_format >(),
prec,
std::ios_base::app );
hw.write( m_it, m_t, m_dt, data[i] );
++i;
}
}
bool
Discretization::fielditer() const
// *****************************************************************************
// Decide if field output iteration count interval is hit
//! \return True if field output iteration count interval is hit
// *****************************************************************************
{
if (g_cfg.get< tag::benchmark >()) return false;
return m_it % g_cfg.get< tag::fieldout_iter >() == 0;
}
bool
Discretization::fieldtime() const
// *****************************************************************************
// Decide if field output physics time interval is hit
//! \return True if field output physics time interval is hit
// *****************************************************************************
{
if (g_cfg.get< tag::benchmark >()) return false;
auto eps = std::numeric_limits< tk::real >::epsilon();
auto ft = g_cfg.get< tag::fieldout_time >();
if (ft < eps) return false;
return std::floor(m_t/ft) - m_physFieldFloor > eps;
}
bool
Discretization::fieldrange() const
// *****************************************************************************
// Decide if physics time falls into a field output time range
//! \return True if physics time falls into a field output time range
// *****************************************************************************
{
if (g_cfg.get< tag::benchmark >()) return false;
const auto eps = std::numeric_limits< tk::real >::epsilon();
bool output = false;
const auto& rf = g_cfg.get< tag::fieldout_range >();
for (std::size_t i=0; i<rf.size(); ++i) {
if (m_t > rf[i][0] and m_t < rf[i][1]) {
output |= std::floor(m_t/rf[i][2]) - m_rangeFieldFloor[i] > eps;
}
}
return output;
}
bool
Discretization::histiter() const
// *****************************************************************************
// Decide if history output iteration count interval is hit
//! \return True if history output iteration count interval is hit
// *****************************************************************************
{
if (g_cfg.get< tag::benchmark >()) return false;
auto hist = g_cfg.get< tag::histout_iter >();
const auto& hist_points = g_cfg.get< tag::histout >();
return m_it % hist == 0 and not hist_points.empty();
}
bool
Discretization::histtime() const
// *****************************************************************************
// Decide if history output physics time interval is hit
//! \return True if history output physics time interval is hit
// *****************************************************************************
{
if (g_cfg.get< tag::benchmark >()) return false;
auto eps = std::numeric_limits< tk::real >::epsilon();
auto ht = g_cfg.get< tag::histout_time >();
if (ht < eps) return false;
return std::floor(m_t/ht) - m_physHistFloor > eps;
}
bool
Discretization::histrange() const
// *****************************************************************************
// Decide if physics time falls into a history output time range
//! \return True if physics time falls into a history output time range
// *****************************************************************************
{
if (g_cfg.get< tag::benchmark >()) return false;
auto eps = std::numeric_limits< tk::real >::epsilon();
bool output = false;
const auto& rh = g_cfg.get< tag::histout_range >();
for (std::size_t i=0; i<rh.size(); ++i) {
if (m_t > rh[i][0] and m_t < rh[i][1]) {
output |= std::floor(m_t/rh[i][2]) - m_rangeHistFloor[i] > eps;
}
}
return output;
}
bool
Discretization::integiter() const
// *****************************************************************************
// Decide if integral output iteration count interval is hit
//! \return True if integral output iteration count interval is hit
// *****************************************************************************
{
if (g_cfg.get< tag::benchmark >()) return false;
auto integ = g_cfg.get< tag::integout_iter >();
const auto& sidesets_integral = g_cfg.get< tag::integout >();
return m_it % integ == 0 and not sidesets_integral.empty();
}
bool
Discretization::integtime() const
// *****************************************************************************
// Decide if integral output physics time interval is hit
//! \return True if integral output physics time interval is hit
// *****************************************************************************
{
if (g_cfg.get< tag::benchmark >()) return false;
auto eps = std::numeric_limits< tk::real >::epsilon();
auto it = g_cfg.get< tag::integout_time >();
if (it < eps) return false;
return std::floor(m_t/it) - m_physIntegFloor > eps;
}
bool
Discretization::integrange() const
// *****************************************************************************
// Decide if physics time falls into a integral output time range
//! \return True if physics time falls into a integral output time range
// *****************************************************************************
{
if (g_cfg.get< tag::benchmark >()) return false;
auto eps = std::numeric_limits< tk::real >::epsilon();
bool output = false;
const auto& ri = g_cfg.get< tag::integout_range >();
for (std::size_t i=0; i<ri.size(); ++i) {
if (m_t > ri[i][0] and m_t < ri[i][1]) {
output |= std::floor(m_t/ri[i][2]) - m_rangeIntegFloor[i] > eps;
}
}
return output;
}
bool
Discretization::finished() const
// *****************************************************************************
// Decide if this is the last time step
//! \return True if this is the last time step
// *****************************************************************************
{
auto eps = std::numeric_limits< tk::real >::epsilon();
auto nstep = g_cfg.get< tag::nstep >();
auto term = g_cfg.get< tag::term >();
auto residual = g_cfg.get< tag::residual >();<--- Shadow variable
return std::abs(m_t-term) < eps or m_it >= nstep or
(m_res > 0.0 and m_res < residual);
}
void
Discretization::residual( tk::real r )
// *****************************************************************************
// Update residual (during convergence to steady state)
//! \param[in] r Current residual
// *****************************************************************************
{
auto ttyi = g_cfg.get< tag::ttyi >();
if (m_it % ttyi == 0) {
m_res0 = m_res1;
m_res1 = r;
}
m_res = r;
}
bool
Discretization::deastart()
// *****************************************************************************
// Decide whether to start the deactivation procedure in this time step
//! \return True to start the deactivation prcedure in this time step
// *****************************************************************************
{
if (not m_deastarted and m_t > g_cfg.get<tag::deatime>() + m_dt) {
m_deastarted = 1;
return true;
}
return false;
}
bool
Discretization::deactivate() const
// *****************************************************************************
// Decide whether to run deactivation procedure in this time step
//! \return True to run deactivation prcedure in this time step
// *****************************************************************************
{
auto dea = g_cfg.get< tag::deactivate >();
auto deafreq = g_cfg.get< tag::deafreq >();
if (dea and !m_nodeCommMap.empty() and m_it % deafreq == 0)
return true;
else
return false;
}
void
Discretization::deactivated( int n )
// *****************************************************************************
// Receive deactivation report
//! \param[in] n Sum of deactivated chares
// *****************************************************************************
{
m_dea = n;
}
bool
Discretization::lb() const
// *****************************************************************************
// Decide whether to do load balancing this time step
//! \return True to do load balancing in this time step
// *****************************************************************************
{
auto lbfreq = g_cfg.get< tag::lbfreq >();
auto lbtime = g_cfg.get< tag::lbtime >();
if ((m_t > lbtime and m_it % lbfreq == 0) or m_it == 2)
return true;
else
return false;
}
void
Discretization::pit( std::size_t it )
// *****************************************************************************
// Update number of pressure linear solve iterations taken
//! \param[in] it Number of pressure linear solve iterations taken
// *****************************************************************************
{
m_pit = it;
}
void
Discretization::mit( std::size_t it )
// *****************************************************************************
// Update number of momentum/transport linear solve iterations taken
//! \param[in] it Number of momentum/transport linear solve iterations taken
// *****************************************************************************
{
m_mit = it;
}
void
Discretization::npoin( std::size_t n )
// *****************************************************************************
// Set number of mesh points (across all meshes)
//! \param[in] n Number of mesh points
// *****************************************************************************
{
m_npoin = n;
}
void
Discretization::status()
// *****************************************************************************
// Output one-liner status report
// *****************************************************************************
{
const auto ttyi = g_cfg.get< tag::ttyi >();
if (thisIndex == 0 and m_meshid == 0 and m_it % ttyi == 0) {
// estimate grind time (taken between this and the previous time step)
using std::chrono::duration_cast;
using us = std::chrono::microseconds;
using clock = std::chrono::high_resolution_clock;
auto grind_time = duration_cast< us >(clock::now() - m_prevstatus).count()
/ static_cast< long >( ttyi );
auto grind_perf = static_cast<tk::real>(grind_time)
/ static_cast<tk::real>(m_npoin);
m_prevstatus = clock::now();
const auto term = g_cfg.get< tag::term >();
const auto t0 = g_cfg.get< tag::t0 >();
const auto nstep = g_cfg.get< tag::nstep >();
const auto diag = g_cfg.get< tag::diag_iter >();
const auto rsfreq = g_cfg.get< tag::rsfreq >();
const auto benchmark = g_cfg.get< tag::benchmark >();
const auto residual = g_cfg.get< tag::residual >();<--- Shadow variable
const auto solver = g_cfg.get< tag::solver >();
const auto pre = solver == "chocg" ? 1 : 0;
const auto theta = g_cfg.get< tag::theta >();
const auto eps = std::numeric_limits< tk::real >::epsilon();
const auto mom = solver == "chocg" and theta > eps ? 1 : 0;
// estimate time elapsed and time for accomplishment
tk::Timer::Watch ete, eta;
m_timer.eta( term-t0, m_t-t0, nstep, m_it, m_res0, m_res1, residual,
ete, eta );
tk::Print print;
// Output one-liner
print << std::setfill(' ') << std::setw(8) << m_it << " "
<< std::scientific << std::setprecision(6)
<< std::setw(12) << m_t << " "
<< m_dt << " "
<< std::setfill('0')
<< std::setw(3) << ete.hrs.count() << ":"
<< std::setw(2) << ete.min.count() << ":"
<< std::setw(2) << ete.sec.count() << " "
<< std::setw(3) << eta.hrs.count() << ":"
<< std::setw(2) << eta.min.count() << ":"
<< std::setw(2) << eta.sec.count() << " "
<< std::scientific << std::setprecision(6) << std::setfill(' ')
<< std::setw(9) << grind_time << " "
<< std::setw(9) << grind_perf << " ";
// Augment one-liner status with output indicators
if (fielditer() or fieldtime() or fieldrange()) print << 'f';
if (not (m_it % diag)) print << 'd';
if (histiter() or histtime() or histrange()) print << 't';
if (integiter() or integtime() or integrange()) print << 'i';
if (m_refined) print << 'h';
if (lb() and not finished()) print << 'l';
if (not benchmark && (not (m_it % rsfreq) || finished())) print << 'c';
if (m_deastarted and deactivate()) {
print << "\te:" << m_dea << '/' << m_nchare;
}
if (pre) print << "\tp:" << m_pit;
if (mom) print << "\tm:" << m_mit;
print << '\n';
}
}
#include "NoWarning/discretization.def.h"
|