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
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
|
/*
* phylokernelnonrev.h
* Kernel based on vectorizing over alignment patterns for non-reversible models
*
* Created on: Nov 4, 2016
* Author: minh
*/
#if !defined(PHYLOKERNELNONREV_H_) || !defined(PHYLOKERNELNONREV_STATE_H_)
#ifdef KERNEL_FIX_STATES
#define PHYLOKERNELNONREV_STATE_H_
#else
#define PHYLOKERNELNONREV_H_
#endif
#include "phylotree.h"
#ifdef _OPENMP
#include <omp.h>
#endif
//#include <thread>
using namespace std;
#ifdef KERNEL_FIX_STATES
template <class VectorClass, const int nstates, const bool FMA>
void PhyloTree::computeNonrevPartialLikelihoodSIMD(TraversalInfo &info, size_t ptn_lower, size_t ptn_upper, int thread_id) {
#else
template <class VectorClass, const bool FMA>
void PhyloTree::computeNonrevPartialLikelihoodGenericSIMD(TraversalInfo &info, size_t ptn_lower, size_t ptn_upper, int thread_id) {
#endif
PhyloNeighbor *dad_branch = info.dad_branch;
PhyloNode *dad = info.dad;
ASSERT(dad);
PhyloNode *node = (PhyloNode*)(dad_branch->node);
// assert(dad_branch->direction != UNDEFINED_DIRECTION);
#ifndef KERNEL_FIX_STATES
size_t nstates = aln->num_states;
#endif
if (node->isLeaf()) {
return;
}
ASSERT(node->degree() >= 3);
size_t ptn, c;
size_t orig_nptn = aln->size();
size_t max_orig_nptn = ((orig_nptn+VectorClass::size()-1)/VectorClass::size())*VectorClass::size();
size_t nptn = max_orig_nptn+model_factory->unobserved_ptns.size();
size_t ncat = site_rate->getNRate();
size_t ncat_mix = (model_factory->fused_mix_rate) ? ncat : ncat*model->getNMixtures();
size_t i, x;
size_t block = nstates * ncat_mix;
// internal node
PhyloNeighbor *left = NULL, *right = NULL; // left & right are two neighbors leading to 2 subtrees
FOR_NEIGHBOR_IT(node, dad, it) {
if (!left) left = (PhyloNeighbor*)(*it); else right = (PhyloNeighbor*)(*it);
}
// precomputed buffer to save times
double *buffer_partial_lh_ptr = buffer_partial_lh + (getBufferPartialLhSize() - 2*block*VectorClass::size()*num_threads);
double *echildren = info.echildren;
double *partial_lh_leaves = info.partial_lh_leaves;
double *eleft = echildren, *eright = echildren + block*nstates;
if ((!left->node->isLeaf() && right->node->isLeaf())) {
PhyloNeighbor *tmp = left;
left = right;
right = tmp;
double *etmp = eleft;
eleft = eright;
eright = etmp;
}
if (node->degree() > 3) {
/*--------------------- multifurcating node ------------------*/
double *vec_tip = buffer_partial_lh_ptr + (block*2)*VectorClass::size()*thread_id;
VectorClass *vtip = (VectorClass*)vec_tip;
// now for-loop computing partial_lh over all site-patterns
for (ptn = ptn_lower; ptn < ptn_upper; ptn+=VectorClass::size()) {
VectorClass *partial_lh_all = (VectorClass*)(dad_branch->partial_lh + ptn*block);
for (i = 0; i < block; i++)
partial_lh_all[i] = 1.0;
memset(&dad_branch->scale_num[ptn], 0, sizeof(UBYTE)*VectorClass::size());
double *partial_lh_leaf = partial_lh_leaves;
double *echild = echildren;
FOR_NEIGHBOR_IT(node, dad, it) {
PhyloNeighbor *child = (PhyloNeighbor*)*it;
if (child->node->isLeaf()) {
// external node
// load data for tip
for (x = 0; x < VectorClass::size(); x++) {
double *tip_child;
if (isRootLeaf(child->node))
tip_child = partial_lh_leaf;
else if (ptn+x < orig_nptn)
tip_child = partial_lh_leaf + block * (aln->at(ptn+x))[child->node->id];
else if (ptn+x < max_orig_nptn)
tip_child = partial_lh_leaf + block * aln->STATE_UNKNOWN;
else if (ptn+x < nptn)
tip_child = partial_lh_leaf + block * model_factory->unobserved_ptns[ptn+x-max_orig_nptn];
else
tip_child = partial_lh_leaf + block * aln->STATE_UNKNOWN;
double *this_vec_tip = vec_tip+x;
for (i = 0; i < block; i++) {
*this_vec_tip = tip_child[i];
this_vec_tip += VectorClass::size();
}
}
for (c = 0; c < block; c++) {
// compute real partial likelihood vector
partial_lh_all[c] *= vtip[c];
}
partial_lh_leaf += (aln->STATE_UNKNOWN+1)*block;
} else {
// internal node
VectorClass *partial_lh = partial_lh_all;
VectorClass *partial_lh_child = (VectorClass*)(child->partial_lh + ptn*block);
for (i = 0; i < VectorClass::size(); i++)
dad_branch->scale_num[ptn+i] += child->scale_num[ptn+i];
double *echild_ptr = echild;
for (c = 0; c < ncat_mix; c++) {
// compute real partial likelihood vector
for (x = 0; x < nstates; x++) {
VectorClass vchild;
// for (i = 0; i < nstates; i++) {
// vchild += echild_ptr[i] * partial_lh_child[i];
// }
#ifdef KERNEL_FIX_STATES
dotProductVec<VectorClass, double, nstates, FMA>(echild_ptr, partial_lh_child, vchild);
#else
dotProductVec<VectorClass, double, FMA>(echild_ptr, partial_lh_child, vchild, nstates);
#endif
echild_ptr += nstates;
partial_lh[x] *= vchild;
}
partial_lh += nstates;
partial_lh_child += nstates;
}
} // if
echild += block*nstates;
} // FOR_NEIGHBOR
VectorClass lh_max = partial_lh_all[0];
for (i = 1; i < block; i++)
lh_max = max(lh_max, partial_lh_all[i]);
// check if one should scale partial likelihoods
auto underflown = (lh_max < SCALING_THRESHOLD);
if (horizontal_or(underflown)) {
// now do the likelihood scaling
for (x = 0; x < VectorClass::size(); x++)
if (underflown[x]) {
double *partial_lh = dad_branch->partial_lh + (ptn*block + x);
// now do the likelihood scaling
for (i = 0; i < block; i++) {
partial_lh[i*VectorClass::size()] = ldexp(partial_lh[i*VectorClass::size()], SCALING_THRESHOLD_EXP);
}
dad_branch->scale_num[ptn+x] += 1;
}
}
} // for ptn
// end multifurcating treatment
} else if (left->node->isLeaf() && right->node->isLeaf()) {
/*--------------------- TIP-TIP (cherry) case ------------------*/
double *partial_lh_left = partial_lh_leaves;
double *partial_lh_right = partial_lh_leaves + (aln->STATE_UNKNOWN+1)*block;
double *vec_left = buffer_partial_lh_ptr + (block*2)*VectorClass::size()*thread_id;
double *vec_right = &vec_left[block*VectorClass::size()];
if (isRootLeaf(right->node)) {
// swap so that left node is the root
PhyloNeighbor *tmp = left;
left = right;
right = tmp;
double *etmp = eleft;
eleft = eright;
eright = etmp;
etmp = partial_lh_left;
partial_lh_left = partial_lh_right;
partial_lh_right = etmp;
}
// scale number must be ZERO
memset(dad_branch->scale_num + ptn_lower, 0, (ptn_upper-ptn_lower) * sizeof(UBYTE));
if (isRootLeaf(left->node)) {
for (ptn = ptn_lower; ptn < ptn_upper; ptn+=VectorClass::size()) {
double *vright = dad_branch->partial_lh + ptn*block;
VectorClass *partial_lh = (VectorClass*)vright;
// load data for tip
for (x = 0; x < VectorClass::size(); x++) {
double *tip_right;
if (ptn+x < orig_nptn)
tip_right = partial_lh_right + block * (aln->at(ptn+x))[right->node->id];
else if (ptn+x < max_orig_nptn)
tip_right = partial_lh_right + block * aln->STATE_UNKNOWN;
else if (ptn+x < nptn)
tip_right = partial_lh_right + block * model_factory->unobserved_ptns[ptn+x-max_orig_nptn];
else
tip_right = partial_lh_right + block * aln->STATE_UNKNOWN;
double *this_vec_right = vright+x;
for (i = 0; i < block; i++) {
*this_vec_right = tip_right[i];
this_vec_right += VectorClass::size();
}
}
for (i = 0; i < block; i++)
partial_lh[i] *= partial_lh_left[i];
}
} else
for (ptn = ptn_lower; ptn < ptn_upper; ptn+=VectorClass::size()) {
VectorClass *partial_lh = (VectorClass*)(dad_branch->partial_lh + ptn*block);
VectorClass *vleft = (VectorClass*)vec_left;
VectorClass *vright = (VectorClass*)vec_right;
// load data for tip
for (x = 0; x < VectorClass::size(); x++) {
double *tip_left, *tip_right;
if (ptn+x < orig_nptn) {
tip_left = partial_lh_left + block * (aln->at(ptn+x))[left->node->id];
tip_right = partial_lh_right + block * (aln->at(ptn+x))[right->node->id];
} else if (ptn+x < max_orig_nptn) {
tip_left = partial_lh_left + block * aln->STATE_UNKNOWN;
tip_right = partial_lh_right + block * aln->STATE_UNKNOWN;
} else if (ptn+x < nptn) {
tip_left = partial_lh_left + block * model_factory->unobserved_ptns[ptn+x-max_orig_nptn];
tip_right = partial_lh_right + block * model_factory->unobserved_ptns[ptn+x-max_orig_nptn];
} else {
tip_left = partial_lh_left + block * aln->STATE_UNKNOWN;
tip_right = partial_lh_right + block * aln->STATE_UNKNOWN;
}
double *this_vec_left = vec_left+x;
double *this_vec_right = vec_right+x;
for (i = 0; i < block; i++) {
*this_vec_left = tip_left[i];
*this_vec_right = tip_right[i];
this_vec_left += VectorClass::size();
this_vec_right += VectorClass::size();
}
}
for (i = 0; i < block; i++)
partial_lh[i] = vleft[i] * vright[i];
}
} else if (isRootLeaf(left->node) && !right->node->isLeaf()) {
// left is root node
/*--------------------- ROOT-INTERNAL NODE case ------------------*/
// only take scale_num from the right subtree
memcpy(dad_branch->scale_num + ptn_lower, right->scale_num + ptn_lower, (ptn_upper-ptn_lower) * sizeof(UBYTE));
double *partial_lh_left = partial_lh_leaves;
for (ptn = ptn_lower; ptn < ptn_upper; ptn+=VectorClass::size()) {
VectorClass *partial_lh = (VectorClass*)(dad_branch->partial_lh + ptn*block);
VectorClass *partial_lh_right = (VectorClass*)(right->partial_lh + ptn*block);
double *eright_ptr = eright;
double *lh_left = partial_lh_left;
for (c = 0; c < ncat_mix; c++) {
// compute real partial likelihood vector
for (x = 0; x < nstates; x++) {
VectorClass vright;
#ifdef KERNEL_FIX_STATES
dotProductVec<VectorClass, double, nstates, FMA>(eright_ptr, partial_lh_right, vright);
#else
dotProductVec<VectorClass, double, FMA>(eright_ptr, partial_lh_right, vright, nstates);
#endif
eright_ptr += nstates;
partial_lh[x] = lh_left[x]*vright;
}
partial_lh_right += nstates;
lh_left += nstates;
partial_lh += nstates;
}
}
} else if (left->node->isLeaf() && !right->node->isLeaf()) {
/*--------------------- TIP-INTERNAL NODE case ------------------*/
// only take scale_num from the right subtree
memcpy(dad_branch->scale_num + ptn_lower, right->scale_num + ptn_lower, (ptn_upper-ptn_lower) * sizeof(UBYTE));
double *partial_lh_left = partial_lh_leaves;
double *vec_left = buffer_partial_lh_ptr + (block*2)*VectorClass::size()*thread_id;
for (ptn = ptn_lower; ptn < ptn_upper; ptn+=VectorClass::size()) {
VectorClass *partial_lh = (VectorClass*)(dad_branch->partial_lh + ptn*block);
VectorClass *partial_lh_right = (VectorClass*)(right->partial_lh + ptn*block);
VectorClass *vleft = (VectorClass*)vec_left;
// load data for tip
for (x = 0; x < VectorClass::size(); x++) {
double *tip;
if (ptn+x < orig_nptn)
tip = partial_lh_left + block*(aln->at(ptn+x))[left->node->id];
else if (ptn+x < max_orig_nptn)
tip = partial_lh_left + block*aln->STATE_UNKNOWN;
else if (ptn+x < nptn)
tip = partial_lh_left + block*model_factory->unobserved_ptns[ptn+x-max_orig_nptn];
else
tip = partial_lh_left + block*aln->STATE_UNKNOWN;
double *this_vec_left = vec_left+x;
for (i = 0; i < block; i++) {
*this_vec_left = tip[i];
this_vec_left += VectorClass::size();
}
}
VectorClass lh_max = 0.0;
double *eright_ptr = eright;
for (c = 0; c < ncat_mix; c++) {
// compute real partial likelihood vector
for (x = 0; x < nstates; x++) {
VectorClass vright;
#ifdef KERNEL_FIX_STATES
dotProductVec<VectorClass, double, nstates, FMA>(eright_ptr, partial_lh_right, vright);
#else
dotProductVec<VectorClass, double, FMA>(eright_ptr, partial_lh_right, vright, nstates);
#endif
eright_ptr += nstates;
lh_max = max(lh_max, (partial_lh[x] = vleft[x]*vright));
}
vleft += nstates;
partial_lh_right += nstates;
partial_lh += nstates;
}
// check if one should scale partial likelihoods
auto underflown = (lh_max < SCALING_THRESHOLD);
if (horizontal_or(underflown)) {
// now do the likelihood scaling
for (x = 0; x < VectorClass::size(); x++)
if (underflown[x]) {
double *partial_lh = dad_branch->partial_lh + (ptn*block + x);
// now do the likelihood scaling
for (i = 0; i < block; i++) {
partial_lh[i*VectorClass::size()] = ldexp(partial_lh[i*VectorClass::size()], SCALING_THRESHOLD_EXP);
}
dad_branch->scale_num[ptn+x] += 1;
}
}
}
} else {
/*--------------------- INTERNAL-INTERNAL NODE case ------------------*/
for (ptn = ptn_lower; ptn < ptn_upper; ptn+=VectorClass::size()) {
VectorClass *partial_lh = (VectorClass*)(dad_branch->partial_lh + ptn*block);
VectorClass *partial_lh_left = (VectorClass*)(left->partial_lh + ptn*block);
VectorClass *partial_lh_right = (VectorClass*)(right->partial_lh + ptn*block);
VectorClass lh_max = 0.0;
for (i = 0; i < VectorClass::size(); i++)
dad_branch->scale_num[ptn+i] = left->scale_num[ptn+i] + right->scale_num[ptn+i];
double *eleft_ptr = eleft;
double *eright_ptr = eright;
for (c = 0; c < ncat_mix; c++) {
// compute real partial likelihood vector
for (x = 0; x < nstates; x++) {
#ifdef KERNEL_FIX_STATES
dotProductDualVec<VectorClass, double, nstates, FMA>(eleft_ptr, partial_lh_left, eright_ptr, partial_lh_right, partial_lh[x]);
#else
dotProductDualVec<VectorClass, double, FMA>(eleft_ptr, partial_lh_left, eright_ptr, partial_lh_right, partial_lh[x], nstates);
#endif
eleft_ptr += nstates;
eright_ptr += nstates;
lh_max=max(lh_max, partial_lh[x]);
}
partial_lh_left += nstates;
partial_lh_right += nstates;
partial_lh += nstates;
}
// check if one should scale partial likelihoods
auto underflown = (lh_max < SCALING_THRESHOLD);
if (horizontal_or(underflown)) {
// now do the likelihood scaling
for (x = 0; x < VectorClass::size(); x++)
if (underflown[x]) {
double *partial_lh = dad_branch->partial_lh + (ptn*block + x);
// now do the likelihood scaling
for (i = 0; i < block; i++) {
partial_lh[i*VectorClass::size()] = ldexp(partial_lh[i*VectorClass::size()], SCALING_THRESHOLD_EXP);
}
dad_branch->scale_num[ptn+x] += 1;
}
}
}
}
}
#ifdef KERNEL_FIX_STATES
template <class VectorClass, const int nstates, const bool FMA>
void PhyloTree::computeNonrevLikelihoodDervSIMD(PhyloNeighbor *dad_branch, PhyloNode *dad, double *df, double *ddf) {
#else
template <class VectorClass, const bool FMA>
void PhyloTree::computeNonrevLikelihoodDervGenericSIMD(PhyloNeighbor *dad_branch, PhyloNode *dad, double *df, double *ddf) {
#endif
// assert(rooted);
PhyloNode *node = (PhyloNode*) dad_branch->node;
PhyloNeighbor *node_branch = (PhyloNeighbor*) node->findNeighbor(dad);
if (!central_partial_lh)
initializeAllPartialLh();
if (node->isLeaf() || (dad_branch->direction == AWAYFROM_ROOT && !isRootLeaf(dad))) {
PhyloNode *tmp_node = dad;
dad = node;
node = tmp_node;
PhyloNeighbor *tmp_nei = dad_branch;
dad_branch = node_branch;
node_branch = tmp_nei;
}
#ifdef KERNEL_FIX_STATES
computeTraversalInfo<VectorClass, nstates>(node, dad, false);
#else
computeTraversalInfo<VectorClass>(node, dad, false);
#endif
#ifndef KERNEL_FIX_STATES
size_t nstates = aln->num_states;
#endif
size_t nstatesqr = nstates*nstates;
size_t ncat = site_rate->getNRate();
size_t ncat_mix = (model_factory->fused_mix_rate) ? ncat : ncat*model->getNMixtures();
size_t denom = (model_factory->fused_mix_rate) ? 1 : ncat;
size_t block = ncat_mix * nstates;
size_t ptn; // for big data size > 4GB memory required
size_t c, i;
size_t orig_nptn = aln->size();
size_t max_orig_nptn = ((orig_nptn+VectorClass::size()-1)/VectorClass::size())*VectorClass::size();
size_t nptn = max_orig_nptn+model_factory->unobserved_ptns.size();
bool isASC = model_factory->unobserved_ptns.size() > 0;
// double *trans_mat = new double[block*nstates*3];
double *trans_mat = buffer_partial_lh;
double *trans_derv1 = buffer_partial_lh + block*nstates;
double *trans_derv2 = trans_derv1 + block*nstates;
double *buffer_partial_lh_ptr = buffer_partial_lh + get_safe_upper_limit(3*block*nstates);
for (c = 0; c < ncat_mix; c++) {
size_t mycat = c%ncat;
size_t m = c/denom;
double cat_rate = site_rate->getRate(mycat);
double len = cat_rate * dad_branch->length;
double prop = site_rate->getProp(mycat) * model->getMixtureWeight(m);
double *this_trans_mat = &trans_mat[c*nstatesqr];
double *this_trans_derv1 = &trans_derv1[c*nstatesqr];
double *this_trans_derv2 = &trans_derv2[c*nstatesqr];
model->computeTransDerv(len, this_trans_mat, this_trans_derv1, this_trans_derv2, m);
double prop_rate = prop * cat_rate;
double prop_rate_2 = prop_rate * cat_rate;
for (i = 0; i < nstatesqr; i++) {
this_trans_mat[i] *= prop;
this_trans_derv1[i] *= prop_rate;
this_trans_derv2[i] *= prop_rate_2;
}
if (!rooted) {
// for unrooted tree, multiply with state_freq
double state_freq[nstates];
model->getStateFrequency(state_freq, m);
for (i = 0; i < nstates; i++) {
for (size_t x = 0; x < nstates; x++) {
this_trans_mat[x] *= state_freq[i];
this_trans_derv1[x] *= state_freq[i];
this_trans_derv2[x] *= state_freq[i];
}
this_trans_mat += nstates;
this_trans_derv1 += nstates;
this_trans_derv2 += nstates;
}
}
}
VectorClass all_df(0.0), all_ddf(0.0);
VectorClass all_prob_const(0.0), all_df_const(0.0), all_ddf_const(0.0);
vector<size_t> limits;
computeBounds<VectorClass>(num_threads, nptn, limits);
// double *buffer_partial_lh_ptr = buffer_partial_lh;
if (dad->isLeaf()) {
// make sure that we do not estimate the virtual branch length from the root
ASSERT(!isRootLeaf(dad));
// special treatment for TIP-INTERNAL NODE case
// double *partial_lh_node = new double[(aln->STATE_UNKNOWN+1)*block*3];
double *partial_lh_node = buffer_partial_lh_ptr;
double *partial_lh_derv1 = partial_lh_node + (aln->STATE_UNKNOWN+1)*block;
double *partial_lh_derv2 = partial_lh_derv1 + (aln->STATE_UNKNOWN+1)*block;
buffer_partial_lh_ptr += get_safe_upper_limit((aln->STATE_UNKNOWN+1)*block*3);
IntVector states_dad = aln->seq_states[dad->id];
states_dad.push_back(aln->STATE_UNKNOWN);
// precompute information from one tip
for (IntVector::iterator it = states_dad.begin(); it != states_dad.end(); it++) {
double *lh_node = partial_lh_node +(*it)*block;
double *lh_derv1 = partial_lh_derv1 +(*it)*block;
double *lh_derv2 = partial_lh_derv2 +(*it)*block;
double *lh_tip = tip_partial_lh + (*it)*nstates;
double *trans_mat_tmp = trans_mat;
double *trans_derv1_tmp = trans_derv1;
double *trans_derv2_tmp = trans_derv2;
for (c = 0; c < ncat_mix; c++) {
for (i = 0; i < nstates; i++) {
lh_node[i] = 0.0;
lh_derv1[i] = 0.0;
lh_derv2[i] = 0.0;
for (size_t x = 0; x < nstates; x++) {
lh_node[i] += trans_mat_tmp[x] * lh_tip[x];
lh_derv1[i] += trans_derv1_tmp[x] * lh_tip[x];
lh_derv2[i] += trans_derv2_tmp[x] * lh_tip[x];
}
trans_mat_tmp += nstates;
trans_derv1_tmp += nstates;
trans_derv2_tmp += nstates;
}
lh_node += nstates;
lh_derv1 += nstates;
lh_derv2 += nstates;
}
}
// now do the real computation
#ifdef _OPENMP
#pragma omp parallel for private(ptn, i, c) schedule(static,1) num_threads(num_threads)
#endif
for (int thread_id = 0; thread_id < num_threads; thread_id++) {
VectorClass my_df(0.0), my_ddf(0.0), vc_prob_const(0.0), vc_df_const(0.0), vc_ddf_const(0.0);
size_t ptn_lower = limits[thread_id];
size_t ptn_upper = limits[thread_id+1];
// first compute partial_lh
for (vector<TraversalInfo>::iterator it = traversal_info.begin(); it != traversal_info.end(); it++)
computePartialLikelihood(*it, ptn_lower, ptn_upper, thread_id);
double *vec_tip = buffer_partial_lh_ptr + block*3*VectorClass::size()*thread_id;
for (ptn = ptn_lower; ptn < ptn_upper; ptn+=VectorClass::size()) {
VectorClass lh_ptn, df_ptn, ddf_ptn;
VectorClass *partial_lh_dad = (VectorClass*)(dad_branch->partial_lh + ptn*block);
//load tip vector
for (i = 0; i < VectorClass::size(); i++) {
size_t state_dad;
if (ptn+i < orig_nptn)
state_dad = block * (aln->at(ptn+i))[dad->id];
else if (ptn+i < max_orig_nptn)
state_dad = block * aln->STATE_UNKNOWN;
else if (ptn+i < nptn)
state_dad = block * model_factory->unobserved_ptns[ptn+i-max_orig_nptn];
else
state_dad = block * aln->STATE_UNKNOWN;
double *lh_tip = partial_lh_node + state_dad;
double *lh_derv1 = partial_lh_derv1 + state_dad;
double *lh_derv2 = partial_lh_derv2 + state_dad;
double *this_vec_tip = vec_tip+i;
double *this_derv1 = this_vec_tip + block*VectorClass::size();
double *this_derv2 = this_derv1 + block*VectorClass::size();
for (c = 0; c < block; c++) {
*this_vec_tip = lh_tip[c];
*this_derv1 = lh_derv1[c];
*this_derv2 = lh_derv2[c];
this_vec_tip += VectorClass::size();
this_derv1 += VectorClass::size();
this_derv2 += VectorClass::size();
}
}
VectorClass *lh_node = (VectorClass*)vec_tip;
VectorClass *lh_derv1 = (VectorClass*)vec_tip + block;
VectorClass *lh_derv2 = (VectorClass*)lh_derv1 + block;
#ifdef KERNEL_FIX_STATES
dotProductTriple<VectorClass, VectorClass, nstates, FMA, false>(lh_node, lh_derv1, lh_derv2, partial_lh_dad, lh_ptn, df_ptn, ddf_ptn, block);
#else
dotProductTriple<VectorClass, VectorClass, FMA, false>(lh_node, lh_derv1, lh_derv2, partial_lh_dad, lh_ptn, df_ptn, ddf_ptn, block, nstates);
#endif
lh_ptn = (lh_ptn + VectorClass().load_a(&ptn_invar[ptn]));
if (ptn < orig_nptn) {
lh_ptn = 1.0 / lh_ptn;
VectorClass df_frac = df_ptn * lh_ptn;
VectorClass ddf_frac = ddf_ptn * lh_ptn;
VectorClass freq;
freq.load_a(&ptn_freq[ptn]);
VectorClass tmp1 = df_frac * freq;
VectorClass tmp2 = ddf_frac * freq;
my_df += tmp1;
my_ddf += nmul_add(tmp1, df_frac, tmp2);
} else {
if (ptn+VectorClass::size() > nptn) {
// cutoff the last entries if going beyond
lh_ptn.cutoff(nptn-ptn);
df_ptn.cutoff(nptn-ptn);
ddf_ptn.cutoff(nptn-ptn);
}
// bugfix 2016-01-21, prob_const can be rescaled
double *lh_ptn_ptr = (double*)&lh_ptn;
for (i = 0; i < VectorClass::size(); i++)
if (dad_branch->scale_num[ptn+i] >= 1)
lh_ptn_ptr[i] *= SCALING_THRESHOLD;
vc_prob_const += lh_ptn;
vc_df_const += df_ptn;
vc_ddf_const += ddf_ptn;
}
} // FOR ptn
#ifdef _OPENMP
#pragma omp critical
#endif
{
all_df += my_df;
all_ddf += my_ddf;
if (isASC) {
all_prob_const += vc_prob_const;
all_df_const += vc_df_const;
all_ddf_const += vc_ddf_const;
}
}
} // FOR thread_id
// delete [] partial_lh_node;
} else {
// both dad and node are internal nodes
#ifdef _OPENMP
#pragma omp parallel for private(ptn, i, c) schedule(static,1) num_threads(num_threads)
#endif
for (int thread_id = 0; thread_id < num_threads; thread_id++) {
VectorClass my_df(0.0), my_ddf(0.0), vc_prob_const(0.0), vc_df_const(0.0), vc_ddf_const(0.0);
size_t ptn_lower = limits[thread_id];
size_t ptn_upper = limits[thread_id+1];
// first compute partial_lh
for (vector<TraversalInfo>::iterator it = traversal_info.begin(); it != traversal_info.end(); it++)
computePartialLikelihood(*it, ptn_lower, ptn_upper, thread_id);
for (ptn = ptn_lower; ptn < ptn_upper; ptn+=VectorClass::size()) {
VectorClass lh_ptn(0.0), df_ptn(0.0), ddf_ptn(0.0);
VectorClass *partial_lh_dad = (VectorClass*)(dad_branch->partial_lh + ptn*block);
VectorClass *partial_lh_node = (VectorClass*)(node_branch->partial_lh + ptn*block);
double *trans_mat_tmp = trans_mat;
double *trans_derv1_tmp = trans_derv1;
double *trans_derv2_tmp = trans_derv2;
for (c = 0; c < ncat_mix; c++) {
for (i = 0; i < nstates; i++) {
VectorClass lh_state;
VectorClass lh_derv1;
VectorClass lh_derv2;
#ifdef KERNEL_FIX_STATES
dotProductTriple<VectorClass, double, nstates, FMA, false>(trans_mat_tmp, trans_derv1_tmp, trans_derv2_tmp, partial_lh_node, lh_state, lh_derv1, lh_derv2, nstates);
#else
dotProductTriple<VectorClass, double, FMA, false>(trans_mat_tmp, trans_derv1_tmp, trans_derv2_tmp, partial_lh_node, lh_state, lh_derv1, lh_derv2, nstates, nstates);
#endif
lh_ptn = mul_add(partial_lh_dad[i], lh_state, lh_ptn);
df_ptn = mul_add(partial_lh_dad[i], lh_derv1, df_ptn);
ddf_ptn = mul_add(partial_lh_dad[i], lh_derv2, ddf_ptn);
trans_mat_tmp += nstates;
trans_derv1_tmp += nstates;
trans_derv2_tmp += nstates;
}
partial_lh_node += nstates;
partial_lh_dad += nstates;
}
lh_ptn = (lh_ptn + VectorClass().load_a(&ptn_invar[ptn]));
if (ptn < orig_nptn) {
lh_ptn = 1.0 / lh_ptn;
VectorClass df_frac = df_ptn * lh_ptn;
VectorClass ddf_frac = ddf_ptn * lh_ptn;
VectorClass freq;
freq.load_a(&ptn_freq[ptn]);
VectorClass tmp1 = df_frac * freq;
VectorClass tmp2 = ddf_frac * freq;
my_df += tmp1;
my_ddf += nmul_add(tmp1, df_frac, tmp2);
} else {
if (ptn+VectorClass::size() > nptn) {
// cutoff the last entries if going beyond
lh_ptn.cutoff(nptn-ptn);
df_ptn.cutoff(nptn-ptn);
ddf_ptn.cutoff(nptn-ptn);
}
// bugfix 2016-01-21, prob_const can be rescaled
double *lh_ptn_ptr = (double*)&lh_ptn;
for (i = 0; i < VectorClass::size(); i++)
if (dad_branch->scale_num[ptn+i] >= 1)
lh_ptn_ptr[i] *= SCALING_THRESHOLD;
vc_prob_const += lh_ptn;
vc_df_const += df_ptn;
vc_ddf_const += ddf_ptn;
}
} // FOR ptn
#ifdef _OPENMP
#pragma omp critical
#endif
{
all_df += my_df;
all_ddf += my_ddf;
if (isASC) {
all_prob_const += vc_prob_const;
all_df_const += vc_df_const;
all_ddf_const += vc_ddf_const;
}
}
} // FOR thread
}
*df = horizontal_add(all_df);
*ddf = horizontal_add(all_ddf);
ASSERT(std::isfinite(*df) && "Numerical underflow for non-rev lh-derivative");
if (isASC) {
double prob_const = 0.0, df_const = 0.0, ddf_const = 0.0;
prob_const = horizontal_add(all_prob_const);
df_const = horizontal_add(all_df_const);
ddf_const = horizontal_add(all_ddf_const);
// ascertainment bias correction
prob_const = 1.0 - prob_const;
double df_frac = df_const / prob_const;
double ddf_frac = ddf_const / prob_const;
int nsites = aln->getNSite();
*df += nsites * df_frac;
*ddf += nsites *(ddf_frac + df_frac*df_frac);
}
}
#ifdef KERNEL_FIX_STATES
template <class VectorClass, const int nstates, const bool FMA>
double PhyloTree::computeNonrevLikelihoodBranchSIMD(PhyloNeighbor *dad_branch, PhyloNode *dad) {
#else
template <class VectorClass, const bool FMA>
double PhyloTree::computeNonrevLikelihoodBranchGenericSIMD(PhyloNeighbor *dad_branch, PhyloNode *dad) {
#endif
// assert(rooted);
PhyloNode *node = (PhyloNode*) dad_branch->node;
PhyloNeighbor *node_branch = (PhyloNeighbor*) node->findNeighbor(dad);
if (!central_partial_lh)
initializeAllPartialLh();
if (node->isLeaf() || (dad_branch->direction == AWAYFROM_ROOT && !isRootLeaf(dad))) {
PhyloNode *tmp_node = dad;
dad = node;
node = tmp_node;
PhyloNeighbor *tmp_nei = dad_branch;
dad_branch = node_branch;
node_branch = tmp_nei;
}
#ifdef KERNEL_FIX_STATES
computeTraversalInfo<VectorClass, nstates>(node, dad, false);
#else
computeTraversalInfo<VectorClass>(node, dad, false);
#endif
double tree_lh = 0.0;
#ifndef KERNEL_FIX_STATES
size_t nstates = aln->num_states;
#endif
size_t nstatesqr = nstates*nstates;
size_t ncat = site_rate->getNRate();
size_t ncat_mix = (model_factory->fused_mix_rate) ? ncat : ncat*model->getNMixtures();
size_t denom = (model_factory->fused_mix_rate) ? 1 : ncat;
size_t block = ncat_mix * nstates;
size_t ptn; // for big data size > 4GB memory required
size_t c, i;
size_t orig_nptn = aln->size();
size_t max_orig_nptn = ((orig_nptn+VectorClass::size()-1)/VectorClass::size())*VectorClass::size();
size_t nptn = max_orig_nptn+model_factory->unobserved_ptns.size();
bool isASC = model_factory->unobserved_ptns.size() > 0;
vector<size_t> limits;
computeBounds<VectorClass>(num_threads, nptn, limits);
// double *trans_mat = new double[block*nstates];
double *trans_mat = buffer_partial_lh;
double *buffer_partial_lh_ptr = buffer_partial_lh + block*nstates;
for (c = 0; c < ncat_mix; c++) {
size_t mycat = c%ncat;
size_t m = c/denom;
double len = site_rate->getRate(mycat) * dad_branch->length;
double prop = site_rate->getProp(mycat) * model->getMixtureWeight(m);
double *this_trans_mat = &trans_mat[c*nstatesqr];
model->computeTransMatrix(len, this_trans_mat, m);
for (i = 0; i < nstatesqr; i++)
this_trans_mat[i] *= prop;
if (!rooted) {
// if unrooted tree, multiply with frequency
double state_freq[nstates];
model->getStateFrequency(state_freq, m);
for (i = 0; i < nstates; i++) {
for (size_t x = 0; x < nstates; x++)
this_trans_mat[x] *= state_freq[i];
this_trans_mat += nstates;
}
}
}
VectorClass all_tree_lh(0.0);
VectorClass all_prob_const(0.0);
if (dad->isLeaf()) {
// special treatment for TIP-INTERNAL NODE case
// double *partial_lh_node = new double[(aln->STATE_UNKNOWN+1)*block];
double *partial_lh_node = buffer_partial_lh_ptr;
buffer_partial_lh_ptr += get_safe_upper_limit((aln->STATE_UNKNOWN+1)*block);
if (isRootLeaf(dad)) {
for (c = 0; c < ncat_mix; c++) {
double *lh_node = partial_lh_node + c*nstates;
size_t m = c/denom;
model->getStateFrequency(lh_node, m);
double prop = site_rate->getProp(c%ncat) * model->getMixtureWeight(m);
for (i = 0; i < nstates; i++)
lh_node[i] *= prop;
}
} else {
IntVector states_dad = aln->seq_states[dad->id];
states_dad.push_back(aln->STATE_UNKNOWN);
// precompute information from one tip
for (IntVector::iterator it = states_dad.begin(); it != states_dad.end(); it++) {
double *lh_node = partial_lh_node +(*it)*block;
double *lh_tip = tip_partial_lh + (*it)*nstates;
double *trans_mat_tmp = trans_mat;
for (c = 0; c < ncat_mix; c++) {
for (i = 0; i < nstates; i++) {
lh_node[i] = 0.0;
for (size_t x = 0; x < nstates; x++)
lh_node[i] += trans_mat_tmp[x] * lh_tip[x];
trans_mat_tmp += nstates;
}
lh_node += nstates;
}
}
}
// now do the real computation
#ifdef _OPENMP
#pragma omp parallel for private(ptn, i, c) schedule(static,1) num_threads(num_threads)
#endif
for (int thread_id = 0; thread_id < num_threads; thread_id++) {
VectorClass vc_tree_lh(0.0), vc_prob_const(0.0);
size_t ptn_lower = limits[thread_id];
size_t ptn_upper = limits[thread_id+1];
// first compute partial_lh
for (vector<TraversalInfo>::iterator it = traversal_info.begin(); it != traversal_info.end(); it++)
computePartialLikelihood(*it, ptn_lower, ptn_upper, thread_id);
// reset memory for _pattern_lh_cat
// memset(_pattern_lh_cat+ptn_lower*ncat_mix, 0, (ptn_upper-ptn_lower)*ncat_mix*sizeof(double));
double *vec_tip = buffer_partial_lh_ptr + block*VectorClass::size()*thread_id;
for (ptn = ptn_lower; ptn < ptn_upper; ptn+=VectorClass::size()) {
VectorClass lh_ptn;
lh_ptn.load_a(&ptn_invar[ptn]);
VectorClass *lh_cat = (VectorClass*)(_pattern_lh_cat + ptn*ncat_mix);
VectorClass *partial_lh_dad = (VectorClass*)(dad_branch->partial_lh + ptn*block);
VectorClass *lh_node = (VectorClass*)vec_tip;
//load tip vector
for (i = 0; i < VectorClass::size(); i++) {
double *lh_tip;
if (isRootLeaf(dad))
lh_tip = partial_lh_node;
else if (ptn+i < orig_nptn)
lh_tip = partial_lh_node + block*(aln->at(ptn+i))[dad->id];
else if (ptn+i < max_orig_nptn)
lh_tip = partial_lh_node + block*aln->STATE_UNKNOWN;
else if (ptn+i < nptn)
lh_tip = partial_lh_node + block*model_factory->unobserved_ptns[ptn+i-max_orig_nptn];
else
lh_tip = partial_lh_node + block*aln->STATE_UNKNOWN;
double *this_vec_tip = vec_tip+i;
for (c = 0; c < block; c++) {
*this_vec_tip = lh_tip[c];
this_vec_tip += VectorClass::size();
}
}
if (_pattern_lh_cat_state) {
// naively compute pattern_lh per category per state
VectorClass *lh_state = (VectorClass*)(_pattern_lh_cat_state + ptn*block);
for (c = 0; c < ncat_mix; c++) {
for (i=0; i < nstates; i++) {
lh_cat[c] += (lh_state[i] = lh_node[i]*partial_lh_dad[i]);
}
lh_node += nstates;
partial_lh_dad += nstates;
lh_state += nstates;
lh_ptn += lh_cat[c];
}
} else {
for (c = 0; c < ncat_mix; c++) {
#ifdef KERNEL_FIX_STATES
dotProductVec<VectorClass, VectorClass, nstates, FMA>(lh_node, partial_lh_dad, lh_cat[c]);
#else
dotProductVec<VectorClass, VectorClass, FMA>(lh_node, partial_lh_dad, lh_cat[c], nstates);
#endif
lh_node += nstates;
partial_lh_dad += nstates;
lh_ptn += lh_cat[c];
}
}
VectorClass vc_min_scale;
double* vc_min_scale_ptr = (double*)&vc_min_scale;
for (i = 0; i < VectorClass::size(); i++) {
vc_min_scale_ptr[i] = dad_branch->scale_num[ptn+i];
}
vc_min_scale *= LOG_SCALING_THRESHOLD;
// lh_ptn = abs(lh_ptn);
// assert(horizontal_and(lh_ptn > 0));
if (ptn < orig_nptn) {
lh_ptn = log(lh_ptn) + vc_min_scale;
lh_ptn.store_a(&_pattern_lh[ptn]);
vc_tree_lh = mul_add(lh_ptn, VectorClass().load_a(&ptn_freq[ptn]), vc_tree_lh);
} else {
// ascertainment bias correction
if (ptn+VectorClass::size() > nptn) {
// cutoff the last entries if going beyond
lh_ptn.cutoff(nptn-ptn);
}
// bugfix 2016-01-21, prob_const can be rescaled
if (horizontal_or(vc_min_scale != 0.0)) {
// some entries are rescaled
double *lh_ptn_dbl = (double*)&lh_ptn;
for (i = 0; i < VectorClass::size(); i++)
if (vc_min_scale_ptr[i] != 0.0)
lh_ptn_dbl[i] *= SCALING_THRESHOLD;
}
vc_prob_const += lh_ptn;
}
} // FOR ptn
#ifdef _OPENMP
#pragma omp critical
#endif
{
all_tree_lh += vc_tree_lh;
if (isASC)
all_prob_const += vc_prob_const;
}
} // FOR thread_id
} else {
// both dad and node are internal nodes
#ifdef _OPENMP
#pragma omp parallel for private(ptn, i, c) schedule(static,1) num_threads(num_threads)
#endif
for (int thread_id = 0; thread_id < num_threads; thread_id++) {
VectorClass vc_tree_lh(0.0), vc_prob_const(0.0);
size_t ptn_lower = limits[thread_id];
size_t ptn_upper = limits[thread_id+1];
// first compute partial_lh
for (vector<TraversalInfo>::iterator it = traversal_info.begin(); it != traversal_info.end(); it++)
computePartialLikelihood(*it, ptn_lower, ptn_upper, thread_id);
// reset memory for _pattern_lh_cat
memset(_pattern_lh_cat+ptn_lower*ncat_mix, 0, (ptn_upper-ptn_lower)*ncat_mix*sizeof(double));
for (ptn = ptn_lower; ptn < ptn_upper; ptn+=VectorClass::size()) {
VectorClass lh_ptn;
lh_ptn.load_a(&ptn_invar[ptn]);
VectorClass *lh_cat = (VectorClass*)(_pattern_lh_cat + ptn*ncat_mix);
VectorClass *partial_lh_dad = (VectorClass*)(dad_branch->partial_lh + ptn*block);
VectorClass *partial_lh_node = (VectorClass*)(node_branch->partial_lh + ptn*block);
double *trans_mat_tmp = trans_mat;
if (_pattern_lh_cat_state) {
VectorClass *lh_state = (VectorClass*)(_pattern_lh_cat_state + ptn*block);
for (c = 0; c < ncat_mix; c++) {
for (i = 0; i < nstates; i++) {
#ifdef KERNEL_FIX_STATES
dotProductVec<VectorClass, double, nstates, FMA>(trans_mat_tmp, partial_lh_node, lh_state[i]);
#else
dotProductVec<VectorClass, double, FMA>(trans_mat_tmp, partial_lh_node, lh_state[i], nstates);
#endif
lh_cat[c] += (lh_state[i] *= partial_lh_dad[i]);
trans_mat_tmp += nstates;
}
lh_ptn += lh_cat[c];
partial_lh_node += nstates;
partial_lh_dad += nstates;
lh_state += nstates;
}
} else {
for (c = 0; c < ncat_mix; c++) {
for (i = 0; i < nstates; i++) {
VectorClass lh_state;
#ifdef KERNEL_FIX_STATES
dotProductVec<VectorClass, double, nstates, FMA>(trans_mat_tmp, partial_lh_node, lh_state);
#else
dotProductVec<VectorClass, double, FMA>(trans_mat_tmp, partial_lh_node, lh_state, nstates);
#endif
lh_cat[c] = mul_add(partial_lh_dad[i], lh_state, lh_cat[c]);
trans_mat_tmp += nstates;
}
lh_ptn += lh_cat[c];
partial_lh_node += nstates;
partial_lh_dad += nstates;
}
}
VectorClass vc_min_scale;
double* vc_min_scale_ptr = (double*)&vc_min_scale;
for (i = 0; i < VectorClass::size(); i++) {
vc_min_scale_ptr[i] = dad_branch->scale_num[ptn+i] + node_branch->scale_num[ptn+i];
}
vc_min_scale *= LOG_SCALING_THRESHOLD;
// lh_ptn = abs(lh_ptn);
ASSERT(horizontal_and(lh_ptn > 0));
if (ptn < orig_nptn) {
lh_ptn = log(lh_ptn) + vc_min_scale;
lh_ptn.store_a(&_pattern_lh[ptn]);
vc_tree_lh = mul_add(lh_ptn, VectorClass().load_a(&ptn_freq[ptn]), vc_tree_lh);
} else {
// ascertainment bias correction
if (ptn+VectorClass::size() > nptn) {
// cutoff the last entries if going beyond
lh_ptn.cutoff(nptn-ptn);
}
// bugfix 2016-01-21, prob_const can be rescaled
if (horizontal_or(vc_min_scale != 0.0)) {
// some entries are rescaled
double *lh_ptn_dbl = (double*)&lh_ptn;
for (i = 0; i < VectorClass::size(); i++)
if (vc_min_scale_ptr[i] != 0.0)
lh_ptn_dbl[i] *= SCALING_THRESHOLD;
}
vc_prob_const += lh_ptn;
}
} // FOR ptn
#ifdef _OPENMP
#pragma omp critical
#endif
{
all_tree_lh += vc_tree_lh;
if (isASC)
all_prob_const += vc_prob_const;
}
} // FOR thread_id
}
tree_lh = horizontal_add(all_tree_lh);
if (!std::isfinite(tree_lh)) {
model->writeInfo(cout);
site_rate->writeInfo(cout);
ASSERT(0 && "Numerical underflow for non-rev lh-branch");
}
if (isASC) {
// ascertainment bias correction
double prob_const = horizontal_add(all_prob_const);
if (prob_const >= 1.0 || prob_const < 0.0) {
printTree(cout, WT_TAXON_ID + WT_BR_LEN + WT_NEWLINE);
model->writeInfo(cout);
}
ASSERT(prob_const < 1.0 && prob_const >= 0.0);
prob_const = log(1.0 - prob_const);
for (ptn = 0; ptn < orig_nptn; ptn+=VectorClass::size())
(VectorClass().load_a(&_pattern_lh[ptn])-prob_const).store_a(&_pattern_lh[ptn]);
tree_lh -= aln->getNSite()*prob_const;
ASSERT(std::isfinite(tree_lh));
}
return tree_lh;
}
#endif
|
