-
Notifications
You must be signed in to change notification settings - Fork 18
/
Copy pathHeavyFlavBaseProducer.py
810 lines (711 loc) · 39.2 KB
/
HeavyFlavBaseProducer.py
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
import os
import itertools
import numpy as np
import ROOT
ROOT.PyConfig.IgnoreCommandLineOptions = True
from PhysicsTools.NanoAODTools.postprocessing.framework.datamodel import Collection, Object
from PhysicsTools.NanoAODTools.postprocessing.framework.eventloop import Module
from ..helpers.utils import deltaR, closest, polarP4, sumP4, get_subjets, corrected_svmass, configLogger
from ..helpers.xgbHelper import XGBEnsemble
from ..helpers.nnHelper import convert_prob, ensemble
from ..helpers.jetmetCorrector import JetMETCorrector, rndSeed
import logging
logger = logging.getLogger('nano')
configLogger('nano', loglevel=logging.INFO)
lumi_dict = {2015: 19.52, 2016: 16.81, 2017: 41.48, 2018: 59.83}
class _NullObject:
'''An null object which does not store anything, and does not raise exception.'''
def __bool__(self):
return False
def __nonzero__(self):
return False
def __getattr__(self, name):
pass
def __setattr__(self, name, value):
pass
class METObject(Object):
def p4(self):
return polarP4(self, eta=None, mass=None)
class HeavyFlavBaseProducer(Module, object):
def __init__(self, channel, **kwargs):
self._channel = channel # 'qcd', 'photon', 'inclusive', 'muon'
self.year = int(kwargs['year'])
self.jetType = kwargs.get('jetType', 'ak8').lower()
self._jmeSysts = {'jec': False, 'jes': None, 'jes_source': '', 'jes_uncertainty_file_prefix': '',
'jer': None, 'jmr': None, 'met_unclustered': None, 'smearMET': True, 'applyHEMUnc': False}
self._opts = {'sfbdt_threshold': -99,
'run_tagger': False, 'tagger_versions': ['V02b', 'V02c', 'V02d'],
'run_mass_regression': False, 'mass_regression_versions': ['V01a', 'V01b', 'V01c'],
'WRITE_CACHE_FILE': False}
for k in kwargs:
if k in self._jmeSysts:
self._jmeSysts[k] = kwargs[k]
else:
self._opts[k] = kwargs[k]
self._needsJMECorr = any([self._jmeSysts['jec'], self._jmeSysts['jes'],
self._jmeSysts['jer'], self._jmeSysts['jmr'],
self._jmeSysts['met_unclustered'], self._jmeSysts['applyHEMUnc']])
logger.info('Running %s channel for %s jets with JME systematics %s, other options %s',
self._channel, self.jetType, str(self._jmeSysts), str(self._opts))
if self.jetType == 'ak8':
self._jetConeSize = 0.8
self._fj_name = 'FatJet'
self._sj_name = 'SubJet'
self._fj_gen_name = 'GenJetAK8'
self._sj_gen_name = 'SubGenJetAK8'
self._sfbdt_files = [
os.path.expandvars(
'$CMSSW_BASE/src/PhysicsTools/NanoHRTTools/data/sfBDT/ak15/xgb_train_qcd.model.%d' % idx)
for idx in range(10)] # FIXME: update to AK8 training
self._sfbdt_vars = ['fj_2_tau21', 'fj_2_sj1_rawmass', 'fj_2_sj2_rawmass',
'fj_2_ntracks_sv12', 'fj_2_sj1_sv1_pt', 'fj_2_sj2_sv1_pt']
elif self.jetType == 'ak15':
self._jetConeSize = 1.5
self._fj_name = 'AK15Puppi'
self._sj_name = 'AK15PuppiSubJet'
self._fj_gen_name = 'GenJetAK15'
self._sj_gen_name = 'GenSubJetAK15'
self._sfbdt_files = [
os.path.expandvars(
'$CMSSW_BASE/src/PhysicsTools/NanoHRTTools/data/sfBDT/ak15/xgb_train_qcd.model.%d' % idx)
for idx in range(10)]
self._sfbdt_vars = ['fj_2_tau21', 'fj_2_sj1_rawmass', 'fj_2_sj2_rawmass',
'fj_2_ntracks_sv12', 'fj_2_sj1_sv1_pt', 'fj_2_sj2_sv1_pt']
else:
raise RuntimeError('Jet type %s is not recognized!' % self.jetType)
self._fill_sv = self._channel in ('qcd', 'photon', 'inclusive') and self._opts['sfbdt_threshold'] > -99
if self._needsJMECorr:
self.jetmetCorr = JetMETCorrector(year=self.year, jetType="AK4PFchs", **self._jmeSysts)
self.fatjetCorr = JetMETCorrector(year=self.year, jetType="AK8PFPuppi", **self._jmeSysts)
self.subjetCorr = JetMETCorrector(year=self.year, jetType="AK4PFPuppi", **self._jmeSysts)
if self._opts['run_tagger'] or self._opts['run_mass_regression']:
from ..helpers.makeInputs import ParticleNetTagInfoMaker
from ..helpers.runPrediction import ParticleNetJetTagsProducer
self.tagInfoMaker = ParticleNetTagInfoMaker(
fatjet_branch=self._fj_name, pfcand_branch='PFCands', sv_branch='SV', jetR=self._jetConeSize)
prefix = os.path.expandvars('$CMSSW_BASE/src/PhysicsTools/NanoHRTTools/data')
if self._opts['run_tagger']:
self.pnTaggers = [ParticleNetJetTagsProducer(
'%s/ParticleNet-MD/%s/{version}/particle-net.onnx' % (prefix, self.jetType),
'%s/ParticleNet-MD/%s/{version}/preprocess.json' % (prefix, self.jetType),
version=ver, cache_suffix='tagger') for ver in self._opts['tagger_versions']]
if self._opts['run_mass_regression']:
self.pnMassRegressions = [ParticleNetJetTagsProducer(
'%s/MassRegression/%s/{version}/particle_net_regression.onnx' % (prefix, self.jetType),
'%s/MassRegression/%s/{version}/preprocess.json' % (prefix, self.jetType),
version=ver, cache_suffix='mass') for ver in self._opts['mass_regression_versions']]
# https://twiki.cern.ch/twiki/bin/viewauth/CMS/BtagRecommendation
self.DeepJet_WP_L = {2015: 0.0508, 2016: 0.0480, 2017: 0.0532, 2018: 0.0490}[self.year]
self.DeepJet_WP_M = {2015: 0.2598, 2016: 0.2489, 2017: 0.3040, 2018: 0.2783}[self.year]
self.DeepJet_WP_T = {2015: 0.6502, 2016: 0.6377, 2017: 0.7476, 2018: 0.7100}[self.year]
def beginJob(self):
if self._needsJMECorr:
self.jetmetCorr.beginJob()
self.fatjetCorr.beginJob()
self.subjetCorr.beginJob()
if self._opts['sfbdt_threshold'] > -99:
self.xgb = XGBEnsemble(self._sfbdt_files, self._sfbdt_vars)
def beginFile(self, inputFile, outputFile, inputTree, wrappedOutputTree):
self.isMC = bool(inputTree.GetBranch('genWeight'))
self.hasParticleNetProb = bool(inputTree.GetBranch(self._fj_name + '_ParticleNetMD_probXbb'))
# remove all possible h5 cache files
for f in os.listdir('.'):
if f.endswith('.h5'):
os.remove(f)
if self._opts['run_tagger']:
for p in self.pnTaggers:
p.load_cache(inputFile)
if self._opts['run_mass_regression']:
for p in self.pnMassRegressions:
p.load_cache(inputFile)
if self._opts['run_tagger'] or self._opts['run_mass_regression']:
self.tagInfoMaker.init_file(inputFile, fetch_step=1000)
self.out = wrappedOutputTree
# NOTE: branch names must start with a lower case letter
# check keep_and_drop_output.txt
self.out.branch("year", "I")
self.out.branch("lumiwgt", "F")
self.out.branch("jetR", "F")
self.out.branch("passmetfilters", "O")
self.out.branch("l1PreFiringWeight", "F")
self.out.branch("l1PreFiringWeightUp", "F")
self.out.branch("l1PreFiringWeightDown", "F")
self.out.branch("nlep", "I")
self.out.branch("ht", "F")
self.out.branch("met", "F")
self.out.branch("metphi", "F")
# Large-R jets
for idx in ([1, 2] if self._channel == 'qcd' else [1]):
prefix = 'fj_%d_' % idx
# fatjet kinematics
self.out.branch(prefix + "is_qualified", "O")
self.out.branch(prefix + "pt", "F")
self.out.branch(prefix + "eta", "F")
self.out.branch(prefix + "phi", "F")
self.out.branch(prefix + "rawmass", "F")
self.out.branch(prefix + "sdmass", "F")
self.out.branch(prefix + "regressed_mass", "F")
self.out.branch(prefix + "tau21", "F")
self.out.branch(prefix + "tau32", "F")
self.out.branch(prefix + "btagcsvv2", "F")
self.out.branch(prefix + "btagjp", "F")
# subjets
self.out.branch(prefix + "deltaR_sj12", "F")
self.out.branch(prefix + "sj1_pt", "F")
self.out.branch(prefix + "sj1_eta", "F")
self.out.branch(prefix + "sj1_phi", "F")
self.out.branch(prefix + "sj1_rawmass", "F")
self.out.branch(prefix + "sj1_btagdeepcsv", "F")
self.out.branch(prefix + "sj2_pt", "F")
self.out.branch(prefix + "sj2_eta", "F")
self.out.branch(prefix + "sj2_phi", "F")
self.out.branch(prefix + "sj2_rawmass", "F")
self.out.branch(prefix + "sj2_btagdeepcsv", "F")
# taggers
self.out.branch(prefix + "DeepAK8_TvsQCD", "F")
self.out.branch(prefix + "DeepAK8_WvsQCD", "F")
self.out.branch(prefix + "DeepAK8_ZvsQCD", "F")
self.out.branch(prefix + "DeepAK8_ZHbbvsQCD", "F")
self.out.branch(prefix + "DeepAK8MD_TvsQCD", "F")
self.out.branch(prefix + "DeepAK8MD_WvsQCD", "F")
self.out.branch(prefix + "DeepAK8MD_ZvsQCD", "F")
self.out.branch(prefix + "DeepAK8MD_ZHbbvsQCD", "F")
self.out.branch(prefix + "DeepAK8MD_ZHccvsQCD", "F")
self.out.branch(prefix + "DeepAK8MD_bbVsLight", "F")
self.out.branch(prefix + "DeepAK8MD_bbVsTop", "F")
self.out.branch(prefix + "ParticleNet_TvsQCD", "F")
self.out.branch(prefix + "ParticleNet_WvsQCD", "F")
self.out.branch(prefix + "ParticleNet_ZvsQCD", "F")
self.out.branch(prefix + "ParticleNetMD_Xbb", "F")
self.out.branch(prefix + "ParticleNetMD_Xcc", "F")
self.out.branch(prefix + "ParticleNetMD_Xqq", "F")
self.out.branch(prefix + "ParticleNetMD_QCD", "F")
self.out.branch(prefix + "ParticleNetMD_XbbVsQCD", "F")
self.out.branch(prefix + "ParticleNetMD_XccVsQCD", "F")
self.out.branch(prefix + "ParticleNetMD_XccOrXqqVsQCD", "F")
if self._opts['run_tagger']:
self.out.branch(prefix + "origParticleNetMD_XccVsQCD", "F")
self.out.branch(prefix + "origParticleNetMD_XbbVsQCD", "F")
# matching variables
if self.isMC:
self.out.branch(prefix + "nbhadrons", "I")
self.out.branch(prefix + "nchadrons", "I")
self.out.branch(prefix + "partonflavour", "I")
self.out.branch(prefix + "sj1_nbhadrons", "I")
self.out.branch(prefix + "sj1_nchadrons", "I")
self.out.branch(prefix + "sj1_partonflavour", "I")
self.out.branch(prefix + "sj2_nbhadrons", "I")
self.out.branch(prefix + "sj2_nchadrons", "I")
self.out.branch(prefix + "sj2_partonflavour", "I")
# info of the closest hadGenH
self.out.branch(prefix + "dr_H", "F")
self.out.branch(prefix + "dr_H_daus", "F")
self.out.branch(prefix + "H_pt", "F")
self.out.branch(prefix + "H_decay", "I")
# info of the closest hadGenZ
self.out.branch(prefix + "dr_Z", "F")
self.out.branch(prefix + "dr_Z_daus", "F")
self.out.branch(prefix + "Z_pt", "F")
self.out.branch(prefix + "Z_decay", "I")
# info of the closest hadGenW
self.out.branch(prefix + "dr_W", "F")
self.out.branch(prefix + "dr_W_daus", "F")
self.out.branch(prefix + "W_pt", "F")
self.out.branch(prefix + "W_decay", "I")
# info of the closest hadGenTop
self.out.branch(prefix + "dr_T", "F")
self.out.branch(prefix + "dr_T_b", "F")
self.out.branch(prefix + "dr_T_Wq_max", "F")
self.out.branch(prefix + "dr_T_Wq_min", "F")
self.out.branch(prefix + "T_Wq_max_pdgId", "I")
self.out.branch(prefix + "T_Wq_min_pdgId", "I")
self.out.branch(prefix + "T_pt", "F")
if self._fill_sv:
# SV variables
self.out.branch(prefix + "nsv", "I")
self.out.branch(prefix + "nsv_ptgt25", "I")
self.out.branch(prefix + "nsv_ptgt50", "I")
self.out.branch(prefix + "ntracks", "I")
self.out.branch(prefix + "ntracks_sv12", "I")
self.out.branch(prefix + "sj1_ntracks", "I")
self.out.branch(prefix + "sj1_nsv", "I")
self.out.branch(prefix + "sj1_sv1_pt", "F")
self.out.branch(prefix + "sj1_sv1_mass", "F")
self.out.branch(prefix + "sj1_sv1_masscor", "F")
self.out.branch(prefix + "sj1_sv1_ntracks", "I")
self.out.branch(prefix + "sj1_sv1_dxy", "F")
self.out.branch(prefix + "sj1_sv1_dxysig", "F")
self.out.branch(prefix + "sj1_sv1_dlen", "F")
self.out.branch(prefix + "sj1_sv1_dlensig", "F")
self.out.branch(prefix + "sj1_sv1_chi2ndof", "F")
self.out.branch(prefix + "sj1_sv1_pangle", "F")
self.out.branch(prefix + "sj2_ntracks", "I")
self.out.branch(prefix + "sj2_nsv", "I")
self.out.branch(prefix + "sj2_sv1_pt", "F")
self.out.branch(prefix + "sj2_sv1_mass", "F")
self.out.branch(prefix + "sj2_sv1_masscor", "F")
self.out.branch(prefix + "sj2_sv1_ntracks", "I")
self.out.branch(prefix + "sj2_sv1_dxy", "F")
self.out.branch(prefix + "sj2_sv1_dxysig", "F")
self.out.branch(prefix + "sj2_sv1_dlen", "F")
self.out.branch(prefix + "sj2_sv1_dlensig", "F")
self.out.branch(prefix + "sj2_sv1_chi2ndof", "F")
self.out.branch(prefix + "sj2_sv1_pangle", "F")
self.out.branch(prefix + "sj12_masscor_dxysig", "F")
# sfBDT
self.out.branch(prefix + "sfBDT", "F")
def endFile(self, inputFile, outputFile, inputTree, wrappedOutputTree):
if self._opts['run_tagger'] and self._opts['WRITE_CACHE_FILE']:
for p in self.pnTaggers:
p.update_cache()
if self._opts['run_mass_regression'] and self._opts['WRITE_CACHE_FILE']:
for p in self.pnMassRegressions:
p.update_cache()
# remove all h5 cache files
if self._opts['run_tagger'] or self._opts['run_mass_regression']:
for f in os.listdir('.'):
if f.endswith('.h5'):
os.remove(f)
def selectLeptons(self, event):
# do lepton selection
event.looseLeptons = [] # used for jet lepton cleaning & lepton counting
electrons = Collection(event, "Electron")
for el in electrons:
el.etaSC = el.eta + el.deltaEtaSC
if el.pt > 10 and abs(el.eta) < 2.5 and abs(el.dxy) < 0.05 and abs(el.dz) < 0.2 \
and el.mvaFall17V2noIso_WP90 and el.miniPFRelIso_all < 0.4:
event.looseLeptons.append(el)
muons = Collection(event, "Muon")
for mu in muons:
if mu.pt > 10 and abs(mu.eta) < 2.4 and abs(mu.dxy) < 0.05 and abs(mu.dz) < 0.2 \
and mu.looseId and mu.miniPFRelIso_all < 0.4:
event.looseLeptons.append(mu)
event.looseLeptons.sort(key=lambda x: x.pt, reverse=True)
def correctJetsAndMET(self, event):
# correct Jets and MET
event.idx = event._entry if event._tree._entrylist is None else event._tree._entrylist.GetEntry(event._entry)
event._allJets = Collection(event, "Jet")
event.met = METObject(event, "MET")
event._allFatJets = Collection(event, self._fj_name)
event.subjets = Collection(event, self._sj_name) # do not sort subjets after updating!!
if self._needsJMECorr:
rho = event.fixedGridRhoFastjetAll
# correct AK4 jets and MET
self.jetmetCorr.setSeed(rndSeed(event, event._allJets))
self.jetmetCorr.correctJetAndMET(jets=event._allJets, lowPtJets=Collection(event, "CorrT1METJet"),
met=event.met, rawMET=METObject(event, "RawMET"),
defaultMET=METObject(event, "MET"),
rho=rho, genjets=Collection(event, 'GenJet') if self.isMC else None,
isMC=self.isMC, runNumber=event.run)
event._allJets = sorted(event._allJets, key=lambda x: x.pt, reverse=True) # sort by pt after updating
# correct fatjets
self.fatjetCorr.setSeed(rndSeed(event, event._allFatJets))
self.fatjetCorr.correctJetAndMET(jets=event._allFatJets, met=None, rho=rho,
genjets=Collection(event, self._fj_gen_name) if self.isMC else None,
isMC=self.isMC, runNumber=event.run)
# correct subjets
self.subjetCorr.setSeed(rndSeed(event, event.subjets))
self.subjetCorr.correctJetAndMET(jets=event.subjets, met=None, rho=rho,
genjets=Collection(event, self._sj_gen_name) if self.isMC else None,
isMC=self.isMC, runNumber=event.run)
# jet mass resolution smearing
if self.isMC and self._jmeSysts['jmr']:
raise NotImplementedError
# link fatjet to subjets and recompute softdrop mass
for idx, fj in enumerate(event._allFatJets):
fj.idx = idx
fj.is_qualified = True
fj.subjets = get_subjets(fj, event.subjets, ('subJetIdx1', 'subJetIdx2'))
fj.msoftdrop = sumP4(*fj.subjets).M()
event._allFatJets = sorted(event._allFatJets, key=lambda x: x.pt, reverse=True) # sort by pt
# select lepton-cleaned jets
event.fatjets = [fj for fj in event._allFatJets if fj.pt > 200 and abs(fj.eta) < 2.4 and (
fj.jetId & 2) and closest(fj, event.looseLeptons)[1] >= self._jetConeSize]
event.ak4jets = [j for j in event._allJets if j.pt > 25 and abs(j.eta) < 2.4 and (
j.jetId & 4) and closest(j, event.looseLeptons)[1] >= 0.4]
event.ht = sum([j.pt for j in event.ak4jets])
def selectSV(self, event):
event._allSV = Collection(event, "SV")
event.secondary_vertices = []
for sv in event._allSV:
# if sv.ntracks > 2 and abs(sv.dxy) < 3. and sv.dlenSig > 4:
# if sv.dlenSig > 4:
if True:
event.secondary_vertices.append(sv)
event.secondary_vertices = sorted(event.secondary_vertices, key=lambda x: x.pt, reverse=True) # sort by pt
# event.secondary_vertices = sorted(event.secondary_vertices, key=lambda x : x.dxySig, reverse=True) # sort by dxysig
def matchSVToFatJets(self, event, fatjets):
# match SV to fatjets
for fj in fatjets:
fj.sv_list = []
for sv in event.secondary_vertices:
if deltaR(sv, fj) < self._jetConeSize:
fj.sv_list.append(sv)
# match SV to subjets
drcut = min(0.4, 0.5 * deltaR(*fj.subjets)) if len(fj.subjets) == 2 else 0.4
for sj in fj.subjets:
sj.sv_list = []
for sv in event.secondary_vertices:
if deltaR(sv, sj) < drcut:
sj.sv_list.append(sv)
fj.nsv_ptgt25 = 0
fj.nsv_ptgt50 = 0
fj.ntracks = 0
fj.ntracks_sv12 = 0
for isv, sv in enumerate(fj.sv_list):
fj.ntracks += sv.ntracks
if isv < 2:
fj.ntracks_sv12 += sv.ntracks
if sv.pt > 25:
fj.nsv_ptgt25 += 1
if sv.pt > 50:
fj.nsv_ptgt50 += 1
# sfBDT & sj12_masscor_dxysig
fj.sfBDT = -1
fj.sj12_masscor_dxysig = 0
if len(fj.subjets) == 2:
sj1, sj2 = fj.subjets
if len(sj1.sv_list) > 0 and len(sj2.sv_list) > 0:
sj1_sv, sj2_sv = sj1.sv_list[0], sj2.sv_list[0]
sfbdt_inputs = {
'fj_2_tau21': fj.tau2 / fj.tau1 if fj.tau1 > 0 else 99,
'fj_2_sj1_rawmass': sj1.mass,
'fj_2_sj2_rawmass': sj2.mass,
'fj_2_ntracks_sv12': fj.ntracks_sv12,
'fj_2_sj1_sv1_pt': sj1_sv.pt,
'fj_2_sj2_sv1_pt': sj2_sv.pt,
}
fj.sfBDT = self.xgb.eval(sfbdt_inputs, model_idx=(event.event % 10))
fj.sj12_masscor_dxysig = corrected_svmass(sj1_sv if sj1_sv.dxySig > sj2_sv.dxySig else sj2_sv)
def loadGenHistory(self, event, fatjets):
# gen matching
if not self.isMC:
return
try:
genparts = event.genparts
except RuntimeError as e:
genparts = Collection(event, "GenPart")
for idx, gp in enumerate(genparts):
if 'dauIdx' not in gp.__dict__:
gp.dauIdx = []
if gp.genPartIdxMother >= 0:
mom = genparts[gp.genPartIdxMother]
if 'dauIdx' not in mom.__dict__:
mom.dauIdx = [idx]
else:
mom.dauIdx.append(idx)
event.genparts = genparts
def isHadronic(gp):
if len(gp.dauIdx) == 0:
raise ValueError('Particle has no daughters!')
for idx in gp.dauIdx:
if abs(genparts[idx].pdgId) < 6:
return True
return False
def getFinal(gp):
for idx in gp.dauIdx:
dau = genparts[idx]
if dau.pdgId == gp.pdgId:
return getFinal(dau)
return gp
lepGenTops = []
hadGenTops = []
hadGenWs = []
hadGenZs = []
hadGenHs = []
for gp in genparts:
if gp.statusFlags & (1 << 13) == 0:
continue
if abs(gp.pdgId) == 6:
for idx in gp.dauIdx:
dau = genparts[idx]
if abs(dau.pdgId) == 24:
genW = getFinal(dau)
gp.genW = genW
if isHadronic(genW):
hadGenTops.append(gp)
else:
lepGenTops.append(gp)
elif abs(dau.pdgId) in (1, 3, 5):
gp.genB = dau
elif abs(gp.pdgId) == 24:
if isHadronic(gp):
hadGenWs.append(gp)
elif abs(gp.pdgId) == 23:
if isHadronic(gp):
hadGenZs.append(gp)
elif abs(gp.pdgId) == 25:
if isHadronic(gp):
hadGenHs.append(gp)
for parton in itertools.chain(lepGenTops, hadGenTops):
parton.daus = (parton.genB, genparts[parton.genW.dauIdx[0]], genparts[parton.genW.dauIdx[1]])
parton.genW.daus = parton.daus[1:]
for parton in itertools.chain(hadGenWs, hadGenZs, hadGenHs):
parton.daus = (genparts[parton.dauIdx[0]], genparts[parton.dauIdx[1]])
for fj in fatjets:
fj.genH, fj.dr_H = closest(fj, hadGenHs)
fj.genZ, fj.dr_Z = closest(fj, hadGenZs)
fj.genW, fj.dr_W = closest(fj, hadGenWs)
fj.genT, fj.dr_T = closest(fj, hadGenTops)
fj.genLepT, fj.dr_LepT = closest(fj, lepGenTops)
def evalTagger(self, event, jets):
for j in jets:
if self._opts['run_tagger']:
outputs = [p.predict_with_cache(self.tagInfoMaker, event.idx, j.idx, j) for p in self.pnTaggers]
outputs = ensemble(outputs, np.mean)
j.pn_Xbb = outputs['probXbb']
j.pn_Xcc = outputs['probXcc']
j.pn_Xqq = outputs['probXqq']
j.pn_QCD = convert_prob(outputs, None, prefix='prob')
else:
if self.hasParticleNetProb:
j.pn_Xbb = j.ParticleNetMD_probXbb
j.pn_Xcc = j.ParticleNetMD_probXcc
j.pn_Xqq = j.ParticleNetMD_probXqq
j.pn_QCD = convert_prob(j, None, prefix='ParticleNetMD_prob')
else:
j.pn_Xbb = j.particleNetMD_Xbb
j.pn_Xcc = j.particleNetMD_Xcc
j.pn_Xqq = j.particleNetMD_Xqq
j.pn_QCD = j.particleNetMD_QCD
j.pn_XbbVsQCD = convert_prob(j, ['Xbb'], ['QCD'], prefix='pn_')
j.pn_XccVsQCD = convert_prob(j, ['Xcc'], ['QCD'], prefix='pn_')
j.pn_XccOrXqqVsQCD = convert_prob(j, ['Xcc', 'Xqq'], ['QCD'], prefix='pn_')
def evalMassRegression(self, event, jets):
for j in jets:
if self._opts['run_mass_regression']:
outputs = [p.predict_with_cache(self.tagInfoMaker, event.idx, j.idx, j) for p in self.pnMassRegressions]
j.regressed_mass = ensemble(outputs, np.median)['mass']
else:
try:
j.regressed_mass = j.particleNet_mass
except RuntimeError:
j.regressed_mass = 0
def fillBaseEventInfo(self, event):
self.out.fillBranch("jetR", self._jetConeSize)
self.out.fillBranch("year", self.year)
self.out.fillBranch("lumiwgt", lumi_dict[self.year])
met_filters = bool(
event.Flag_goodVertices and
event.Flag_globalSuperTightHalo2016Filter and
event.Flag_HBHENoiseFilter and
event.Flag_HBHENoiseIsoFilter and
event.Flag_EcalDeadCellTriggerPrimitiveFilter and
event.Flag_BadPFMuonFilter and
event.Flag_BadPFMuonDzFilter and
event.Flag_eeBadScFilter
)
if self.year in (2017, 2018):
met_filters = met_filters and event.Flag_ecalBadCalibFilter
self.out.fillBranch("passmetfilters", met_filters)
# L1 prefire weights
if self.year <= 2017:
self.out.fillBranch("l1PreFiringWeight", event.L1PreFiringWeight_Nom)
self.out.fillBranch("l1PreFiringWeightUp", event.L1PreFiringWeight_Up)
self.out.fillBranch("l1PreFiringWeightDown", event.L1PreFiringWeight_Dn)
else:
self.out.fillBranch("l1PreFiringWeight", 1.0)
self.out.fillBranch("l1PreFiringWeightUp", 1.0)
self.out.fillBranch("l1PreFiringWeightDown", 1.0)
self.out.fillBranch("nlep", len(event.looseLeptons))
self.out.fillBranch("ht", event.ht)
self.out.fillBranch("met", event.met.pt)
self.out.fillBranch("metphi", event.met.phi)
def _get_filler(self, obj):
def filler(branch, value, default=0):
self.out.fillBranch(branch, value if obj else default)
return filler
def fillFatJetInfo(self, event, fatjets):
for idx in ([1, 2] if self._channel == 'qcd' else [1]):
prefix = 'fj_%d_' % idx
fj = fatjets[idx - 1]
if not fj.is_qualified:
# fill zeros if fatjet fails probe selection
for b in self.out._branches.keys():
if b.startswith(prefix):
self.out.fillBranch(b, 0)
continue
# fatjet kinematics
self.out.fillBranch(prefix + "is_qualified", fj.is_qualified)
self.out.fillBranch(prefix + "pt", fj.pt)
self.out.fillBranch(prefix + "eta", fj.eta)
self.out.fillBranch(prefix + "phi", fj.phi)
self.out.fillBranch(prefix + "rawmass", fj.mass)
self.out.fillBranch(prefix + "sdmass", fj.msoftdrop)
self.out.fillBranch(prefix + "regressed_mass", fj.regressed_mass)
self.out.fillBranch(prefix + "tau21", fj.tau2 / fj.tau1 if fj.tau1 > 0 else 99)
self.out.fillBranch(prefix + "tau32", fj.tau3 / fj.tau2 if fj.tau2 > 0 else 99)
self.out.fillBranch(prefix + "btagcsvv2", fj.btagCSVV2)
try:
self.out.fillBranch(prefix + "btagjp", fj.btagJP)
except RuntimeError:
self.out.fillBranch(prefix + "btagjp", -1)
# subjets
self.out.fillBranch(prefix + "deltaR_sj12", deltaR(*fj.subjets) if len(fj.subjets) == 2 else 99)
for idx_sj, sj in enumerate(fj.subjets):
prefix_sj = prefix + 'sj%d_' % (idx_sj + 1)
self.out.fillBranch(prefix_sj + "pt", sj.pt)
self.out.fillBranch(prefix_sj + "eta", sj.eta)
self.out.fillBranch(prefix_sj + "phi", sj.phi)
self.out.fillBranch(prefix_sj + "rawmass", sj.mass)
try:
self.out.fillBranch(prefix_sj + "btagdeepcsv", sj.btagDeepB)
except RuntimeError:
self.out.fillBranch(prefix_sj + "btagdeepcsv", -1)
# taggers
try:
# Full
self.out.fillBranch(prefix + "DeepAK8_TvsQCD", fj.deepTag_TvsQCD)
self.out.fillBranch(prefix + "DeepAK8_WvsQCD", fj.deepTag_WvsQCD)
self.out.fillBranch(prefix + "DeepAK8_ZvsQCD", fj.deepTag_ZvsQCD)
# MD
self.out.fillBranch(prefix + "DeepAK8MD_TvsQCD", fj.deepTagMD_TvsQCD)
self.out.fillBranch(prefix + "DeepAK8MD_WvsQCD", fj.deepTagMD_WvsQCD)
self.out.fillBranch(prefix + "DeepAK8MD_ZvsQCD", fj.deepTagMD_ZvsQCD)
self.out.fillBranch(prefix + "DeepAK8MD_ZHbbvsQCD", fj.deepTagMD_ZHbbvsQCD)
self.out.fillBranch(prefix + "DeepAK8MD_ZHccvsQCD", fj.deepTagMD_ZHccvsQCD)
self.out.fillBranch(prefix + "DeepAK8MD_bbVsLight", fj.deepTagMD_bbvsLight)
try:
bbVsTop = (1 / (1 + (fj.deepTagMD_TvsQCD / fj.deepTagMD_HbbvsQCD) * (1 - fj.deepTagMD_HbbvsQCD) / (1 - fj.deepTagMD_TvsQCD))) # noqa
except ZeroDivisionError:
bbVsTop = 0
self.out.fillBranch(prefix + "DeepAK8MD_bbVsTop", bbVsTop)
except RuntimeError:
# if no DeepAK8 branches
self.out.fillBranch(prefix + "DeepAK8_TvsQCD", -1)
self.out.fillBranch(prefix + "DeepAK8_WvsQCD", -1)
self.out.fillBranch(prefix + "DeepAK8_ZvsQCD", -1)
self.out.fillBranch(prefix + "DeepAK8MD_TvsQCD", -1)
self.out.fillBranch(prefix + "DeepAK8MD_WvsQCD", -1)
self.out.fillBranch(prefix + "DeepAK8MD_ZvsQCD", -1)
self.out.fillBranch(prefix + "DeepAK8MD_ZHbbvsQCD", -1)
self.out.fillBranch(prefix + "DeepAK8MD_ZHccvsQCD", -1)
self.out.fillBranch(prefix + "DeepAK8MD_bbVsLight", -1)
self.out.fillBranch(prefix + "DeepAK8MD_bbVsTop", -1)
try:
self.out.fillBranch(prefix + "DeepAK8_ZHbbvsQCD",
convert_prob(fj, ['Zbb', 'Hbb'], prefix='deepTag_prob'))
except RuntimeError:
# if no DeepAK8 raw probs
self.out.fillBranch(prefix + "DeepAK8_ZHbbvsQCD", -1)
# ParticleNet
if self.hasParticleNetProb:
self.out.fillBranch(prefix + "ParticleNet_TvsQCD",
convert_prob(fj, ['Tbcq', 'Tbqq'], prefix='ParticleNet_prob'))
self.out.fillBranch(prefix + "ParticleNet_WvsQCD",
convert_prob(fj, ['Wcq', 'Wqq'], prefix='ParticleNet_prob'))
self.out.fillBranch(prefix + "ParticleNet_ZvsQCD",
convert_prob(fj, ['Zbb', 'Zcc', 'Zqq'], prefix='ParticleNet_prob'))
else:
try:
# nominal ParticleNet from official NanoAOD
self.out.fillBranch(prefix + "ParticleNet_TvsQCD", fj.particleNet_TvsQCD)
self.out.fillBranch(prefix + "ParticleNet_WvsQCD", fj.particleNet_WvsQCD)
self.out.fillBranch(prefix + "ParticleNet_ZvsQCD", fj.particleNet_ZvsQCD)
except RuntimeError:
# if no nominal ParticleNet
self.out.fillBranch(prefix + "ParticleNet_TvsQCD", -1)
self.out.fillBranch(prefix + "ParticleNet_WvsQCD", -1)
self.out.fillBranch(prefix + "ParticleNet_ZvsQCD", -1)
# ParticleNet-MD
self.out.fillBranch(prefix + "ParticleNetMD_Xbb", fj.pn_Xbb)
self.out.fillBranch(prefix + "ParticleNetMD_Xcc", fj.pn_Xcc)
self.out.fillBranch(prefix + "ParticleNetMD_Xqq", fj.pn_Xqq)
self.out.fillBranch(prefix + "ParticleNetMD_QCD", fj.pn_QCD)
self.out.fillBranch(prefix + "ParticleNetMD_XbbVsQCD", fj.pn_XbbVsQCD)
self.out.fillBranch(prefix + "ParticleNetMD_XccVsQCD", fj.pn_XccVsQCD)
self.out.fillBranch(prefix + "ParticleNetMD_XccOrXqqVsQCD", fj.pn_XccOrXqqVsQCD)
if self._opts['run_tagger']:
self.out.fillBranch(prefix + "origParticleNetMD_XccVsQCD",
convert_prob(fj, ['Xcc'], None, prefix='ParticleNetMD_prob'))
self.out.fillBranch(prefix + "origParticleNetMD_XbbVsQCD",
convert_prob(fj, ['Xbb'], None, prefix='ParticleNetMD_prob'))
# matching variables
if self.isMC:
try:
sj1 = fj.subjets[0]
except IndexError:
sj1 = None
try:
sj2 = fj.subjets[1]
except IndexError:
sj2 = None
self.out.fillBranch(prefix + "nbhadrons", fj.nBHadrons)
self.out.fillBranch(prefix + "nchadrons", fj.nCHadrons)
self.out.fillBranch(prefix + "sj1_nbhadrons", sj1.nBHadrons if sj1 else -1)
self.out.fillBranch(prefix + "sj1_nchadrons", sj1.nCHadrons if sj1 else -1)
self.out.fillBranch(prefix + "sj2_nbhadrons", sj2.nBHadrons if sj2 else -1)
self.out.fillBranch(prefix + "sj2_nchadrons", sj2.nCHadrons if sj2 else -1)
try:
self.out.fillBranch(prefix + "partonflavour", fj.partonFlavour)
self.out.fillBranch(prefix + "sj1_partonflavour", sj1.partonFlavour if sj1 else -1)
self.out.fillBranch(prefix + "sj2_partonflavour", sj2.partonFlavour if sj2 else -1)
except RuntimeError:
self.out.fillBranch(prefix + "partonflavour", -1)
self.out.fillBranch(prefix + "sj1_partonflavour", -1)
self.out.fillBranch(prefix + "sj2_partonflavour", -1)
# info of the closest hadGenH
self.out.fillBranch(prefix + "dr_H", fj.dr_H)
self.out.fillBranch(prefix + "dr_H_daus",
max([deltaR(fj, dau) for dau in fj.genH.daus]) if fj.genH else 99)
self.out.fillBranch(prefix + "H_pt", fj.genH.pt if fj.genH else -1)
self.out.fillBranch(prefix + "H_decay", abs(fj.genH.daus[0].pdgId) if fj.genH else 0)
# info of the closest hadGenZ
self.out.fillBranch(prefix + "dr_Z", fj.dr_Z)
self.out.fillBranch(prefix + "dr_Z_daus",
max([deltaR(fj, dau) for dau in fj.genZ.daus]) if fj.genZ else 99)
self.out.fillBranch(prefix + "Z_pt", fj.genZ.pt if fj.genZ else -1)
self.out.fillBranch(prefix + "Z_decay", abs(fj.genZ.daus[0].pdgId) if fj.genZ else 0)
# info of the closest hadGenW
self.out.fillBranch(prefix + "dr_W", fj.dr_W)
self.out.fillBranch(prefix + "dr_W_daus",
max([deltaR(fj, dau) for dau in fj.genW.daus]) if fj.genW else 99)
self.out.fillBranch(prefix + "W_pt", fj.genW.pt if fj.genW else -1)
self.out.fillBranch(prefix + "W_decay", max([abs(d.pdgId) for d in fj.genW.daus]) if fj.genW else 0)
# info of the closest hadGenTop
drwq1, drwq2 = [deltaR(fj, dau) for dau in fj.genT.genW.daus] if fj.genT else [99, 99]
wq1_pdgId, wq2_pdgId = [dau.pdgId for dau in fj.genT.genW.daus] if fj.genT else [0, 0]
if drwq1 < drwq2:
drwq1, drwq2 = drwq2, drwq1
wq1_pdgId, wq2_pdgId = wq2_pdgId, wq1_pdgId
self.out.fillBranch(prefix + "dr_T", fj.dr_T)
self.out.fillBranch(prefix + "dr_T_b", deltaR(fj, fj.genT.genB) if fj.genT else 99)
self.out.fillBranch(prefix + "dr_T_Wq_max", drwq1)
self.out.fillBranch(prefix + "dr_T_Wq_min", drwq2)
self.out.fillBranch(prefix + "T_Wq_max_pdgId", wq1_pdgId)
self.out.fillBranch(prefix + "T_Wq_min_pdgId", wq2_pdgId)
self.out.fillBranch(prefix + "T_pt", fj.genT.pt if fj.genT else -1)
if self._fill_sv:
# SV variables
self.out.fillBranch(prefix + "nsv", len(fj.sv_list))
self.out.fillBranch(prefix + "nsv_ptgt25", fj.nsv_ptgt25)
self.out.fillBranch(prefix + "nsv_ptgt50", fj.nsv_ptgt50)
self.out.fillBranch(prefix + "ntracks", fj.ntracks)
self.out.fillBranch(prefix + "ntracks_sv12", fj.ntracks_sv12)
for idx_sj in (0, 1):
prefix_sj = prefix + 'sj%d_' % (idx_sj + 1)
try:
sj = fj.subjets[idx_sj]
except IndexError:
# fill zeros if not enough subjets
for b in self.out._branches.keys():
if b.startswith(prefix):
self.out.fillBranch(b, 0)
continue
self.out.fillBranch(prefix_sj + "ntracks", sum([sv.ntracks for sv in sj.sv_list]))
self.out.fillBranch(prefix_sj + "nsv", len(sj.sv_list))
sv = sj.sv_list[0] if len(sj.sv_list) else _NullObject()
fill_sv = self._get_filler(sv) # wrapper, fill default value if sv=None
fill_sv(prefix_sj + "sv1_pt", sv.pt)
fill_sv(prefix_sj + "sv1_mass", sv.mass)
fill_sv(prefix_sj + "sv1_masscor", corrected_svmass(sv) if sv else 0)
fill_sv(prefix_sj + "sv1_ntracks", sv.ntracks)
fill_sv(prefix_sj + "sv1_dxy", sv.dxy)
fill_sv(prefix_sj + "sv1_dxysig", sv.dxySig)
fill_sv(prefix_sj + "sv1_dlen", sv.dlen)
fill_sv(prefix_sj + "sv1_dlensig", sv.dlenSig)
fill_sv(prefix_sj + "sv1_chi2ndof", sv.chi2)
fill_sv(prefix_sj + "sv1_pangle", sv.pAngle)
self.out.fillBranch(prefix + "sj12_masscor_dxysig", fj.sj12_masscor_dxysig)
# sfBDT
self.out.fillBranch(prefix + "sfBDT", fj.sfBDT)