forked from Wilsker/TTbarCalib
-
Notifications
You must be signed in to change notification settings - Fork 1
/
runUnfoldDiscriminatorShape.py
436 lines (379 loc) · 17.2 KB
/
runUnfoldDiscriminatorShape.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
import optparse
import os,sys
import json
import commands
import ROOT
import random
import pickle
from Templated_btagEffFitter import SLICEBINS,SLICEVAR,SYSTVARS
"""
"""
def showFitResult(model,data,frame,pullFrame,nfreePars,labels=[]):
c = ROOT.TCanvas('c','c',500,500)
p1 = ROOT.TPad('p1','p1',0.0,0.85,1.0,0.0)
p1.Draw()
p1.SetTopMargin(0.01)
p1.SetRightMargin(0.05)
c.cd()
p2 = ROOT.TPad('p2','p2',0.0,0.85,1.0,1.0)
p2.SetBottomMargin(0.01)
p2.SetRightMargin(0.05)
p2.Draw()
#compute chi2 from comparison of data and model
p1.cd()
p1.Clear()
data.plotOn(frame)
model.plotOn(frame,ROOT.RooFit.ProjWData(data))
frame.Draw()
chi2=frame.chiSquare(nfreePars)
label = ROOT.TLatex()
label.SetNDC()
label.SetTextFont(42)
label.SetTextSize(0.035)
label.DrawLatex(0.2,0.95,'#bf{CMS} #it{simulation}')
label.DrawLatex(0.2,0.90,' %d variables' % nfreePars)
label.DrawLatex(0.2,0.85,'#chi^{2}/ndof=%3.2f' % (chi2) )
for i in xrange(0,len(labels)):
label.DrawLatex(0.2,0.80-i*5,labels[i])
#show pull
p2.cd()
p2.Clear()
hpull = frame.pullHist()
pullFrame.addPlotable(hpull,"P") ;
pullFrame.Draw()
pullFrame.GetYaxis().SetTitle("Pull")
pullFrame.GetYaxis().SetTitleSize(0.2)
pullFrame.GetYaxis().SetLabelSize(0.2)
pullFrame.GetXaxis().SetTitleSize(0)
pullFrame.GetXaxis().SetLabelSize(0)
pullFrame.GetYaxis().SetTitleOffset(0.15)
pullFrame.GetYaxis().SetNdivisions(4)
pullFrame.GetYaxis().SetRangeUser(-3.1,3.1)
pullFrame.GetXaxis().SetTitleOffset(0.8)
c.Modified()
c.Update()
return c,chi2
"""
get distributions from file
"""
def getDiscriminators(opt,islice,syst=''):
fIn=ROOT.TFile.Open(opt.input)
othersH=fIn.Get('other_pass0_slice%d%s' % (islice,syst)).Clone('others')
othersH.Add(fIn.Get('c_pass0_slice%d%s' % (islice,syst)))
othersH.SetDirectory(0)
bH=fIn.Get('b_pass0_slice%d%s' % (islice,syst)).Clone('b')
bH.SetDirectory(0)
fIn.Close()
return othersH,bH
"""
"""
def runSystematics(opt,tag,ws):
print 'Running systematics for',tag
biasResults={'sig':{},'bkg':{}}
#fix parameters
ws.loadSnapshot('b_pars')
ws.loadSnapshot('nonb_pars')
allVars=ws.allVars()
varIter = allVars.createIterator()
var=varIter.Next()
while var :
varName=var.GetName()
if not varName in ['nsig','nbkg'] :
ws.var(varName).setConstant()
var=varIter.Next()
#import data
variables=ROOT.RooArgSet()
variables.add(ws.var('KIN'))
variables.add(ws.factory('bias[0,0,1]'))
variables.add(ws.factory("wgt[0,0,9999999.]") )
lumi=1000.
for syst in SYSTLIST:
othersH,bH=getDiscriminators(opt,sliceCond=tag,syst=syst)
#create pseudo_data
pseudoData=ROOT.RooDataSet("pseudo_data_%s"%syst,"pseudo_data_%s"%syst,variables,"wgt")
nexp=int(lumi*othersH.Integral())
for i in xrange(0,nexp):
ws.var('KIN').setVal( othersH.GetRandom() )
ws.var('bias').setVal( random.uniform(0,1) )
ws.var('wgt').setVal(1.0)
argset = ROOT.RooArgSet()
for var in ['KIN','bias','wgt'] : argset.add( ws.var(var) )
pseudoData.add(argset, ws.var("wgt").getVal())
nexp=int(lumi*bH.Integral())
for i in xrange(0,nexp):
ws.var('KIN').setVal( bH.GetRandom())
ws.var('bias').setVal( random.uniform(0,1) )
ws.var('wgt').setVal(1.0)
argset = ROOT.RooArgSet()
for var in ['KIN','bias','wgt']: argset.add( ws.var(var) )
pseudoData.add(argset, ws.var("wgt").getVal())
#unfold distribution with the sPlot technique
sData = ROOT.RooStats.SPlot("s"+pseudoData.GetName(),"An SPlot from KIN",
pseudoData,ws.pdf('model'),
ROOT.RooArgList(ws.var('nsig'),ws.var('nbkg'))
)
#the weighted data for signal and background species
for ipop in ['sig','bkg']:
ipopData = ROOT.RooDataSet(ipop+'bias',ipop+'bias',pseudoData,pseudoData.get(),'','n%s_sw'%ipop)
biasH=ROOT.TH1F('biasH',';bias;',25,0,1)
ipopData.fillHistogram(biasH,ROOT.RooArgList(ws.var('bias')))
maxProb,maxProbFunc=0.,'pol0'
for iord in xrange(0,10):
func='pol%d'%iord
biasH.Fit(func,'0MRQ+')
iprob=biasH.GetFunction(func).GetProb()
if iprob<maxProb : continue
maxProb=iprob
maxProbFunc=func
biasResults[ipop][syst]=biasH.GetFunction(maxProbFunc).Clone('bias_%s_%s'%(ipop,syst))
biasH.Delete()
sData.Delete()
pseudoData.Delete()
fOut=ROOT.TFile.Open('bias_%s.root'%tag,'RECREATE')
for ipop in biasResults:
for syst in biasResults[ipop]:
biasResults[ipop][syst].Write()
fOut.Close()
"""
"""
def unfoldDiscriminatorShape(opt,tag,ws):
#fix parameters
ws.loadSnapshot('b_pars')
ws.loadSnapshot('nonb_pars')
allVars=ws.allVars()
varIter = allVars.createIterator()
var=varIter.Next()
while var :
varName=var.GetName()
if not varName in ['nsig','nbkg'] :
ws.var(varName).setConstant()
var=varIter.Next()
#import data from trees
files=os.listdir(opt.data)
chain=ROOT.TChain('kin')
for f in files:
if 'Data13' in f: chain.Add(os.path.join(opt.data,f))
variables=ROOT.RooArgSet()
variables.add(ws.var('KIN'))
variables.add(ws.factory('csv[0,0,1]'))
variables.add(ws.factory("wgt[0,0,9999999.]") )
data=ROOT.RooDataSet("data","data",variables,'wgt')
for i in xrange(0,chain.GetEntries()):
chain.GetEntry(i)
if not chain.ttbar_chan in [-11*13]: continue #-13*13, -11*13 -11*11] : continue
if chain.csv<0 : continue
ws.var('KIN').setVal( chain.kindisc[0] )
ws.var('csv').setVal( chain.csv )
ws.var('wgt').setVal(1.0)
argset = ROOT.RooArgSet()
for var in ['KIN','csv','wgt'] : argset.add( ws.var(var) )
data.add(argset, ws.var("wgt").getVal())
frame=ws.var('KIN').frame()
pullFrame=ws.var('KIN').frame()
c,_ = showFitResult(ws.pdf('model'),data,frame,pullFrame,2)
c.SaveAs('datafit_%s.png'%tag)
c.SaveAs('datafit_%s.pdf'%tag)
#unfold distribution with the sPlot technique
sData = ROOT.RooStats.SPlot("s"+data.GetName(),"An SPlot from KIN",
data,ws.pdf('model'),
ROOT.RooArgList(ws.var('nsig'),ws.var('nbkg'))
)
#the weighted data for signal and background species
biasF=ROOT.TFile('bias_%s.root'%tag)
for ipop in ['sig','bkg']:
ipopdata = ROOT.RooDataSet('%scsv'%ipop,'%scsv'%ipop,data,data.get(),'','n%s_sw'%ipop)
csvH = ROOT.TH1F('%s_csv'%ipop,';CSV;PDF',15,0,1)
ipopdata.fillHistogram(csvH,ROOT.RooArgList(ws.var('csv')))
histos={}
for syst in SYSTLIST:
corrF=biasF.Get('bias_%s_%s'%(ipop,syst))
histos[syst]=csvH.Clone('%s_csv%s'%(ipop,syst))
histos[syst].Reset('ICE')
for xbin in xrange(1,csvH.GetXaxis().GetNbins()+1):
corr=corrF.Eval(csvH.GetXaxis().GetBinCenter(xbin))
val=csvH.GetBinContent(xbin)*corr
err=csvH.GetBinError(xbin)*corr
histos[syst].SetBinContent(xbin,val)
histos[syst].SetBinError(xbin,err)
histos[syst].Scale(1./histos[syst].Integral())
errUp=[0]*csvH.GetXaxis().GetNbins()
errDn=[0]*csvH.GetXaxis().GetNbins()
for xbin in xrange(1,csvH.GetXaxis().GetNbins()+1):
cen=histos[''].GetBinContent(xbin)
for syst in SYSTLIST:
if syst=='':
statUnc=histos[''].GetBinError(xbin)
errUp[xbin-1]=errUp[xbin-1]+statUnc**2
errDn[xbin-1]=errDn[xbin-1]+statUnc**2
else:
diff=histos[syst].GetBinContent(xbin)-cen
if diff>0 : errUp[xbin-1] = errUp[xbin-1]+diff**2
else : errDn[xbin-1] = errDn[xbin-1]+diff**2
csvStatGr=ROOT.TGraphAsymmErrors(histos[''])
csvStatGr.SetName('csvstatunc')
csvTotalGr=csvStatGr.Clone('csvtotalunc')
for np in xrange(0,csvTotalGr.GetN()):
ex=0.5*csvH.GetXaxis().GetBinWidth(np+1)
csvTotalGr.SetPointError(np,ex,ex,ROOT.TMath.Sqrt(errDn[np]),ROOT.TMath.Sqrt(errUp[np]))
c.Clear()
csvTotalGr.SetFillStyle(1001)
csvTotalGr.SetFillColor(ROOT.kGray)
csvTotalGr.Draw('a2')
csvTotalGr.GetXaxis().SetTitle('CSV discriminator')
csvTotalGr.GetYaxis().SetTitle('Events')
csvTotalGr.GetXaxis().SetRangeUser(0,1)
csvTotalGr.GetYaxis().SetRangeUser(0,histos[''].GetMaximum()*1.3)
csvStatGr.SetMarkerStyle(20)
csvStatGr.SetFillStyle(1001)
csvStatGr.SetFillColor(ROOT.kOrange+1)
csvStatGr.Draw('2')
histos[''].Draw('histsame')
label = ROOT.TLatex()
label.SetNDC()
label.SetTextFont(42)
label.SetTextSize(0.035)
label.DrawLatex(0.12,0.96,'#bf{CMS} #it{preliminary}')
leg=ROOT.TLegend(0.12,0.91,0.5,0.94)
leg.SetTextFont(42)
leg.SetNColumns(2)
leg.SetBorderSize(0)
leg.SetTextSize(0.035)
leg.SetFillStyle(0)
leg.AddEntry(csvStatGr,'stat','lf')
leg.AddEntry(csvTotalGr,'total','f')
leg.Draw()
c.Modified()
c.Update()
raw_input()
"""
"""
def prepareTemplates(opt,islice):
ws = ROOT.RooWorkspace('w_slice%d'%islice)
KIN = ws.factory('KIN[0,-1,1]')
ws.factory("RooFormulaVar::KIN_shift('@0+1.0',{KIN})")
othersH,bH = getDiscriminators(opt,islice)
data={
'b':ROOT.RooDataHist('b', 'b', ROOT.RooArgList(KIN),bH),
'nonb':ROOT.RooDataHist('nonb','nonb',ROOT.RooArgList(KIN),othersH)
}
#prepare options for the fit
rllist = ROOT.RooLinkedList()
rllist.Add(ROOT.RooFit.Save())
rllist.Add(ROOT.RooFit.Strategy(2))
#prepare base models
ws.factory("SUM::nonb_model("
"nonb_frac1[0,0,1]*RooArgusBG::nonb_argus(KIN_shift,nonb_endpoint[2.0,1.90,2.1],nonb_shape[-0.5,-10,-0.]),"
"nonb_frac2[0,0,1]*RooExponential::nonb_expo(KIN_shift,nonb_lambda[-0.5,-10,-0.]),"
"Chebychev::nonb_pol(KIN,{nonb_a[0,-10,10]})"
")")
ws.factory("SUM::b_model("
"b_frac1[0,0,1]*RooArgusBG::b_argus(KIN_shift,b_endpoint[2.0,1.90,2.1],b_shape[-0.5,-50,-0.]),"
"b_frac2[0,0,0.1]*RooGaussian::b_gauss(KIN_shift,b_mean[0.0],b_sigma[0.05,0.01,0.1]),"
"Chebychev::b_pol(KIN,{b_a[0,-10,10]})"
")")
#loop to find smallest bias model
for dist in ['nonb','b']:
polArgs=''
ws.factory('%s_polorder[1]'%dist)
minChi2=999999.
for i in xrange(1,8):
if len(polArgs)>0 : polArgs+=','
#specialize model
modelName='%s_%d' % (dist,i)
polArgs+='%s_a%d[0,-10,10]' % (dist,i)
ws.factory("EDIT::%s(%s_model,%s_pol=Chebychev::%s_pol_%d(KIN,{%s}))" % (modelName,dist,dist,dist,i,polArgs))
#fit
fitRes=ws.pdf(modelName).chi2FitTo(data[dist],rllist)
floatParsFinal=fitRes.floatParsFinal()
nPars=len(floatParsFinal)
c,chi2 = showFitResult(model=ws.pdf(modelName),
data=data[dist],
frame=KIN.frame(),
pullFrame=KIN.frame(),
nfreePars=nPars,
labels=['%s distribution' % dist])
#save if it minimizes the chi2
if chi2>minChi2 : continue
minChi2=chi2
ws.var('%s_polorder'%dist).setVal(i)
parsToFreeze=ROOT.RooArgSet(floatParsFinal)
ws.saveSnapshot('%s_pars' % dist,parsToFreeze)
c.SaveAs('%s_%s_model.pdf' % (dist,sliceCond))
#build model to fit data
ws.factory('nsig[10,0,9999999999999.]')
ws.factory('nbkg[10,0,9999999999999.]')
model=ROOT.RooAddPdf("model","b+nonb model",
ROOT.RooArgList( ws.pdf("b_%d" % ws.var('b_polorder').getVal()), ws.pdf("nonb_%d" % ws.var('nonb_polorder').getVal()) ),
ROOT.RooArgList( ws.var("nsig"), ws.var("nbkg") )
)
getattr(ws,'import')(model)
#save workspace
ws.writeToFile(os.path.join(opt.outDir,'KINworkspace.root'),False)
"""
Wrapper to be used when run in parallel
"""
def runSystematicsPacked(args):
opt, tag, ws = args
return runSystematics(opt=opt,tag=tag,ws=ws)
"""
steer the script
"""
def main():
ROOT.gStyle.SetOptStat(0)
ROOT.gStyle.SetOptTitle(0)
ROOT.gROOT.SetBatch(True)
ROOT.RooMsgService.instance().setSilentMode(True)
ROOT.RooMsgService.instance().getStream(0).removeTopic(ROOT.RooFit.Minimization)
ROOT.RooMsgService.instance().getStream(1).removeTopic(ROOT.RooFit.Minimization)
ROOT.RooMsgService.instance().getStream(1).removeTopic(ROOT.RooFit.ObjectHandling)
ROOT.RooMsgService.instance().getStream(1).removeTopic(ROOT.RooFit.DataHandling)
ROOT.RooMsgService.instance().getStream(1).removeTopic(ROOT.RooFit.Fitting)
ROOT.RooMsgService.instance().getStream(1).removeTopic(ROOT.RooFit.Plotting)
ROOT.RooMsgService.instance().getStream(0).removeTopic(ROOT.RooFit.Plotting)
ROOT.RooMsgService.instance().getStream(0).removeTopic(ROOT.RooFit.InputArguments)
ROOT.RooMsgService.instance().getStream(1).removeTopic(ROOT.RooFit.InputArguments)
ROOT.RooMsgService.instance().getStream(0).removeTopic(ROOT.RooFit.Eval)
ROOT.RooMsgService.instance().getStream(1).removeTopic(ROOT.RooFit.Eval)
ROOT.RooMsgService.instance().getStream(1).removeTopic(ROOT.RooFit.Integration)
ROOT.RooMsgService.instance().getStream(1).removeTopic(ROOT.RooFit.NumIntegration)
ROOT.RooMsgService.instance().getStream(1).removeTopic(ROOT.RooFit.NumIntegration)
#configuration
usage = 'usage: %prog [options]'
parser = optparse.OptionParser(usage)
parser.add_option('-i', '--input', dest='input', help='input file', default=None, type='string')
parser.add_option('-o', '--output', dest='outDir', help='output directory', default=None, type='string')
parser.add_option( '--ws', dest='ws', help='input file with workspace', default=None, type='string')
parser.add_option( '--data', dest='data', help='input directory with data', default=None, type='string')
parser.add_option( '--doBias', dest='doBias', help='compute bias', action='store_true')
parser.add_option('-n', '--njobs', dest='njobs', help='# jobs to run in parallel', default=0, type='int')
(opt, args) = parser.parse_args()
#recreate workspace
if opt.ws is None:
fOut=ROOT.TFile('KINworkspace.root','RECREATE')
fOut.Close()
for islice in xrange(0,len(SLICEBINS)):
prepareTemplates(opt=opt,islice=islice)
opt.ws='KINworkspace.root'
#read workspace from file
fIn=ROOT.TFile(opt.ws)
#compute bias, if required
# if opt.doBias:
# task_list=[]
# for sliceCond in SLICES:
# task_list.append( (opt,sliceCond,fIn.Get('w_%s'%sliceCond) ) )
#run the analysis jobs
# if opt.njobs == 0:
# for opt, tag, ws in task_list:
# runSystematics(opt,tag,ws)
# else:
# from multiprocessing import Pool
# pool = Pool(opt.njobs)
# pool.map(runSystematicsPacked, task_list)
#unfold
# for sliceCond in SLICES:
# unfoldDiscriminatorShape(opt,sliceCond,fIn.Get('w_%s'%sliceCond))
"""
for execution from another script
"""
if __name__ == "__main__":
sys.exit(main())