def buildSmoothBkgCombA(self):
"""Build with RooWorkspace.factory. See also RooFactoryWSTool.factory"""
wspace = self.getWspace()
f_bkgCombAAltA = wspace.pdf("f_bkgCombAAltA")
if f_bkgCombAAltA == None:
f_bkgCombAAltKUp = RooKeysPdf(
"f_bkgCombAAltKUp", "f_bkgCombAAltKUp", CosThetaK,
self.process.sourcemanager.get('dataReader.USB'),
RooKeysPdf.MirrorBoth, 1.0)
f_bkgCombAAltKLo = RooKeysPdf(
"f_bkgCombAAltKLo", "f_bkgCombAAltKLo", CosThetaK,
self.process.sourcemanager.get('dataReader.LSB'),
RooKeysPdf.MirrorBoth, 1.0)
f_bkgCombAAltLUp = RooKeysPdf(
"f_bkgCombAAltLUp", "f_bkgCombAAltLUp", CosThetaL,
self.process.sourcemanager.get('dataReader.USB'),
RooKeysPdf.MirrorBoth, 1.0)
f_bkgCombAAltLLo = RooKeysPdf(
"f_bkgCombAAltLLo", "f_bkgCombAAltLLo", CosThetaL,
self.process.sourcemanager.get('dataReader.LSB'),
RooKeysPdf.MirrorBoth, 1.0)
for f in f_bkgCombAAltKLo, f_bkgCombAAltKUp, f_bkgCombAAltLLo, f_bkgCombAAltLUp:
getattr(wspace, 'import')(f)
wspace.factory(
"PROD::f_bkgCombAAltAUp(f_bkgCombAAltKUp,f_bkgCombAAltLUp)")
wspace.factory(
"PROD::f_bkgCombAAltALo(f_bkgCombAAltKLo,f_bkgCombAAltLLo)")
wspace.factory(
"SUM::f_bkgCombAAltA(frac_bkgCombAAltA[0.5,0,1]*f_bkgCombAAltALo,f_bkgCombAAltAUp)"
)
f_bkgCombAAltA = wspace.pdf("f_bkgCombAAltA")
frac_bkgCombAAltA = wspace.var("frac_bkgCombAAltA")
frac_bkgCombAAltA.setVal(
self.process.sourcemanager.get('dataReader.LSB').sumEntries() /
(self.process.sourcemanager.get('dataReader.LSB').sumEntries() +
self.process.sourcemanager.get('dataReader.USB').sumEntries()))
frac_bkgCombAAltA.setConstant(True)
self.cfg['source']['frac_bkgCombAAltA'] = frac_bkgCombAAltA
self.cfg['source']['f_bkgCombAAltA'] = f_bkgCombAAltA
def main(options, args):
from ROOT import gSystem, gROOT, gStyle
gROOT.SetBatch()
gSystem.Load("libRooFitCore")
if options.doWebPage:
from lip.Tools.rootutils import loadToolsLib, apply_modifs
loadToolsLib()
from ROOT import TFile, RooFit, RooArgSet, RooDataHist, RooKeysPdf, RooHistPdf, TCanvas, TLegend, TLatex, TArrow, TPaveText, RooAddPdf, RooArgList
from ROOT import kWhite, kBlue, kOpenSquare
if options.doWebPage:
from ROOT import HtmlHelper, HtmlTag, HtmlTable, HtmlPlot
rootglobestyle.setTDRStyle()
gStyle.SetMarkerSize(1.5)
gStyle.SetTitleYOffset(1.5)
gStyle.SetPadLeftMargin(0.16)
gStyle.SetPadRightMargin(0.05)
gStyle.SetPadTopMargin(0.05)
gStyle.SetPadBottomMargin(0.13)
gStyle.SetLabelFont(42, "XYZ")
gStyle.SetLabelOffset(0.007, "XYZ")
gStyle.SetLabelSize(0.05, "XYZ")
gStyle.SetTitleSize(0.06, "XYZ")
gStyle.SetTitleXOffset(0.9)
gStyle.SetTitleYOffset(1.24)
gStyle.SetTitleFont(42, "XYZ")
##
## Read files
##
options.outdir = "%s_m%1.0f" % (options.outdir, options.mH)
if options.fp:
options.outdir += "_fp"
ncat = options.ncat
cats = options.cats
if cats is "":
categories = ["_cat%d" % i for i in range(0, ncat)]
else:
categories = ["_cat%s" % i for i in cats.split(",")]
if options.mva:
clables = {
"_cat0": ("MVA > 0.89", ""),
"_cat1": ("0.74 #leq MVA", "MVA < 0.89"),
"_cat2": ("0.545 #leq MVA", "MVA < 0.74"),
"_cat3": ("0.05 #leq MVA", "MVA < 0.545"),
"_cat4": ("Di-jet", "Tagged"),
"_cat5": ("Di-jet", "Tagged"),
"_combcat": ("All Classes", "Combined")
}
else:
clables = {
"_cat0": ("max(|#eta|<1.5", "min(R_{9})>0.94"),
"_cat1": ("max(|#eta|<1.5", "min(R_{9})<0.94"),
"_cat2": ("max(|#eta|>1.5", "min(R_{9})>0.94"),
"_cat3": ("max(|#eta|>1.5", "min(R_{9})<0.94"),
"_cat4": ("Di-jet", "Tagged"),
"_cat5": ("Di-jet", "Tagged"),
"_combcat": ("All Classes", "Combined")
}
helper = Helper()
fin = TFile.Open(options.infile)
helper.files.append(fin)
ws = fin.Get("cms_hgg_workspace")
mass = ws.var("CMS_hgg_mass")
mass.SetTitle("m_{#gamma#gamma}")
mass.setUnit("GeV")
mass.setRange(100., 150.)
mass.setBins(100, "plot")
mass.setBins(5000)
print ws
aset = RooArgSet(mass)
helper.objs.append(mass)
helper.objs.append(aset)
fitopt = (RooFit.Minimizer("Minuit2", ""), RooFit.Minos(False),
RooFit.SumW2Error(False), RooFit.NumCPU(8))
if not options.binned and not options.refit:
finpdf = TFile.Open(options.infilepdf)
helper.files.append(finpdf)
wspdf = finpdf.Get("wsig")
else:
wspdf = ws
for c in categories:
processes = ["ggh", "vbf", "wzh"]
if options.fp:
processes = ["vbf", "wzh"]
### elif clables[c][0] == "Di-jet":
### processes = [ "vbf", "ggh" ]
dsname = "sig_mass_m%1.0f%s" % (options.mH, c)
print dsname
print ws
ds = ws.data("sig_%s_mass_m%1.0f%s" %
(processes[0], options.mH, c)).Clone(dsname)
for proc in processes[1:]:
ds.append(ws.data("sig_%s_mass_m%1.0f%s" % (proc, options.mH, c)))
helper.dsets.append(ds)
if options.binned:
binned_ds = RooDataHist("binned_%s" % dsname, "binned_%s" % dsname,
aset, ds)
pdf = RooKeysPdf("pdf_%s_%s" % (dsname, f), "pdf_%s" % dsname,
mass, ds)
plot_pdf = RooHistPdf("pdf_%s" % dsname, "pdf_%s" % dsname, aset,
plot_ds)
helper.add(binned_ds, binned_ds.GetName())
else:
if options.refit:
if options.refitall and clables[c][0] != "Di-jet":
rpdfs = []
for proc in processes:
for ngaus in range(1, 4):
pp = build_pdf(ws, "%s_%s" % (c, proc), ngaus,
ngaus == 3)
pp.fitTo(
ws.data("sig_%s_mass_m%1.0f%s" %
(proc, options.mH, c)),
RooFit.Strategy(0), *fitopt)
rpdfs.append(pp)
pdf = RooAddPdf("hggpdfrel%s" % c, "hggpdfrel%s" % c,
RooArgList(*tuple(rpdfs)))
else:
if options.refitall and clables[c][0] == "Di-jet":
for ngaus in range(1, 5):
pdf = build_pdf(ws, c, ngaus, ngaus == 5)
pdf.fitTo(ds, RooFit.Strategy(0), *fitopt)
else:
for ngaus in range(1, 4):
pdf = build_pdf(ws, c, ngaus, ngaus == 3)
pdf.fitTo(ds, RooFit.Strategy(0), *fitopt)
else:
pdfs = (wspdf.pdf("hggpdfrel%s_%s" % (c, p))
for p in processes)
pdf = RooAddPdf("hggpdfrel%s" % c, "hggpdfrel%s" % c,
RooArgList(*pdfs))
helper.add(pdf, pdf.GetName())
plot_pdf = pdf.Clone("pdf_%s" % dsname)
plot_ds = RooDataHist("plot_%s" % dsname, "plot_%s" % dsname, aset,
"plot")
plot_ds.add(ds)
cdf = pdf.createCdf(aset)
hmin, hmax, hm = get_FWHM(mass, pdf, cdf, options.mH - 10.,
options.mH + 10.)
wmin, wmax = get_eff_sigma(mass, pdf, cdf, options.mH - 10.,
options.mH + 10.)
### hmin, hmax, hm = get_FWHM( points )
helper.add(plot_ds, plot_ds.GetName())
helper.add(plot_pdf, plot_pdf.GetName())
helper.add((wmin, wmax), "eff_sigma%s" % c)
helper.add((hmin, hmax, hm), "FWHM%s" % c)
helper.add(ds.sumEntries(),
"sumEntries%s" % c) # signal model integral
# data integral for PAS tables
data = ws.data("data_mass%s" % c)
helper.add(
data.sumEntries("CMS_hgg_mass>=%1.4f && CMS_hgg_mass<=%1.4f" %
(options.mH - 10., options.mH + 10.)),
"data_sumEntries%s" % c)
del cdf
del pdf
dsname = "sig_mass_m%1.0f_combcat" % options.mH
print dsname
combined_ds = helper.dsets[0].Clone(dsname)
for d in helper.dsets[1:]:
combined_ds.append(d)
if options.binned:
binned_ds = RooDataHist("binned_%s" % dsname, "binned_%s" % dsname,
aset, combined_ds)
pdf = RooKeysPdf("pdf_%s" % (dsname), "pdf_%s" % dsname, mass,
combined_ds)
plot_pdf = RooHistPdf("pdf_%s" % dsname, "pdf_%s" % dsname, aset,
plot_ds)
helper.add(binned_ds, binned_ds.GetName())
else:
#### pdf = build_pdf(ws,"_combcat")
#### pdf.fitTo(combined_ds, RooFit.Strategy(0), *fitopt )
#### plot_pdf = pdf.Clone( "pdf_%s" % dsname )
pdf = RooAddPdf(
"pdf_%s" % dsname, "pdf_%s" % dsname,
RooArgList(*(helper.histos["hggpdfrel%s" % c]
for c in categories)))
plot_pdf = pdf
cdf = pdf.createCdf(aset)
plot_ds = RooDataHist("plot_%s" % dsname, "plot_%s" % dsname, aset, "plot")
plot_ds.add(combined_ds)
wmin, wmax = get_eff_sigma(mass, pdf, cdf, options.mH - 10.,
options.mH + 10.)
hmin, hmax, hm = get_FWHM(mass, pdf, cdf, options.mH - 10.,
options.mH + 10.)
helper.add(plot_ds, plot_ds.GetName())
helper.add(plot_pdf, plot_pdf.GetName())
helper.add((wmin, wmax), "eff_sigma_combcat")
helper.add((hmin, hmax, hm), "FWHM_combcat")
helper.add(plot_ds.sumEntries(), "sumEntries_combcat")
mass.setRange("higgsrange", options.mH - 25., options.mH + 15.)
del cdf
del pdf
del helper.dsets
### label = TLatex(0.1812081,0.8618881,"#scale[0.8]{#splitline{CMS preliminary}{Simulation}}")
label = TLatex(0.7, 0.86,
"#scale[0.65]{#splitline{CMS preliminary}{Simulation}}")
label.SetNDC(1)
##
## Make web page with plots
##
if options.doWebPage:
hth = HtmlHelper(options.outdir)
hth.navbar().cell(HtmlTag("a")).firstChild().txt("..").set(
"href", "../?C=M;O=D")
hth.navbar().cell(HtmlTag("a")).firstChild().txt("home").set(
"href", "./")
tab = hth.body().add(HtmlTable())
ip = 0
for c in ["_combcat"] + categories:
### for c in categories:
if options.doWebPage and ip % 4 == 0:
row = tab.row()
ip = ip + 1
dsname = "sig_mass_m%1.0f%s" % (options.mH, c)
canv = TCanvas(dsname, dsname, 600, 600)
helper.objs.append(canv)
### leg = TLegend(0.4345638,0.6835664,0.9362416,0.9178322)
leg = TLegend(0.2, 0.96, 0.5, 0.55)
#apply_modifs( leg, [("SetLineColor",kWhite),("SetFillColor",kWhite),("SetFillStyle",0),("SetLineStyle",0)] )
hplotcompint = mass.frame(RooFit.Bins(250), RooFit.Range("higgsrange"))
helper.objs.append(hplotcompint)
helper.objs.append(leg)
plot_ds = helper.histos["plot_%s" % dsname]
plot_pdf = helper.histos["pdf_%s" % dsname]
wmin, wmax = helper.histos["eff_sigma%s" % c]
hmin, hmax, hm = helper.histos["FWHM%s" % c]
print hmin, hmax, hm
style = (RooFit.LineColor(kBlue), RooFit.LineWidth(2),
RooFit.FillStyle(0))
style_seff = (
RooFit.LineWidth(2),
RooFit.FillStyle(1001),
RooFit.VLines(),
RooFit.LineColor(15),
)
style_ds = (RooFit.MarkerStyle(kOpenSquare), )
plot_ds.plotOn(hplotcompint, RooFit.Invisible())
plot_pdf.plotOn(hplotcompint, RooFit.NormRange("higgsrange"),
RooFit.Range(wmin, wmax), RooFit.FillColor(19),
RooFit.DrawOption("F"), *style_seff)
seffleg = hplotcompint.getObject(int(hplotcompint.numItems() - 1))
plot_pdf.plotOn(hplotcompint, RooFit.NormRange("higgsrange"),
RooFit.Range(wmin, wmax), RooFit.LineColor(15),
*style_seff)
plot_pdf.plotOn(hplotcompint, RooFit.NormRange("higgsrange"),
RooFit.Range("higgsrange"), *style)
pdfleg = hplotcompint.getObject(int(hplotcompint.numItems() - 1))
plot_ds.plotOn(hplotcompint, *style_ds)
pointsleg = hplotcompint.getObject(int(hplotcompint.numItems() - 1))
iob = int(hplotcompint.numItems() - 1)
leg.AddEntry(pointsleg, "Simulation", "pe")
leg.AddEntry(pdfleg, "Parametric model", "l")
leg.AddEntry(seffleg,
"#sigma_{eff} = %1.2f GeV " % (0.5 * (wmax - wmin)), "fl")
clabel = TLatex(0.74, 0.65,
"#scale[0.65]{#splitline{%s}{%s}}" % clables[c])
clabel.SetNDC(1)
helper.objs.append(clabel)
hm = hplotcompint.GetMaximum() * 0.5 * 0.9
### hm = pdfleg.GetMaximum()*0.5
fwhmarrow = TArrow(hmin, hm, hmax, hm)
fwhmarrow.SetArrowSize(0.03)
helper.objs.append(fwhmarrow)
fwhmlabel = TPaveText(0.20, 0.58, 0.56, 0.48, "brNDC")
fwhmlabel.SetFillStyle(0)
fwhmlabel.SetLineColor(kWhite)
reducedFWHM = (hmax - hmin) / 2.3548200
fwhmlabel.AddText("FWHM/2.35 = %1.2f GeV" % reducedFWHM)
helper.objs.append(fwhmlabel)
hplotcompint.SetTitle("")
hplotcompint.GetXaxis().SetNoExponent(True)
hplotcompint.GetXaxis().SetTitle("m_{#gamma#gamma} (GeV)")
hplotcompint.GetXaxis().SetNdivisions(509)
## hplotcompint.GetYaxis().SetTitle("A.U.");
## hplotcompint.GetYaxis().SetRangeUser(0.,hplotcompint.GetMaximum()*1.4);
hplotcompint.Draw()
leg.Draw("same")
label.Draw("same")
clabel.Draw("same")
fwhmarrow.Draw("<>")
fwhmlabel.Draw("same")
plot_ds.sumEntries()
if options.doWebPage:
hpl = HtmlPlot(canv, False, "", True, True, True)
hpl.caption("<i>%s</i>" % canv.GetTitle())
row.cell(hpl)
else:
if os.path.isdir(options.outdir) is False:
os.mkdir(options.outdir)
for ext in "C", "png", "pdf":
canv.SaveAs(
os.path.join(options.outdir,
"%s.%s" % (canv.GetName(), ext)))
if "comb" in c:
ip = 0
if options.doWebPage:
print "Creating pages..."
hth.dump()
for f in helper.files:
f.Close()
gROOT.Reset()
from pprint import pprint
pprint(helper)
print 'Summary statistics per event class'
print 'Cat\tSignal\t\tData/GeV (in %3.1f+/-10)\tsigEff\tFWHM/2.35' % options.mH
sigTotal = 0.
dataTotal = 0.
for c in categories:
sigVal = helper.histos["sumEntries%s" % c]
datVal = helper.histos["data_sumEntries%s" % c]
sigTotal += sigVal
dataTotal += datVal
for c in categories:
sigVal = helper.histos["sumEntries%s" % c]
datVal = helper.histos["data_sumEntries%s" % c]
effSig = 0.5 * (helper.histos["eff_sigma%s" % c][1] -
helper.histos["eff_sigma%s" % c][0])
fwhm = (helper.histos["FWHM%s" % c][1] -
helper.histos["FWHM%s" % c][0]) / 2.3548200
print c, '\t%3.1f (%3.1f%%)\t%3.1f (%3.1f%%)\t\t\t%2.2f\t%2.2f' % (
sigVal, 100. * sigVal / sigTotal, datVal /
(10. + 10.), 100. * datVal / dataTotal, effSig, fwhm)
print "Done."
mmgData = data.reduce(RooArgSet(mmgMass))
## Put the nominal data in the workspace
mmgData.SetName('mmgData')
# w.Import(mmgData)
## Define the translated mass
mmgMassFunc = w.factory('''FormulaVar::mmgMassFunc(
"((mmgMass - mmgMode) / mmgScale + mmgMode) - fPho*phoScale/100.",
{mmgMass, mmgMode[91.2,70,110], mmgScale[1.,0.1,5], fPho, phoScale}
)''')
mmgMode = w.var('mmgMode')
mmgScale = w.var('mmgScale')
## Get the nominal model for translation, use rho=2
model = RooKeysPdf('model', 'model', mmgMass, mmgData, RooKeysPdf.NoMirror, 2)
## Sample nominal model to binned data, no Poisson fluctuation
modelHist = model.createHistogram('mmgMass', 10000)
modelDH = RooDataHist('modelDH', 'modelDH', RooArgList(mmgMass), modelHist)
## Use the sampled model to build a transformed model, interpolation order=2
tmodel = RooHistPdf('tmodel', 'tmodel', RooArgList(mmgMassFunc),
RooArgList(mmgMass), modelDH, 2)
## Find the mode of the tmodel.
mmgMode.setVal(modelHist.GetBinCenter(modelHist.GetMaximumBin()))
mmgMode.setConstant(True)
fPho.setConstant(True)
loadDatasetLb = RooDataSet(
'loadData16', 'loadData16', inN16,
RooArgSet(massLb, massBd, massbD, massBs, massKK, massKp, masskP,
tk1Pt, tk2Pt))
frame = "lbtkMass>5.3&&lbtkMass<6."
ptsel = "tk1Pt>2. && tk2Pt>2."
Bdbkg = "((bdMass>5.25&&bdMass<5.31)&&(kpiMass>0.86&&kpiMass<0.93))"
bDbkg = "(bdbarMass>5.25&&bdbarMass<5.31&&kpibarMass<0.93)"
bsbkg = "(bsMass>5.25&&bsMass<5.4&&kkMass>1.01&&kkMass<1.03)"
cutData = loadDatasetLb.reduce('&&'.join([frame, ptsel]))
bkgComb = loadDatasetLb.reduce('&&'.join([frame, ptsel]) + '&&' +
'||'.join([Bdbkg, bDbkg, bsbkg]))
space.factory('Polynomial::pPDF(lbtkMass,{c1[0.01,-5.,5.]})')
keyPDF = RooKeysPdf('kPDF', 'kPDF', massLb, bkgComb, RooKeysPdf.NoMirror,
0.1)
getattr(space, 'import')(keyPDF)
space.factory(
'SUM::totPDF(numBKG[{initNum},0.,{maxNum}]*pPDF,{bkgNum}*kPDF)'.format(
initNum=cutData.sumEntries() - bkgComb.sumEntries(),
maxNum=cutData.sumEntries() * 1.05,
bkgNum=bkgComb.sumEntries()))
fitModel = space.factory('totPDF')
massLb.setRange('leftSideband', 5.3, 5.619 - 0.03)
massLb.setRange('rightSideband', 5.619 + 0.03, 6.)
fitRes = fitModel.fitTo(cutData, RooFit.Minos(True),
RooFit.Range('leftSideband,rightSideband'),
RooFit.Save())
dataset_COMB_CORR = RooDataSet("dataset_COMB_CORR","dataset_COMB_CORR",newtree,varset_small)
key_COMB = []
hist_COMB = []
for i, bin in enumerate(decaytime_binnning):
if i == 0:
continue
start = datetime.now()
dataset_COMB_CORR_dtb_init = RooDataSet("dataset_COMB_CORR_dtb_init","Decaytime bin"+str(i),dataset_COMB_CORR,varset_small,"Dst_DTF_D0_CTAU>"+str(bin[0]*ctau)+"&&Dst_DTF_D0_CTAU<"+str(bin[1]*ctau))
dataset_COMB_CORR_dtb = Subtract_Distribution(dataset_COMB_CORR_dtb_init, DTF_D0sPi_M, LOG_D0_IPCHI2_OWNPV, str(i)+"_comb", True)
dataset_COMB_CORR_dtb.SetName("dataset_COMB_CORR_dtb")
print "Background substraction from combinatorial tool "+str(datetime.now()-start)+" \n"
hist_COMB.append(RooDataHist("hist_COMB"+str(i),"hist_COMB", RooArgSet(LOG_D0_IPCHI2_OWNPV), dataset_COMB_CORR_dtb))
key_COMB.append(RooKeysPdf("key_COMB_"+str(i), "key_COMB", LOG_D0_IPCHI2_OWNPV, dataset_COMB_CORR_dtb))
#We store shapes of Log(IPCHI2) for mathced candidates in for each decay time bins here.
wspace_2 = RooWorkspace("wspace_key_shapes")
wspace_2.Print("t")
wsfile = TFile("/afs/cern.ch/user/"+prefix+"/"+user+"/eos/lhcb/user/"+prefix+"/"+user+"/WrongSign/2015/Secondary_Key_Shapes.root", "recreate")
for k in key_COMB:
wspace_2.rfimport(k)
wspace_2.Write("wspace")
wspace_3 = RooWorkspace("wspace_hist_shapes")
wspace_3.Print("t")
wsfile = TFile("/afs/cern.ch/user/"+prefix+"/"+user+"/eos/lhcb/user/"+prefix+"/"+user+"/WrongSign/2015/Secondary_Hist_Shapes.root", "recreate")
for s in hist_COMB:
wspace_3.rfimport(s)
wspace_3.Write("wspace")
## 1. Build a gaussian model g(x|m,s)
g = w.factory('Gaussian::g(x[-10,10],gm[0,-3,3],gs[1,0.1,3])')
x = w.var('x')
## 2. Generate unbinned data from g 1 and 2 with (s,m) = (s1,m1) and (m2,s2)
data1 = g.generate(ArgSet(x), 10000)
w.var('gm').setVal(-3.)
w.var('gs').setVal(2)
data2 = g.generate(ArgSet(x), 10000)
w.var('gm').setVal(0.)
w.var('gs').setVal(1.)
## 3. Use data 1 to build Keys PDF k1(x|s1,m1)
k = KeysPdf('k', 'k', x, data1, KeysPdf.NoMirror, 2)
## 4. Use k to build HistPDF with _transformed_ x variable t = (x-m)/s
## h1((x-hm)/hs|m1,s1)
t = w.factory('FormulaVar::t("(x - hm) / hs", {x, hm[0,-5,5], hs[1,0.1,3]})')
hist = k.createHistogram('x', 10000)
dh = DataHist('dh', 'dh', ArgList(x), hist)
h = HistPdf('h', 'h', ArgList(t), ArgList(x), dh, 2)
## 5. Plot results
f1 = x.frame()
f1.SetTitle('Training Data')
data1.plotOn(f1)
g.plotOn(f1)
k.plotOn(f1, LineColor(kRed), LineStyle(kDashed))
h.plotOn(f1, LineColor(kBlack), LineStyle(kDashed))
ws_file0 = TFile("/afs/cern.ch/user/"+prefix+"/"+user+"/eos/lhcb/user/"+prefix+"/"+user+"/WrongSign/2015/WorkSpaces/Merged_WS_Bin_"+str(i)+".root", "read")
wsp = ws_file0.Get("wspace")
ws_file0.Close()
dataset_RS_dtb = wsp.data("dataset_RS")
dataset_RS_dtb.SetName("dataset_RS_dtb")
#Open secondary shape for given in
start = datetime.now()
dataset_COMB_CORR_dtb_init = RooDataSet("dataset_COMB_CORR_dtb_init","Decaytime bin"+str(i),dataset_COMB_CORR,varset_small,"Dst_DTF_D0_CTAU>"+str(bin[0]*ctau)+"&&Dst_DTF_D0_CTAU<"+str(bin[1]*ctau))
dataset_COMB_CORR_dtb = Subtract_Distribution(dataset_COMB_CORR_dtb_init, DTF_D0sPi_M, LOG_D0_IPCHI2_OWNPV, str(i)+"_comb", True)
dataset_COMB_CORR_dtb.SetName("dataset_COMB_CORR_dtb")
print "Background substraction from combinatorial tool "+str(datetime.now()-start)+" \n"
hist_COMB = RooDataHist("hist_COMB","hist_COMB", RooArgSet(LOG_D0_IPCHI2_OWNPV), dataset_COMB_CORR_dtb)
shape_COMB = RooHistPdf("shape_COMB", "shape_COMB", RooArgSet(LOG_D0_IPCHI2_OWNPV), hist_COMB)
key_COMB = RooKeysPdf("key_COMB", "key_COMB", LOG_D0_IPCHI2_OWNPV, dataset_COMB_CORR_dtb)
Prim_Yield = RooRealVar("Prim_Yield", "Prim_Yield", 1000000,0,1e7);
Sec_Yield = RooRealVar("Sec_Yield", "Sec_Yield", 0,0,1e7);
#Choise of fit model used to fit primary decays: fage (s_fage_ or histogram of 0 bin (shape_RS_0)
#Two_hists = RooAddPdf("Two_hists", "Two_hists", RooArgList(shape_RS_0, key_COMB), RooArgList(Prim_Yield, Sec_Yield))
Two_hists = RooAddPdf("Two_hists", "Two_hists", RooArgList(s_fage, key_COMB), RooArgList(Prim_Yield, Sec_Yield))
LOG_D0_IPCHI2_OWNPV.setRange("All", -10, 10)
LOG_D0_IPCHI2_OWNPV.setRange("Signal", -10, Log(9))
LOG_D0_IPCHI2_OWNPV.setRange("Background", -10, 10)
#N.B. Here, Signal region is one that we use for analysis, bagkround - is one with secondaries.
#We can play with fit limits if we fail to describe signal shape properly
# fit lbtkMass in antiLb MC {{{
if fitToLbMass_AntiLbTkMC:
fitLabel='lbtkDist_antilbtkMC'
mass=space.var('lbtkMass')
inN=inFF.Get('pLbTk/AntiLbTk')
#dataset=RooDataSet('dataset','dataset',inN,RooArgSet(mass,tk1Pt,tk2Pt)).reduce('tk1Pt>3.&&tk2Pt>2.')
dataset=RooDataSet('dataset','dataset',inN,RooArgSet(mass))
fitBkg=dataset.reduce(RooFit.Name('fitData_{0}'.format(fitLabel)))
#getattr(space,'import')(fitBkg)
# create target PDF.
#space.factory('KeysPdf::kPDF_{0}(lbtkMass,fitData_{0},NoMirror,2.0)'.format(fitLabel))
#myPDF=space.pdf('kPDF_{0}'.format(fitLabel))
keyPDF=RooKeysPdf('kPDF_{0}'.format(fitLabel),'kPDF_{0}'.format(fitLabel),mass,fitBkg,RooKeysPdf.NoMirror,2.0)
getattr(space,'import')(keyPDF)
myPDF=keyPDF
myFrame=mass.frame(RooFit.Title(fitLabel),RooFit.Title(fitLabel))
if setAxis: myFrame=mass.frame(xaxisMin,xaxisMax)
myFrame.GetXaxis().SetTitle('J/#psi p K^{-} Mass(GeV)')
dataset.plotOn(myFrame)
myPDF.plotOn(myFrame,RooFit.LineColor(2))
myFrame.Draw()
figDir.cd()
myFrame.Write(fitLabel)
canv.SaveAs(outFig)
def createKDE(self, data, htype):
starttime = time.time()
keys = RooKeysPdf('keys%s' % htype, 'keys%s' % htype, self.roopred, data)
self.pdfs[htype] = keys
print 'Creating KDE pdf for %s took %s' % (htype, time.strftime("%H:%M:%S", time.gmtime(time.time()-starttime)))
def create_pdfs(self, paths_to_data, save=True):
"""
Create KDE pdfs for network output distribution
"""
if not isinstance(paths_to_data, list):
paths_to_data = [paths_to_data]
files_H = []
files_A = []
for path in paths_to_data:
# These are merged, so only one of each
if not path.endswith('/'): path += '/'
files_H.append(path+'model_H_merged.h5')
files_A.append(path+'model_A_merged.h5')
# Find background samples
files_bgnd = []
for path in paths_to_data:
for sample in self.samples:
if sample not in ['H', 'A']:
files_bgnd.append(path+'/model_%s_merged.h5' % sample)
# Read in the data
dataH_X, _, feats = self.read_array_from_files(files_H)
dataH_X, _ = select_features(dataH_X, feats, include_mass_variables=False)
dataH_X = self.scaler.transform(dataH_X)
rho = 0.8
#ntemplateevents = 1000000
ntemplateevents = 500000
#ntemplateevents = 50000
print 'Limiting events for pdf creation to %d' % ntemplateevents
# 700k -> 4h 42m
# 500k -> 2h 30m
# 200k -> 0h 27m
# 100k -> 0h 7m
# 50k -> 0h 2m
dataH_X = dataH_X[:ntemplateevents]
# Predict, fill a RooDataHist
starttime = time.time()
predsH = self.model.predict(dataH_X)[:,1]
self.roopred = RooRealVar('roopred', 'roopred', 0, 1)
roopreddataH = RooDataSet('roopreddataH', 'roopreddataH', RooArgSet(self.roopred))
for pred in predsH:
self.roopred.setVal(pred)
roopreddataH.add(RooArgSet(self.roopred))
# Create the KDE pdfs
def createKDE(self, data, htype):
starttime = time.time()
keys = RooKeysPdf('keys%s' % htype, 'keys%s' % htype, self.roopred, data)
self.pdfs[htype] = keys
print 'Creating KDE pdf for %s took %s' % (htype, time.strftime("%H:%M:%S", time.gmtime(time.time()-starttime)))
from ROOT.RooKeysPdf import NoMirror
keysH = RooKeysPdf('keysH', 'keysH', self.roopred, roopreddataH, NoMirror, rho)
self.pdfs['H'] = keysH
# Do the same for A
dataA_X, _, feats = self.read_array_from_files(files_A)
dataA_X, _ = select_features(dataA_X, feats, include_mass_variables=False)
dataA_X = self.scaler.transform(dataA_X)
dataA_X = dataA_X[:ntemplateevents]
starttime = time.time()
predsA = self.model.predict(dataA_X)[:,1]
roopreddataA = RooDataSet('roopreddataA', 'roopreddataA', RooArgSet(self.roopred))
for pred in predsA:
self.roopred.setVal(pred)
roopreddataA.add(RooArgSet(self.roopred))
keysA = RooKeysPdf('keysA', 'keysA', self.roopred, roopreddataA, NoMirror, rho)
self.pdfs['A'] = keysA
if save:
ws = RooWorkspace('ws', 'ws')
getattr(ws, 'import')(self.roopred)
getattr(ws, 'import')(self.pdfs['H'])
getattr(ws, 'import')(self.pdfs['A'])
ws.writeToFile('%s/pdfs_%s.root' % (self.outdir, self.title))
def main(options,args):
from ROOT import gSystem, gROOT, gStyle
gROOT.SetBatch()
gSystem.Load("libRooFitCore")
if options.doWebPage:
from lip.Tools.rootutils import loadToolsLib, apply_modifs
loadToolsLib()
from ROOT import TFile, RooFit, RooArgSet, RooDataHist, RooKeysPdf, RooHistPdf, TCanvas, TLegend, TLatex, TArrow, TPaveText, RooAddPdf, RooArgList
from ROOT import kWhite, kBlue, kOpenSquare
if options.doWebPage:
from ROOT import HtmlHelper, HtmlTag, HtmlTable, HtmlPlot
rootglobestyle.setTDRStyle()
gStyle.SetMarkerSize(1.5)
gStyle.SetTitleYOffset(1.5)
gStyle.SetPadLeftMargin(0.16)
gStyle.SetPadRightMargin(0.05)
gStyle.SetPadTopMargin(0.05)
gStyle.SetPadBottomMargin(0.13)
gStyle.SetLabelFont(42,"XYZ")
gStyle.SetLabelOffset(0.007, "XYZ")
gStyle.SetLabelSize(0.05,"XYZ")
gStyle.SetTitleSize(0.06,"XYZ")
gStyle.SetTitleXOffset(0.9)
gStyle.SetTitleYOffset(1.24)
gStyle.SetTitleFont(42,"XYZ")
##
## Read files
##
options.outdir = "%s_m%1.0f" % ( options.outdir, options.mH )
if options.fp:
options.outdir += "_fp"
ncat=options.ncat
cats=options.cats
if cats is "":
categories =[ "_cat%d" % i for i in range(0,ncat) ]
else:
categories =[ "_cat%s" % i for i in cats.split(",") ]
if options.mva:
clables = { "_cat0" : ("MVA > 0.89",""),
"_cat1" : ("0.74 #leq MVA","MVA < 0.89"),
"_cat2" : ("0.545 #leq MVA","MVA < 0.74"),
"_cat3" : ("0.05 #leq MVA","MVA < 0.545"),
"_cat4" : ("Di-jet","Tagged"),
"_cat5" : ("Di-jet","Tagged"),
"_combcat" : ("All Classes","Combined")
}
else:
clables = { "_cat0" : ("max(|#eta|<1.5","min(R_{9})>0.94"),
"_cat1" : ("max(|#eta|<1.5","min(R_{9})<0.94"),
"_cat2" : ("max(|#eta|>1.5","min(R_{9})>0.94"),
"_cat3" : ("max(|#eta|>1.5","min(R_{9})<0.94"),
"_cat4" : ("Di-jet","Tagged"),
"_cat5" : ("Di-jet","Tagged"),
"_combcat" : ("All Classes","Combined")
}
helper = Helper()
fin = TFile.Open(options.infile)
helper.files.append(fin)
ws = fin.Get("cms_hgg_workspace")
mass = ws.var("CMS_hgg_mass")
mass.SetTitle("m_{#gamma#gamma}");
mass.setUnit("GeV");
mass.setRange(100.,150.)
mass.setBins(100,"plot")
mass.setBins(5000)
print ws
aset = RooArgSet(mass)
helper.objs.append( mass )
helper.objs.append( aset )
fitopt = ( RooFit.Minimizer("Minuit2", ""), RooFit.Minos(False), RooFit.SumW2Error(False), RooFit.NumCPU(8) )
if not options.binned and not options.refit:
finpdf = TFile.Open(options.infilepdf)
helper.files.append(finpdf)
wspdf = finpdf.Get("wsig")
else:
wspdf = ws
for c in categories:
processes = [ "ggh", "vbf", "wzh" ]
if options.fp:
processes = [ "vbf", "wzh" ]
### elif clables[c][0] == "Di-jet":
### processes = [ "vbf", "ggh" ]
dsname = "sig_mass_m%1.0f%s" % (options.mH,c)
print dsname
print ws
ds = ws.data( "sig_%s_mass_m%1.0f%s" % (processes[0],options.mH,c) ).Clone(dsname)
for proc in processes[1:]:
ds.append( ws.data( "sig_%s_mass_m%1.0f%s" % (proc,options.mH,c) ) )
helper.dsets.append( ds )
if options.binned:
binned_ds = RooDataHist( "binned_%s" % dsname,"binned_%s" % dsname,aset, ds)
pdf = RooKeysPdf( "pdf_%s_%s" % (dsname, f), "pdf_%s" % dsname, mass, ds )
plot_pdf = RooHistPdf( "pdf_%s" % dsname, "pdf_%s" % dsname, aset, plot_ds )
helper.add( binned_ds, binned_ds.GetName() )
else:
if options.refit:
if options.refitall and clables[c][0] != "Di-jet":
rpdfs = []
for proc in processes:
for ngaus in range(1,4):
pp = build_pdf(ws,"%s_%s" % (c,proc),ngaus,ngaus==3 )
pp.fitTo( ws.data( "sig_%s_mass_m%1.0f%s" % (proc,options.mH,c)), RooFit.Strategy(0), *fitopt )
rpdfs.append(pp)
pdf = RooAddPdf("hggpdfrel%s" % c, "hggpdfrel%s" % c, RooArgList(*tuple(rpdfs) ))
else:
if options.refitall and clables[c][0] == "Di-jet":
for ngaus in range(1,5):
pdf = build_pdf(ws,c,ngaus,ngaus==5)
pdf.fitTo(ds, RooFit.Strategy(0), *fitopt )
else:
for ngaus in range(1,4):
pdf = build_pdf(ws,c,ngaus,ngaus==3)
pdf.fitTo(ds, RooFit.Strategy(0), *fitopt )
else:
pdfs = (wspdf.pdf( "hggpdfrel%s_%s" % (c, p)) for p in processes )
pdf = RooAddPdf("hggpdfrel%s" % c, "hggpdfrel%s" % c, RooArgList(*pdfs ))
helper.add(pdf,pdf.GetName())
plot_pdf = pdf.Clone("pdf_%s" % dsname)
plot_ds = RooDataHist( "plot_%s" % dsname,"plot_%s" % dsname, aset, "plot")
plot_ds.add( ds )
cdf = pdf.createCdf(aset)
hmin, hmax, hm = get_FWHM( mass, pdf, cdf, options.mH-10., options.mH+10. )
wmin, wmax = get_eff_sigma( mass, pdf, cdf, options.mH-10., options.mH+10. )
### hmin, hmax, hm = get_FWHM( points )
helper.add( plot_ds, plot_ds.GetName() )
helper.add( plot_pdf, plot_pdf.GetName() )
helper.add( (wmin,wmax), "eff_sigma%s" % c )
helper.add( (hmin, hmax, hm), "FWHM%s" % c )
helper.add( ds.sumEntries(), "sumEntries%s" %c ) # signal model integral
# data integral for PAS tables
data = ws.data( "data_mass%s"%c)
helper.add( data.sumEntries("CMS_hgg_mass>=%1.4f && CMS_hgg_mass<=%1.4f"%(options.mH-10.,options.mH+10.)),"data_sumEntries%s"%c)
del cdf
del pdf
dsname = "sig_mass_m%1.0f_combcat" % options.mH
print dsname
combined_ds = helper.dsets[0].Clone(dsname)
for d in helper.dsets[1:]:
combined_ds.append(d)
if options.binned:
binned_ds = RooDataHist( "binned_%s" % dsname,"binned_%s" % dsname,aset, combined_ds)
pdf = RooKeysPdf( "pdf_%s" % (dsname), "pdf_%s" % dsname, mass, combined_ds )
plot_pdf = RooHistPdf( "pdf_%s" % dsname, "pdf_%s" % dsname, aset, plot_ds )
helper.add( binned_ds, binned_ds.GetName() )
else:
#### pdf = build_pdf(ws,"_combcat")
#### pdf.fitTo(combined_ds, RooFit.Strategy(0), *fitopt )
#### plot_pdf = pdf.Clone( "pdf_%s" % dsname )
pdf = RooAddPdf( "pdf_%s" % dsname, "pdf_%s" % dsname, RooArgList( *(helper.histos["hggpdfrel%s" % c] for c in categories) ) )
plot_pdf = pdf
cdf = pdf.createCdf(aset)
plot_ds = RooDataHist( "plot_%s" % dsname,"plot_%s" % dsname, aset, "plot")
plot_ds.add( combined_ds )
wmin, wmax = get_eff_sigma( mass, pdf, cdf, options.mH-10., options.mH+10. )
hmin, hmax, hm = get_FWHM( mass, pdf, cdf, options.mH-10., options.mH+10. )
helper.add( plot_ds, plot_ds.GetName() )
helper.add( plot_pdf, plot_pdf.GetName() )
helper.add( (wmin,wmax), "eff_sigma_combcat" )
helper.add( (hmin, hmax, hm), "FWHM_combcat" )
helper.add( plot_ds.sumEntries(), "sumEntries_combcat" )
mass.setRange("higgsrange",options.mH-25.,options.mH+15.);
del cdf
del pdf
del helper.dsets
### label = TLatex(0.1812081,0.8618881,"#scale[0.8]{#splitline{CMS preliminary}{Simulation}}")
label = TLatex(0.7,0.86,"#scale[0.65]{#splitline{CMS preliminary}{Simulation}}")
label.SetNDC(1)
##
## Make web page with plots
##
if options.doWebPage:
hth = HtmlHelper(options.outdir)
hth.navbar().cell( HtmlTag("a") ).firstChild().txt("..").set("href","../?C=M;O=D")
hth.navbar().cell( HtmlTag("a") ).firstChild().txt("home").set("href","./")
tab = hth.body().add( HtmlTable() )
ip = 0
for c in ["_combcat"]+categories:
### for c in categories:
if options.doWebPage and ip % 4 == 0:
row = tab.row()
ip = ip + 1
dsname = "sig_mass_m%1.0f%s" % (options.mH,c)
canv = TCanvas(dsname,dsname,600,600)
helper.objs.append(canv)
### leg = TLegend(0.4345638,0.6835664,0.9362416,0.9178322)
leg = TLegend(0.2,0.96,0.5,0.55)
#apply_modifs( leg, [("SetLineColor",kWhite),("SetFillColor",kWhite),("SetFillStyle",0),("SetLineStyle",0)] )
hplotcompint = mass.frame(RooFit.Bins(250),RooFit.Range("higgsrange"))
helper.objs.append(hplotcompint)
helper.objs.append(leg)
plot_ds =helper.histos["plot_%s" % dsname ]
plot_pdf =helper.histos["pdf_%s" % dsname ]
wmin,wmax = helper.histos["eff_sigma%s" % c ]
hmin, hmax, hm = helper.histos["FWHM%s" % c ]
print hmin, hmax, hm
style = ( RooFit.LineColor(kBlue), RooFit.LineWidth(2), RooFit.FillStyle(0) )
style_seff = ( RooFit.LineWidth(2), RooFit.FillStyle(1001), RooFit.VLines(), RooFit.LineColor(15), )
style_ds = ( RooFit.MarkerStyle(kOpenSquare), )
plot_ds.plotOn(hplotcompint,RooFit.Invisible())
plot_pdf.plotOn(hplotcompint,RooFit.NormRange("higgsrange"),RooFit.Range(wmin,wmax), RooFit.FillColor(19),
RooFit.DrawOption("F"), *style_seff)
seffleg = hplotcompint.getObject(int(hplotcompint.numItems()-1))
plot_pdf.plotOn(hplotcompint,RooFit.NormRange("higgsrange"),RooFit.Range(wmin,wmax), RooFit.LineColor(15), *style_seff)
plot_pdf.plotOn(hplotcompint,RooFit.NormRange("higgsrange"),RooFit.Range("higgsrange"),*style)
pdfleg = hplotcompint.getObject(int(hplotcompint.numItems()-1))
plot_ds.plotOn(hplotcompint,*style_ds)
pointsleg = hplotcompint.getObject(int(hplotcompint.numItems()-1))
iob = int( hplotcompint.numItems() - 1 )
leg.AddEntry( pointsleg, "Simulation", "pe" )
leg.AddEntry( pdfleg, "Parametric model", "l" )
leg.AddEntry( seffleg, "#sigma_{eff} = %1.2f GeV " % ( 0.5*(wmax-wmin) ), "fl" )
clabel = TLatex(0.74,0.65,"#scale[0.65]{#splitline{%s}{%s}}" % clables[c])
clabel.SetNDC(1)
helper.objs.append(clabel)
hm = hplotcompint.GetMaximum()*0.5*0.9
### hm = pdfleg.GetMaximum()*0.5
fwhmarrow = TArrow(hmin,hm,hmax,hm)
fwhmarrow.SetArrowSize(0.03)
helper.objs.append(fwhmarrow)
fwhmlabel = TPaveText(0.20,0.58,0.56,0.48,"brNDC")
fwhmlabel.SetFillStyle(0)
fwhmlabel.SetLineColor(kWhite)
reducedFWHM = (hmax-hmin)/2.3548200
fwhmlabel.AddText("FWHM/2.35 = %1.2f GeV" % reducedFWHM)
helper.objs.append(fwhmlabel)
hplotcompint.SetTitle("");
hplotcompint.GetXaxis().SetNoExponent(True);
hplotcompint.GetXaxis().SetTitle("m_{#gamma#gamma} (GeV)");
hplotcompint.GetXaxis().SetNdivisions(509);
## hplotcompint.GetYaxis().SetTitle("A.U.");
## hplotcompint.GetYaxis().SetRangeUser(0.,hplotcompint.GetMaximum()*1.4);
hplotcompint.Draw();
leg.Draw("same")
label.Draw("same")
clabel.Draw("same")
fwhmarrow.Draw("<>")
fwhmlabel.Draw("same")
plot_ds.sumEntries()
if options.doWebPage:
hpl = HtmlPlot(canv,False,"",True,True,True)
hpl.caption("<i>%s</i>" % canv.GetTitle())
row.cell( hpl )
else:
if os.path.isdir(options.outdir) is False:
os.mkdir(options.outdir)
for ext in "C","png","pdf":
canv.SaveAs( os.path.join(options.outdir,"%s.%s" % (canv.GetName(), ext)) )
if "comb" in c:
ip = 0
if options.doWebPage:
print "Creating pages..."
hth.dump()
for f in helper.files:
f.Close()
gROOT.Reset()
from pprint import pprint
pprint(helper)
print 'Summary statistics per event class'
print 'Cat\tSignal\t\tData/GeV (in %3.1f+/-10)\tsigEff\tFWHM/2.35'%options.mH
sigTotal=0.
dataTotal=0.
for c in categories:
sigVal = helper.histos["sumEntries%s"%c]
datVal = helper.histos["data_sumEntries%s"%c]
sigTotal+=sigVal
dataTotal+=datVal
for c in categories:
sigVal = helper.histos["sumEntries%s"%c]
datVal = helper.histos["data_sumEntries%s"%c]
effSig = 0.5*(helper.histos["eff_sigma%s"%c][1]-helper.histos["eff_sigma%s"%c][0])
fwhm = (helper.histos["FWHM%s"%c][1]-helper.histos["FWHM%s"%c][0]) / 2.3548200
print c, '\t%3.1f (%3.1f%%)\t%3.1f (%3.1f%%)\t\t\t%2.2f\t%2.2f'%(sigVal,100.*sigVal/sigTotal,datVal/(10.+10.),100.*datVal/dataTotal,effSig,fwhm)
print "Done."
mmgData = data.reduce(RooArgSet(mmgMass))
## Put the nominal data in the workspace
mmgData.SetName('mmgData')
# w.Import(mmgData)
## Define the translated mass
mmgMassFunc = w.factory('''FormulaVar::mmgMassFunc(
"((mmgMass - mmgMode) / mmgScale + mmgMode) - fPho*phoScale/100.",
{mmgMass, mmgMode[91.2,70,110], mmgScale[1.,0.1,5], fPho, phoScale}
)''')
mmgMode = w.var('mmgMode')
mmgScale = w.var('mmgScale')
## Get the nominal model for translation, use rho=2
model = RooKeysPdf('model', 'model', mmgMass, mmgData, RooKeysPdf.NoMirror, 2)
## Sample nominal model to binned data, no Poisson fluctuation
modelHist = model.createHistogram('mmgMass', 10000)
modelDH = RooDataHist('modelDH', 'modelDH', RooArgList(mmgMass), modelHist)
## Use the sampled model to build a transformed model, interpolation order=2
tmodel = RooHistPdf('tmodel', 'tmodel', RooArgList(mmgMassFunc),
RooArgList(mmgMass), modelDH, 2)
## Find the mode of the tmodel.
mmgMode.setVal(modelHist.GetBinCenter(modelHist.GetMaximumBin()))
mmgMode.setConstant(True)
fPho.setConstant(True)
