def analyze(datasets, inputDir='.'):
gROOT.SetBatch()
gStyle.SetOptStat(0)
gStyle.SetOptTitle(0)
canvas = TCanvas("c1", "c1", 800, 800)
input_ele = TFile(inputDir + '/output_electrons.root', 'read')
input_muon = TFile(inputDir + '/output_muons.root', 'read')
input_fake = TFile(inputDir + '/output_fake.root', 'read')
h_disc_ele = input_ele.Get('disc')
h_disc_muon = input_muon.Get('disc')
h_disc_fake = input_fake.Get('disc')
h_sigma_ele = input_ele.Get('sigma')
h_sigma_muon = input_muon.Get('sigma')
h_sigma_fake = input_fake.Get('sigma')
h_disc_signal = {}
h_sigma_signal = {}
for dataset in datasets:
if not dataset.startswith('higgsino'): continue
fin = TFile(inputDir + '/output_' + dataset + '.root', 'read')
h_disc = fin.Get('disc')
h_sigma = fin.Get('sigma')
h_disc.SetDirectory(0)
h_sigma.SetDirectory(0)
h_disc_signal[dataset] = h_disc
h_sigma_signal[dataset] = h_sigma
fin.Close()
roc_curves = { 'disc_%d_%s' % (i, sig) : getROC(h_disc_ele, h_disc_muon, h_disc_fake, h_disc_signal[sig], i) for i in [4, 5, 6] for sig in h_disc_signal }
roc_curves.update(
{ 'sigma_%d_%s' % (i, sig) : getROC(h_sigma_ele, h_sigma_muon, h_sigma_fake, h_sigma_signal[sig], i) for i in [4, 5, 6] for sig in h_disc_signal }
)
fout = TFile('durp.root', 'recreate')
for curve_name in roc_curves:
roc_curves[curve_name][0].Write(curve_name)
print 'Optimal cut for', curve_name, '=', roc_curves[curve_name][1]
m = TMarker(roc_curves[curve_name][2], roc_curves[curve_name][3], 29)
m.Write(curve_name + '_optimal')
fout.Close()
for i in [4, 5, 6]:
plotStack(h_disc_ele, h_disc_muon, h_disc_fake, h_disc_signal['higgsino_300_1000'], i, 'discriminant', canvas, rebin=5)
plotStack(h_sigma_ele, h_sigma_muon, h_sigma_fake, h_sigma_signal['higgsino_300_1000'], i, 'sigma', canvas, rebin=5)
for dataset in h_disc_signal:
for i in [4, 5, 6]:
wp = getWP(h_disc_ele, h_sigma_ele, h_disc_muon, h_sigma_muon, h_disc_fake, h_sigma_fake, h_disc_signal[dataset], h_sigma_signal[dataset], i)
print dataset, '--', wp
iClone = -1
print '\nComparing disc < 0.3 to fiducial map method:\n'
printChanges('Electrons nLayers=4', h_sigma_ele, h_disc_ele, 0.3, 4, iClone)
printChanges('Electrons nLayers=5', h_sigma_ele, h_disc_ele, 0.3, 5, iClone)
printChanges('Electrons nLayers>=6', h_sigma_ele, h_disc_ele, 0.3, 6, iClone)
printChanges('Muons nLayers=4', h_sigma_muon, h_disc_muon, 0.3, 4, iClone)
printChanges('Muons nLayers=5', h_sigma_muon, h_disc_muon, 0.3, 5, iClone)
printChanges('Muons nLayers>=6', h_sigma_muon, h_disc_muon, 0.3, 6, iClone)
printChanges('Fakes nLayers=4', h_sigma_fake, h_disc_fake, 0.3, 4, iClone)
printChanges('Fakes nLayers=5', h_sigma_fake, h_disc_fake, 0.3, 5, iClone)
printChanges('Fakes nLayers>=6', h_sigma_fake, h_disc_fake, 0.3, 6, iClone)
extra_samples = {
10 : ['0p' + str(i) for i in range(2, 10)] + [str(i) for i in range(0, 10)],
100 : [str(i*10) for i in range(2, 10)],
1000 : [str(i*100) for i in range(2, 10)],
10000 : [str(i*1000) for i in range(2, 10)],
}
for i in [4, 5, 6]:
print 'NLAYERS: ', i
for mass in range(100, 1000, 100):
for lifetime in [10, 100, 1000, 10000]:
datasetName = 'higgsino_%d_%d' % (mass, lifetime)
change_string = printChanges('Higgsino_%dGeV_%dcm_94X' % (mass, lifetime),
h_sigma_signal[datasetName],
h_disc_signal[datasetName],
0.3,
i,
iClone)
for extra_lifetime in extra_samples[lifetime]:
print change_string.replace('Higgsino_%dGeV_%dcm_94X' % (mass, lifetime), 'Higgsino_%dGeV_%scm_94X' % (mass, extra_lifetime))
continue
chi2 /= N
if chi2 < minChi2 or minChi2 < 0.0:
minChi2 = chi2
minChi2HitAndTOBDrop = i
minChi2Index = copy.deepcopy(index)
g.SetPoint(n, i, chi2)
n += 1
print "(" + str(iteration + 1) + " / " + str(
nIterations) + ") minimum chi2: " + str(minChi2) + " at " + str(
minChi2HitAndTOBDrop)
m = TMarker(minChi2HitAndTOBDrop, minChi2, 29)
m.SetMarkerSize(3)
hitDropGridLower = hitDropGrid[
minChi2Index - 1] if minChi2Index > 0 else hitDropGrid[
minChi2Index] - hitDropGrid[minChi2Index + 1]
hitDropGridUpper = hitDropGrid[minChi2Index + 1] if minChi2Index < len(
hitDropGrid) - 1 else hitDropGrid[minChi2Index] + hitDropGrid[
minChi2Index - 1]
hitDropGrid = numpy.linspace(max(hitDropGridLower, 0.0),
hitDropGridUpper, 5)
foutMode = "update" if iteration > 0 else "recreate"
fout = TFile.Open("hipChi2.root", foutMode)
fout.cd()
g.Write("chi2VsHitDropProbability_" + str(iteration))
m.Write("bestFitPoint_" + str(iteration))
fout.Close()
