def nevents_plot():
in_f = ROOT.TFile(gp.fn)
out_f = ROOT.TFile('nevents.root', 'recreate')
kinds, masses, taus, nmasses, ntaus = axes(in_f)
for kind in kinds:
h = ROOT.TH2D('nevents_%s' % kind, ';mass (GeV);#tau (mm)', nmasses, masses, ntaus, taus)
for ibin in xrange(1, nmasses+1):
mass = h.GetXaxis().GetBinLowEdge(ibin)
for jbin in xrange(1, ntaus+1):
tau = h.GetYaxis().GetBinLowEdge(jbin)
try:
isample = name2isample(in_f, details2name(kind, tau, mass))
except ValueError:
continue
nev = nevents(in_f, isample)
h.SetBinContent(ibin, jbin, nev)
h.SetBinError (ibin, jbin, nev**0.5)
out_f.cd()
h.Write()
out_f.Write()
out_f.Close()
def __init__(self,
data_fn,
mc_fn,
data_path='pileup',
mc_path='PileupDist/h_npu',
tol=1e-9,
raise_tol=True):
from JMTucker.Tools.ROOTTools import ROOT
self.data_f = ROOT.TFile(data_fn)
self.mc_f = ROOT.TFile(mc_fn)
self.data_h = self.data_f.Get(data_path)
self.mc_h = self.mc_f.Get(mc_path)
def norm(h):
h = h.Clone(h.GetName() + '_norm')
h.Sumw2()
h.Scale(1 / h.Integral(1, h.GetNbinsX() + 1))
return h
self.data_h_orig = self.data_h
self.mc_h_orig = self.mc_h
self.data_h = norm(self.data_h)
self.mc_h = norm(self.mc_h)
self.ndata = self.data_h.GetNbinsX()
assert self.ndata == self.mc_h.GetNbinsX() # JMTBAD
self.weights = []
for i in xrange(1, self.ndata + 1):
d = self.data_h.GetBinContent(i)
m = self.mc_h.GetBinContent(i)
w = -1
if m == 0:
if d > tol:
msg = 'm == 0 and d = %g > tol=%g for i = %i' % (d, tol, i)
if raise_tol:
raise ValueError(msg)
else:
print msg
w = 0
else:
w = d / m
self.weights.append(w)
while self.weights[-1] == 0:
self.weights.pop()
def points(f=None):
if f is None:
f = ROOT.TFile(gp.fn)
kinds = sorted(set(s.kind for s in sample_iterator(f)))
masses = sorted(set(s.mass for s in sample_iterator(f)))
taus = sorted(set(s.tau for s in sample_iterator(f)))
return kinds, masses, taus
def event_scale_factors():
from signal_efficiency import SignalEfficiencyCombiner
for which, years in ('2017p8', [2017,2018]), :
combiner = SignalEfficiencyCombiner(years)
in_f = combiner.inputs[0].f
out_f = ROOT.TFile('event_TrackMover_scale_factors_%s.root' % which, 'recreate')
kinds, masses, taus, nmasses, ntaus = axes(in_f)
taus.remove(30.)
ntaus -= 1
for kind in kinds:
h = ROOT.TH2D('event_TrackMover_scale_factors_%s' % kind, ';mass (GeV);#tau (mm)', nmasses, masses, ntaus, taus)
for ibin in xrange(1, nmasses+1):
mass = h.GetXaxis().GetBinLowEdge(ibin)
for jbin in xrange(1, ntaus+1):
tau = h.GetYaxis().GetBinLowEdge(jbin)
pt = kind, tau, mass
try:
name = details2name(*pt)
isample = name2isample(in_f, name)
except ValueError as exc:
if tau < 100:
print colors.warning('problem getting %r : isample %r exc %s' % (pt, isample, exc))
continue # leave holes in the plot to know which samples are missing
event_SF = None
if which == '2017p8' :
vtx_SF_2017 = sig_datamcSF_2017p8(name+"_2017")
vtx_SF_2018 = sig_datamcSF_2017p8(name+"_2018")
event_SF_2017 = vtx_SF_2017**2
event_SF_2018 = vtx_SF_2018**2
lumi2017 = gp.int_lumis[gp.years.index("2017")]
lumi2018 = gp.int_lumis[gp.years.index("2018")]
event_SF = (event_SF_2017*lumi2017 + event_SF_2018*lumi2018) / (lumi2017+lumi2018)
# for table of SFs, with bins 0.1--0.3\mm & 0.3--1\mm & 1--10\mm & 10--100\mm,
# take the ~midpoint of each bin and put its SF in the table (averaged where necessary)
if mass == 1600 :
if tau == 0.2 or tau == 0.6 or tau == 0.7 or tau == 4 or tau == 7 or tau == 55 :
print("for SF table:",name, round(event_SF, 4))
else :
sys.exit('event_scale_factors not set up to handle %s! Exiting.' % which)
h.SetBinContent(ibin, jbin, event_SF)
h.SetBinError (ibin, jbin, 0) # JMTBAD
out_f.cd()
h.Write()
out_f.Write()
out_f.Close()
def make():
def warning():
return
for i in xrange(20):
print colors.error("don't forget: anything?")
warning()
assert not os.path.exists(gp.fn)
ROOT.TH1.AddDirectory(0)
f = ROOT.TFile(gp.fn, 'recreate')
make_bkg(f)
name_list = {}
make_signals_2015p6(f, name_list)
make_signals_2017p8(f, name_list)
title = name_list.pop('title')
isamples = sorted([s.isample for s in name_list.itervalues()],
reverse=True)
assert isamples == range(-1, isamples[-1] - 1, -1)
nsigs = len(isamples)
h = ROOT.TH1C('name_list', title, nsigs, 0, nsigs)
missing = defaultdict(list)
for s in name_list.itervalues():
miss = tuple(sorted(set(gp.years) - s.years))
missing[miss].append(s.name)
c = sum((int(year in s.years) << y) for y, year in enumerate(gp.years))
h.GetXaxis().SetBinLabel(-s.isample, s.name)
h.SetBinContent(-s.isample, c)
f.cd()
h.Write()
f.Close()
if missing:
w = colors.warning
print w('missing summary:')
for k in sorted(missing):
if not k:
continue
l = sorted(missing[k])
n = len(l)
print w(
repr(tuple(int(y) for y in sorted(k))) + ': %i samples' % n)
for i in xrange(0, n, 6):
for j in xrange(i, min(i + 6, n)):
print w(' ' + l[j]),
print
warning()
def draw():
ps = plot_saver(plot_dir('o2t_templates_run2'), size=(600,600))
f = ROOT.TFile(gp.fn)
whiches = [
('multijet', [
(name2isample(f, 'mfv_neu_tau00100um_M0800'), ROOT.kRed, 'multijet M = 800 GeV, #tau = 100 #mum'),
(name2isample(f, 'mfv_neu_tau00300um_M0800'), ROOT.kGreen+2, 'multijet M = 800 GeV, #tau = 300 #mum'),
(name2isample(f, 'mfv_neu_tau01000um_M0800'), ROOT.kBlue, 'multijet M = 800 GeV, #tau = 1 mm'),
(name2isample(f, 'mfv_neu_tau10000um_M0800'), ROOT.kMagenta, 'multijet M = 800 GeV, #tau = 10 mm'),
(name2isample(f, 'mfv_neu_tau30000um_M0800'), ROOT.kOrange+2, 'multijet M = 800 GeV, #tau = 30 mm'),
]),
('dijet', [
(name2isample(f, 'mfv_ddbar_tau00100um_M0800'), ROOT.kRed, 'dijet M = 800 GeV, #tau = 100 #mum'),
(name2isample(f, 'mfv_ddbar_tau00300um_M0800'), ROOT.kGreen+2, 'dijet M = 800 GeV, #tau = 300 #mum'),
(name2isample(f, 'mfv_ddbar_tau01000um_M0800'), ROOT.kBlue, 'dijet M = 800 GeV, #tau = 1 mm'),
(name2isample(f, 'mfv_ddbar_tau10000um_M0800'), ROOT.kMagenta, 'dijet M = 800 GeV, #tau = 10 mm'),
(name2isample(f, 'mfv_ddbar_tau30000um_M0800'), ROOT.kOrange+2, 'dijet M = 800 GeV, #tau = 30 mm'),
]),
]
def fmt(h, name, color, save=[]):
binning = to_array(0., 0.04, 0.07, 0.15)
h = h.Rebin(len(binning)-1, name, binning)
h.Sumw2()
h.SetStats(0)
h.SetLineWidth(3)
h.SetLineColor(color)
h.SetTitle(';d_{VV} (cm);event rate (unit norm.)')
move_overflow_into_last_bin(h)
h.Scale(1./h.Integral(0,h.GetNbinsX()+2))
save.append(h)
return h
for which_name, which in whiches:
hbkg = fmt(f.Get('h_bkg_dvv'), 'bkg', ROOT.kBlack)
hbkg.Draw('hist e')
hbkg.GetYaxis().SetRangeUser(0,1.5)
leg = ROOT.TLegend(0.142, 0.657, 0.702, 0.857)
leg.SetBorderSize(0)
leg.AddEntry(hbkg, 'Background template d_{VV}^{C}', 'LE')
for isample, color, title in which:
h = fmt(f.Get('h_signal_%i_dvv' % isample), title, color)
h.Draw('hist e same')
leg.AddEntry(h, title, 'LE')
leg.Draw()
ps.save(which_name, log=False)
def signal_efficiency():
from signal_efficiency import SignalEfficiencyCombiner
for which, years in ('run2', [2016, 2017, 2018]), ('2017p8', [2017, 2018]):
combiner = SignalEfficiencyCombiner(years)
in_f = combiner.inputs[0].f
out_f = ROOT.TFile('signal_efficiency_%s.root' % which, 'recreate')
kinds, masses, taus, nmasses, ntaus = axes(in_f)
taus.remove(30.)
ntaus -= 1
for kind in kinds:
h = ROOT.TH2D('signal_efficiency_%s' % kind,
';mass (GeV);#tau (mm)', nmasses, masses, ntaus,
taus)
for ibin in xrange(1, nmasses + 1):
mass = h.GetXaxis().GetBinLowEdge(ibin)
for jbin in xrange(1, ntaus + 1):
tau = h.GetYaxis().GetBinLowEdge(jbin)
pt = kind, tau, mass
isample, r = None, None
try:
isample = name2isample(in_f, details2name(*pt))
r = combiner.combine(isample)
except ValueError as exc:
if tau < 100:
print colors.warning(
'problem getting %r : isample %r exc %s' %
(pt, isample, exc))
continue # leave holes in the plot to know which samples are missing
h.SetBinContent(ibin, jbin, r.total_sig_eff)
h.SetBinError(ibin, jbin, 0) # JMTBAD
out_f.cd()
h.Write()
out_f.Write()
out_f.Close()
def fromtree(fn, _jobre=re.compile(r'_(\d+)\.root$')):
bn = os.path.basename(fn)
f = ROOT.TFile(fn)
ll = [x for x in detree(f.Get('limit'), 'limit:iToy', xform=(float, int))]
if bn.startswith('expected'):
mo = _jobre.search(bn)
if mo:
njobsexp = 1
ntoysexp = ntoysperjob
else:
njobsexp = njobs
ntoysexp = njobs * ntoysperjob
if len(
ll
) == 2 * ntoysexp: # for some reason, there are two entries for every iToy when --saveToys used
ll2 = []
for mi, (l, i) in enumerate(ll):
assert (mi / 2 % ntoysperjob) + 1 == i
if mi % 2 == 1:
ll2.append(l)
elif mi % 2 * ntoysperjob == 0: # and the limit value saved in the first entry for the first toy in each batch is ~1e-300!
assert l < 1e-300
elif mi > 0:
assert l == ll2[-1]
assert len(ll2) == ntoysexp
return ll2
else:
if len(ll) != ntoysexp:
raise ValueError('fromtree %s failed: len(ll) = %i != %i' %
(fn, len(ll), ntoysexp))
jobordercheck = [b for _, b in ll]
jobordershouldbe = range(1, ntoysperjob + 1) * njobsexp
if jobordercheck != jobordershouldbe:
raise ValueError('fromtree %s failed: jobordercheck %r != $r' %
(fn, jobordercheck, jobordershouldbe))
return [a for a, _ in ll]
else:
assert bn == 'observed.root' and len(ll) == 1 and ll[0][1] == 0
return [ll[0][0]]
def cutplayMangle(filename):
debug = False
file = ROOT.TFile.Open(filename)
if not file.IsOpen():
raise RuntimeError('could not open %s' % filename)
norm = norm_from_file(filename)
x = file.Get('SimpleTriggerEfficiency/triggers_pass_num')
cutscans = {}
nm1 = None
# Extract from the big list of bins the numbers separately for each
# cut.
for i in xrange(1, x.GetNbinsX() + 1):
label = x.GetXaxis().GetBinLabel(i)
if label == 'nm1':
nm1 = x.GetBinContent(i), x.GetBinError(i)
continue
try:
cut, cut_val = label.split('X')
except ValueError:
print 'warning: could not parse label %s' % label
continue
if not cutscans.has_key(cut):
cutscans[cut] = []
cutscans[cut].append((cut_val, x.GetBinContent(i)))
h_intot = file.Get('SimpleTriggerEfficiency/triggers_pass_den')
ntot = h_intot.GetBinContent(1), h_intot.GetBinError(1)
file.Close()
output_fn = filename.replace('.root', '_mangled.root')
output_file = ROOT.TFile(output_fn, 'RECREATE')
sample_name = os.path.basename(filename).replace('.root', '')
try:
sample = getattr(Samples, sample_name)
except AttributeError:
sample = None
h_ntot = ROOT.TH1F('ntot', '', 1, 0, 1)
scale = ntot[0] / norm if sample is not None else 1.
h_ntot.SetBinContent(1, ntot[0] / scale)
h_ntot.SetBinError(1, ntot[1] / scale)
if nm1 is not None:
h_nm1 = ROOT.TH1F('nm1', '', 1, 0, 1)
h_nm1.SetBinContent(1, nm1[0])
h_nm1.SetBinError(1, nm1[1])
else:
raise ValueError('did not find nm1 value in input')
output_hists = []
for cut_name, scan in cutscans.iteritems():
if debug:
print 'cut: %s' % cut_name
hmin = depize(scan[0][0])
hminplus1 = depize(scan[1][0])
binwidth = hminplus1 - hmin
hmaxminus1 = depize(scan[-1][0])
hmax = hmaxminus1 + binwidth
if debug:
print 'hmin: ', hmin, ' hminplus1: ', hminplus1, ' binwidth: ', binwidth, ' hmaxminus1: ', hmaxminus1, ' hmax: ', hmax
hist = ROOT.TH1F(cut_name, '', len(scan), hmin, hmax)
output_hists.append(hist)
for cut_bin_name, nevents in scan:
bin = hist.FindBin(depize(cut_bin_name) + 0.00001)
bin_edge = hist.GetBinLowEdge(bin)
if debug:
print 'cut bin name: %s ibin: %i bin_edge: %s number of events: %f' % (
cut_bin_name, bin, bin_edge, nevents)
hist.SetBinContent(bin, nevents)
output_file.Write()
output_file.Close()
import sys
from JMTucker.Tools.ROOTTools import ROOT
from binning import binning
def book(out_name, hs=[]):
bins = binning()
h = ROOT.TH1F(out_name, '', len(bins) - 1, bins)
hs.append(h)
return h
f = ROOT.TFile('export.root', 'recreate')
execfile(sys.argv[1])
f.Write()
f.Close()
mass = float(mass)
tau, tauunit = float(tau_s[:-2]), tau_s[-2:]
if tauunit == 'um':
tau *= 1e-4
elif tauunit == 'mm':
tau *= 1e-1
nevents_expected = 10000
if model == 'GluinoGluinoToNeutralinoNeutralinoTo2T2B2S':
nevents_missing_allowed = 0
folder_name = 'Lsps'
elif model == 'StopStopTo2Dbar2Dbar':
nevents_missing_allowed = 10
folder_name = 'Hs'
f = ROOT.TFile(fn)
hs = [f.Get('mfvGenHistos/%s#%i/M' % (folder_name, i)) for i in 0, 1]
ok = all(nevents_expected - h.GetEntries() <= nevents_missing_allowed
and abs(h.GetMean() - mass) < 0.01 for h in hs)
h = f.Get('mfvGenHistos/h_ctaubig')
xmax = h.GetXaxis().GetXmax()
mean_expected = (tau - exp(-xmax / tau) *
(xmax + tau)) / (1 - exp(-xmax / tau))
ok = ok and nevents_expected - h.GetEntries(
) / 2. <= nevents_missing_allowed and abs(
h.GetMean() - mean_expected) < 2 * h.GetMeanError()
print(colors.green if ok else colors.red)(
'%s (%s %s %s) %s (%s %s) %s (%s)' %
(fn, hs[0].GetEntries(), hs[1].GetEntries(), h.GetEntries(), mass,
h.Write()
out_f.Write()
out_f.Close()
if __name__ == '__main__':
if 'make' in sys.argv:
make()
elif 'uncert' in sys.argv:
sig_uncert_2017p8(sys.argv[sys.argv.index('uncert')+1], debug=True)
elif 'draw' in sys.argv:
draw()
elif 'compare' in sys.argv:
compare(*sys.argv[2:5])
elif 'ngen' in sys.argv:
f = ROOT.TFile(gp.fn)
years = sys.argv[sys.argv.index('ngen'):]
for s in sample_iterator(f, years, slices_1d=True):
print '%6i' % s.isample, s.name.ljust(35), ' '.join(['%10s']*gp.nyears) % tuple(ngen(f, s.isample, year) for year in gp.years)
#nevents_plot()
elif 'points' in sys.argv:
print 'kinds = %r\nmasses = %r\ntaus = %r' % points()
elif 'iter' in sys.argv:
include_2016 = 'include_2016' in sys.argv
years = ('2016','2017','2018') if include_2016 else ('2017','2018')
samples = sample_iterator(ROOT.TFile('limitsinput.root'),
require_years=years,
test='test_batch' in sys.argv,
slices_1d='slices_1d' in sys.argv,
)
names = set(s.name for s in samples)
#!/usr/bin/env python
import sys
from pprint import pprint
from JMTucker.Tools.ROOTTools import ROOT, flatten_directory, check_consistency
ROOT.TH1.AddDirectory(0)
fn1, fn2 = fns = [x for x in sys.argv[1:] if '.root' in x]
print 'comparing', fn1, fn2
f1, f2 = fs = [ROOT.TFile(fn) for fn in fns]
d1 = set(flatten_directory(f1))
d2 = set(flatten_directory(f2))
# could remap
#assert sorted(d1) == sorted(d2)
d = sorted(d1 & d2)
in1not2 = sorted(d1 - d2)
in2not1 = sorted(d2 - d1)
if 'small' in sys.argv:
print 'comparing %i hists (and not %i+%i of them)' % (len(d), len(in1not2),
len(in2not1))
elif in1not2 or in2not1:
print 'warning: not comparing these:'
print 'in f1 not f2:'
pprint(in1not2)
print 'in f2 not f1:'
pprint(in2not1)
print
out_f.Write()
out_f.Close()
if __name__ == '__main__':
if 'make' in sys.argv:
make()
elif 'uncert' in sys.argv:
sig_uncert_2017p8(sys.argv[sys.argv.index('uncert') + 1], debug=True)
elif 'draw' in sys.argv:
draw()
elif 'compare' in sys.argv:
compare(*sys.argv[2:5])
elif 'ngen' in sys.argv:
f = ROOT.TFile(gp.fn)
years = sys.argv[sys.argv.index('ngen'):]
for s in sample_iterator(f, years, slices_1d=True):
print '%6i' % s.isample, s.name.ljust(35), ' '.join(
['%10s'] * gp.nyears) % tuple(
ngen(f, s.isample, year) for year in gp.years)
#nevents_plot()
elif 'points' in sys.argv:
print 'kinds = %r\nmasses = %r\ntaus = %r' % points()
elif 'iter' in sys.argv:
include_2016 = 'include_2016' in sys.argv
years = ('2016', '2017', '2018') if include_2016 else ('2017', '2018')
samples = sample_iterator(
ROOT.TFile('limitsinput.root'),
require_years=years,
test='test_batch' in sys.argv,