def _setConfiguration(self, trigger, usetag):
self.baseDir = common.getBaseDirectory()
# config nbjet and selections
if trigger == "mu":
self.selections = ["emu", "mumu", "mutau", "mu4j"]
elif trigger == "e":
self.selections = ["ee", "emu2", "etau", "e4j"]
if usetag == "1b":
self.nbjet = "==1"
if usetag == "2b":
self.nbjet = ">1"
# read nGen from file
dfNGen = pd.read_pickle(self.baseDir + "data/pickles/ngen.pkl")
self.dfNGen = dfNGen
self.ttxs, self.txs = 832, 35.85 * 2
if self.variation == 'TTXSUp':
self.ttxs = self.ttxs * 1.05
if self.variation == 'TWXSUp':
self.txs = self.txs * 1.05
self.c_ttxs = self.ttxs / (self.ttxs + self.txs)
self.c_txs = self.txs / (self.ttxs + self.txs)
self.arrayToMatrix = np.array([[0, 2, 9, 10, 11, 15],
[2, 1, 12, 13, 14, 16],
[9, 12, 3, 5, 6, 17],
[10, 13, 5, 4, 7, 18],
[11, 14, 6, 7, 8, 19],
[15, 16, 17, 18, 19, 20]])
def _setConfiguration(self):
self.baseDir = common.getBaseDirectory()
if self.selection == "mumu":
if self.nbjet == '==1':
self.workingPoint = 0.07
if self.nbjet == '>1':
self.workingPoint = 0.07
if self.selection == "ee":
if self.nbjet == '==1':
self.workingPoint = 0.05
if self.nbjet == '>1':
self.workingPoint = 0.2
# read nGen from file
dfNGen = pd.read_pickle(self.baseDir + "data/pickles/ngen.pkl")
self.dfNGen = dfNGen
self.ttxs, self.txs = 832, 35.85 * 2
if self.variation == 'TTXSUp':
self.ttxs = self.ttxs * 1.05
if self.variation == 'TWXSUp':
self.txs = self.txs * 1.05
self.c_ttxs = self.ttxs / (self.ttxs + self.txs)
self.c_txs = self.txs / (self.ttxs + self.txs)
self.arrayToMatrix = np.array([[0, 2, 9, 10, 11, 15],
[2, 1, 12, 13, 14, 16],
[9, 12, 3, 5, 6, 17],
[10, 13, 5, 4, 7, 18],
[11, 14, 6, 7, 8, 19],
[15, 16, 17, 18, 19, 20]])
def templateVariation(region, shaping):
baseDir = common.getBaseDirectory()
# config
if shaping:
folderType = 'shaping'
else:
folderType = 'counting'
# nominal - dwon
dx_list = []
for variation in ['EPt', 'MuPt', 'TauPt']:
x0 = np.load(baseDir + "data/templates/{}_{}Region/X_{}.npy".format(
folderType, region, ''))
x1 = np.load(baseDir +
"data/templates/{}_{}Region/X_{}Down.npy".format(
folderType, region, variation))
dx = tc.from_numpy((x0 - x1)).type(tc.FloatTensor).to(device)
dx_list.append(dx)
# (up-down)/2
for variation in ['JES', 'JER', 'BTag', 'Mistag']:
x1 = np.load(baseDir +
"data/templates/{}_{}Region/X_{}Down.npy".format(
folderType, region, variation))
x2 = np.load(baseDir + "data/templates/{}_{}Region/X_{}Up.npy".format(
folderType, region, variation))
dx = tc.from_numpy((x2 - x1) / 2).type(tc.FloatTensor).to(device)
dx_list.append(dx)
return dx_list
def __init__(self):
self.tb = BFSolver_Toolbox()
self.baseDir = common.getBaseDirectory()
counts = pd.read_pickle(self.baseDir + "data/counts/count_.pkl")
self.a, self.aVar = counts.acc, counts.accVar
self.ndata, self.ndataVar = counts.ndata, counts.ndataVar
self.nmcbg, self.nmcbgVar = counts.nmcbg, counts.nmcbgVar
self.nfake, self.nfakeVar = counts.nfake, counts.nfakeVar
def __init__(self, selection, nbjet):
self.selection = selection
self.nbjet = nbjet
self.var_list = common.featureList()
self.nvar = len(self.var_list)
if nbjet == '==1':
self.bname = '1b'
if nbjet == '>1':
self.bname = '2b'
if self.selection[-1] == 'p':
self.selection = self.selection[0:-1]
self.norm = np.load(common.getBaseDirectory() +
"data/networks/{}{}_norm.npy".format(
self.selection, self.bname)).item()
self.net = torch.load(
common.getBaseDirectory() +
"data/networks/{}{}.pt".format(self.selection, self.bname))
def __init__(self, inputRootFileName, selection):
self.baseDir = common.getBaseDirectory()
self.inputRootFile = root.TFile(self.baseDir + "data/root/" +
inputRootFileName)
self.selection = selection
self.lumin = 35.864
self._getCrossection()
self._getNameList()
def loadData(self):
MCzz = DFCutter(self.selection, self.nbjet, "mcdiboson").getDataFrame()
MCdy = DFCutter(self.selection, self.nbjet, "mcdy").getDataFrame()
MCt = DFCutter(self.selection, self.nbjet, "mct").getDataFrame()
MCtt = DFCutter(self.selection, self.nbjet, "mctt").getDataFrame()
MCsg = pd.concat([MCt, MCtt], ignore_index=True)
if self.selection == 'mumu':
MCsg0 = MCsg.query('(genCategory != 14) & (genCategory != 5)')
MCsg1 = MCsg.query('(genCategory == 14) | (genCategory == 5)')
if self.selection == 'ee':
MCsg0 = MCsg.query('(genCategory != 10) & (genCategory != 4)')
MCsg1 = MCsg.query('(genCategory == 10) | (genCategory == 4)')
MClist = [MCzz, MCdy, MCsg0, MCsg1]
self.df_list, self.N_list, self.NRaw_list = [], [], []
for i in range(len(MClist)):
#MClist[i] = MClist[i].reset_index(drop=True)
n = int(np.sum(MClist[i].eventWeight))
nRaw = int(np.sum(MClist[i].eventWeight / MClist[i].eventWeightSF))
df = MClist[i].sample(int(n), replace=False)
self.N_list.append(n)
self.NRaw_list.append(nRaw)
self.df_list.append(df)
MCbkg = pd.concat(self.df_list[0:3], ignore_index=True)
MCsig = self.df_list[3]
drop_list = [v for v in MCsig.columns if not v in self.var_list]
MCbkg.drop(drop_list, axis=1, inplace=True)
MCsig.drop(drop_list, axis=1, inplace=True)
MCbkg['label'] = 0
MCsig['label'] = 1
df_train = pd.concat([MCbkg, MCsig], ignore_index=True)
df_train = df_train.reset_index(drop=True)
self.norm = {}
for v in df_train.columns:
if v != 'label':
mu = df_train[v].mean()
sigma = df_train[v].std()
df_train[v] = (df_train[v] - mu) / sigma
self.norm[v] = (mu, sigma)
np.save(
common.getBaseDirectory() +
"data/networks/{}{}_norm.npy".format(self.selection, self.bname),
self.norm)
self.df_train = df_train
def configTemplateVariation(self, region, shaping=False):
baseDir = common.getBaseDirectory()
if shaping:
folderType = 'shaping'
else:
folderType = 'counting'
# nominal - dwon
dx_list = []
for variation in ['EPt', 'MuPt', 'TauPt']:
x0 = np.load(baseDir +
"data/templates/{}_{}Region/X_{}.npy".format(
folderType, region, ''))
x1 = np.load(baseDir +
"data/templates/{}_{}Region/X_{}Down.npy".format(
folderType, region, variation))
dx = (x0 - x1)
dx_list.append(dx)
self.dx_energye = dx_list[0]
self.dx_energym = dx_list[1]
self.dx_energyt = dx_list[2]
# (up-down)/2
dx_list = []
for variation in ['JES', 'JER', 'BTag', 'Mistag']:
x1 = np.load(baseDir +
"data/templates/{}_{}Region/X_{}Down.npy".format(
folderType, region, variation))
x2 = np.load(baseDir +
"data/templates/{}_{}Region/X_{}Up.npy".format(
folderType, region, variation))
dx = (x2 - x1) / 2
dx_list.append(dx)
self.dx_jes = dx_list[0]
self.dx_jer = dx_list[1]
self.dx_btag = dx_list[2]
self.dx_mistag = dx_list[3]
def __init__(self, selection, nbjet, name, njet=None, TightTau=False):
'''
initialize a DFCutter with selection and name
For Example, cutter = DFCutter("mumu","mctt")
'''
self.selection = selection
self.name = name
self.baseDir = common.getBaseDirectory()
folderOfSelection = self.selection
if self.selection == "emu2":
folderOfSelection = 'emu'
if self.selection == "mutau_fakes":
folderOfSelection = 'mutau'
if self.selection == "etau_fakes":
folderOfSelection = 'etau'
if self.selection == 'mumuc':
folderOfSelection = 'mumu'
if self.selection == 'eec':
folderOfSelection = 'ee'
self.pickleDirectry = self.baseDir + "data/pickles/{}/".format(
folderOfSelection)
if TightTau:
self.pickleDirectry = self.baseDir + "data/pickles_tightTau/{}/".format(
folderOfSelection)
# default jet requirments
self.nb = " & (nBJets{})".format(nbjet)
if njet is None:
if '4j' in self.selection:
self.nj = " & (nJets >= 4)"
else:
self.nj = " & (nJets >= 2)"
else:
self.nj = " & (nJets {})".format(njet)
def __init__(self,beta0, controlRegion=None):
self.beta0 = beta0
# read sigma of beta0
baseDir = common.getBaseDirectory()
sig = np.load(baseDir + 'data/combine/sigma.npy')
var = np.load(baseDir + 'data/combine/covar.npy')
self.sig_syst = sig[1:]
self.var_stat = var[0]
self.ivar_stat = np.linalg.pinv(self.var_stat)
# some configuration
self.n = var.shape[0]-1
self.param0 = np.r_[0.1081*np.ones(3), np.zeros(self.n)]
# config control region
self.controlRegion = controlRegion
if not controlRegion is None:
self.model1 = controlRegion[0]
self.X1 = controlRegion[1]
self.Y1 = controlRegion[2]
self.lsEstimator()
def summaryCounts():
counts = {}
cols = [
'acc', 'accVar', 'nmcbg', 'nmcbgVar', 'nfake', 'nfakeVar', 'ndata',
'ndataVar'
]
# nominal count
df0 = pd.read_pickle(common.getBaseDirectory() + "data/counts/count_.pkl")
count0 = []
for c in cols:
count0.append(np.r_[df0[c][0], df0[c][1], df0[c][2], df0[c][3]])
counts['nominal'] = count0
# variate up
for variation in ['TTXS', 'TWXS']:
dfup = pd.read_pickle(common.getBaseDirectory() +
"data/counts/count_{}Up.pkl".format(variation))
count = []
for c in cols:
tempup = np.r_[dfup[c][0], dfup[c][1], dfup[c][2], dfup[c][3]]
temp0 = np.r_[df0[c][0], df0[c][1], df0[c][2], df0[c][3]]
count.append(tempup - temp0)
counts[variation] = count
# variate down
for variation in ['EPt', 'MuPt', 'TauPt']:
dfdw = pd.read_pickle(common.getBaseDirectory() +
"data/counts/count_{}Down.pkl".format(variation))
count = []
for c in cols:
tempdw = np.r_[dfdw[c][0], dfdw[c][1], dfdw[c][2], dfdw[c][3]]
temp0 = np.r_[df0[c][0], df0[c][1], df0[c][2], df0[c][3]]
count.append(temp0 - tempdw)
counts[variation] = count
# variate up-down
for variation in ['JES', 'JER', 'BTag', "Mistag"]:
dfdw = pd.read_pickle(common.getBaseDirectory() +
"data/counts/count_{}Down.pkl".format(variation))
dfup = pd.read_pickle(common.getBaseDirectory() +
"data/counts/count_{}Up.pkl".format(variation))
count = []
for c in cols:
tempdw = np.r_[dfdw[c][0], dfdw[c][1], dfdw[c][2], dfdw[c][3]]
tempup = np.r_[dfup[c][0], dfup[c][1], dfup[c][2], dfup[c][3]]
count.append((tempup - tempdw) / 2)
counts[variation] = count
# lepton eff
for variation in ['EffE', 'EffMu', 'EffTau']:
if variation == 'EffE':
effup = 1.01
effup = np.array([effup, 1, 1, 1] * 2 +
[effup**2, effup, effup, effup] * 2)
if variation == 'EffMu':
effup = 1.01
effup = np.array([effup, effup**2, effup, effup] * 2 +
[1, effup, 1, 1] * 2)
if variation == 'EffTau':
effup = 1.05
effup = np.array([1, 1, effup, 1] * 2 + [1, 1, effup, 1] * 2)
count = []
for c in cols:
temp0 = np.r_[df0[c][0], df0[c][1], df0[c][2], df0[c][3]]
if c == 'acc':
tempup = np.r_[df0[c][0], df0[c][1], df0[c][2],
df0[c][3]] * effup.reshape(16, 1, 1)
elif c == 'nmcbg':
tempup = np.r_[df0[c][0], df0[c][1], df0[c][2],
df0[c][3]] * effup.reshape(16, 1)
else:
tempup = np.r_[df0[c][0], df0[c][1], df0[c][2], df0[c][3]]
count.append(tempup - temp0)
counts[variation] = count
# background xs
for variation in ['VVXS', 'ZXS', 'WXS']:
if variation == 'VVXS':
effup = np.array([[1.1, 1, 1]])
if variation == 'EffMu':
effup = np.array([[1, 1.05, 1]])
if variation == 'EffTau':
effup = np.array([[1, 1, 1.05]])
count = []
for c in cols:
temp0 = np.r_[df0[c][0], df0[c][1], df0[c][2], df0[c][3]]
if c == 'nmcbg':
tempup = np.r_[df0[c][0], df0[c][1], df0[c][2],
df0[c][3]] * effup
else:
tempup = np.r_[df0[c][0], df0[c][1], df0[c][2], df0[c][3]]
count.append(tempup - temp0)
counts[variation] = count
# fake rate
for variation in ['fakee', 'fakemu', 'faketau']:
if variation == 'fakee':
effup = 1.25
effup = np.array([1, 1, 1, 1] * 2 + [1, 1, 1, effup] * 2)
if variation == 'fakemu':
effup = 1.25
effup = np.array([1, 1, 1, effup] * 2 + [1, 1, 1, 1] * 2)
if variation == 'faketau':
effup = 1.25
effup = np.array([1, 1, effup, 1] * 4)
count = []
for c in cols:
temp0 = np.r_[df0[c][0], df0[c][1], df0[c][2], df0[c][3]]
if c == 'nfake':
tempup = np.r_[df0[c][0], df0[c][1], df0[c][2],
df0[c][3]] * effup
else:
tempup = np.r_[df0[c][0], df0[c][1], df0[c][2], df0[c][3]]
count.append(tempup - temp0)
counts[variation] = count
counts = pd.DataFrame.from_dict(counts, orient='index', columns=cols)
return counts
def _setConfiguration(self):
self.outputPlotDir = None
baseDirectory = common.getBaseDirectory()
# MARK -- config output plot directory
if self.nbjet == "==1":
self.outputPlotDir = baseDirectory + "plots/kinematics/{}/1b/".format(
self.selection)
elif self.nbjet == ">1":
self.outputPlotDir = baseDirectory + "plots/kinematics/{}/2b/".format(
self.selection)
elif self.nbjet == "<1":
self.outputPlotDir = baseDirectory + "plots/kinematics/{}/0b/".format(
self.selection)
# MARK -- config plotting parameters for each selection
# mumu
if self.selection in ["mumu", "mumuc"]:
self.mcsgQueryList = [
'genCategory>=16', 'genCategory in [1,3,4,5,6,7,8,9,10,11,12]',
'genCategory in [2]', 'genCategory in [13,14,15]'
]
self.labelList = [
'Diboson', 'Z+Jets', 'W+Jets', r'$tt/tW \rightarrow$ other',
r'$tt/tW \rightarrow ll$ other',
r'$tt/tW \rightarrow \mu + \mu$',
r'$tt/tW \rightarrow \mu+ \tau$', 'data'
]
self.colorList = [
"#a32020", "#e0301e", "#eb8c00", "gold", "#49feec",
"deepskyblue", "mediumpurple", "k"
]
self.pp = pd.read_csv(
baseDirectory + "scripts/plotterItemTables/itemTable_mumu.csv")
self.adjust = [1, 1, 1, 1, 1, 1, 1]
#self.hasFake = False
# ee
elif self.selection in ["ee", "eec"]:
self.mcsgQueryList = [
'genCategory>=16', 'genCategory in [2,3,4,5,6,7,8,9,13,14,15]',
'genCategory in [1]', 'genCategory in [10,11,12]'
]
self.labelList = [
'Diboson', 'Z+Jets', 'W+Jets', r'$tt/tW \rightarrow$ other',
r'$tt/tW \rightarrow ll$ other', r'$tt/tW \rightarrow e + e$',
r'$tt/tW \rightarrow e + \tau$', 'data'
]
self.colorList = [
"#a32020", "#e0301e", "#eb8c00", "gold", "#49feec",
"deepskyblue", "mediumpurple", "k"
]
self.pp = pd.read_csv(baseDirectory +
"scripts/plotterItemTables/itemTable_ee.csv")
self.adjust = [1, 1, 1, 1, 1, 1, 1]
#self.hasFake = False
# mue and emu
elif self.selection in ["emu", "emu2"]:
self.mcsgQueryList = [
'genCategory>=16', 'genCategory in [1,2,4,5,6,7,8,9]',
'genCategory in [3]', 'genCategory in [10,11,12]',
'genCategory in [13,14,15]'
]
self.labelList = [
'Diboson', 'Z+Jets', 'W+Jets', r'$tt/tW \rightarrow$ (other)',
r'$tt/tW \rightarrow l + l$ (other)',
r'$tt/tW \rightarrow e + \mu$',
r'$tt/tW \rightarrow e + \tau$',
r'$tt/tW \rightarrow \mu + \tau$', 'data'
]
self.colorList = [
"#a32020", "#e0301e", "#eb8c00", "gold", "springgreen",
"#49feec", "deepskyblue", "mediumpurple", "k"
]
self.pp = pd.read_csv(
baseDirectory + "scripts/plotterItemTables/itemTable_emu.csv")
self.adjust = [1, 1, 1, 1, 1, 1, 1, 1]
#self.hasFake = False
# mutau
elif "mutau" in self.selection:
self.mcsgQueryList = [
'genCategory in [16,21]',
'genCategory in [1,2,3, 4,5,6,7,8,9, 10,11,12]',
'genCategory in [17]', 'genCategory in [18,19,20]',
'genCategory in [13,14,15]'
]
self.labelList = [
'Diboson', 'Z+Jets', 'W+Jets', r'$tt/tW \rightarrow$ other',
r'$tt/tW \rightarrow l + l$ (other) ',
r'$tt/tW \rightarrow \mu + h$',
r'$tt/tW \rightarrow \tau + h$',
r'$tt/tW \rightarrow \mu + \tau$', 'data'
]
self.colorList = [
"#a32020", "#e0301e", "#eb8c00", "gold", "springgreen",
"#49feec", "deepskyblue", "mediumpurple", "k"
]
self.pp = pd.read_csv(
baseDirectory +
"scripts/plotterItemTables/itemTable_mutau.csv")
self.adjust = [1, 1, 1, 1, 1, 1, 1, 1, 1]
#self.adjust = [1/.95,1/.95,1/.95,1/.95,1/.95,1/.95,.89/.95]
#self.hasFake = False
if self.selection == 'mutau':
self.fakeSF = common.getFakeSF('tau')
self.colorList = ['grey'] + self.colorList
self.adjust = [self.fakeSF] + self.adjust
self.labelList = ['Fakes'] + self.labelList
# etau
elif "etau" in self.selection:
self.mcsgQueryList = [
'genCategory in [17,21]',
'genCategory in [1,2,3, 4,5,6,7,8,9, 13,14,15]',
'genCategory in [16]', 'genCategory in [18,19,20]',
'genCategory in [10,11,12]'
]
self.labelList = [
'Diboson', 'Z+Jets', 'W+Jets', r'$tt/tW \rightarrow$ other',
r'$tt/tW \rightarrow l + l$ (other) ',
r'$tt/tW \rightarrow e + h$', r'$tt/tW \rightarrow \tau + h$',
r'$tt/tW \rightarrow e + \tau$', 'data'
]
self.colorList = [
"#a32020", "#e0301e", "#eb8c00", "gold", "springgreen",
"#49feec", "deepskyblue", "mediumpurple", "k"
]
self.pp = pd.read_csv(
baseDirectory + "scripts/plotterItemTables/itemTable_etau.csv")
self.adjust = [1, 1, 1, 1, 1, 1, 1, 1, 1]
#self.adjust = [1/.95,1/.95,1/.95,1/.95,1/.95,1/.95,.89/.95]
#self.hasFake = False
if self.selection == 'etau':
self.fakeSF = common.getFakeSF('tau')
self.colorList = ['grey'] + self.colorList
self.adjust = [self.fakeSF] + self.adjust
self.labelList = ['Fakes'] + self.labelList
# mu4j
elif "mu4j" in self.selection:
self.mcsgQueryList = [
'genCategory in [16,18,19,20,21]',
'genCategory in [1,2,3,4,5,6,7,8,9,10,11,12]',
'genCategory in [17]', 'genCategory in [13,14,15]'
]
self.labelList = [
'Diboson', 'Z+Jets', 'W+Jets', r'$tt/tW \rightarrow$ other',
r'$tt/tW \rightarrow ll$ other',
r'$tt/tW \rightarrow \mu + h$',
r'$tt/tW \rightarrow \mu + \tau$', 'data'
]
self.pp = pd.read_csv(
baseDirectory + "scripts/plotterItemTables/itemTable_mu4j.csv")
self.colorList = [
"#a32020", "#e0301e", "#eb8c00", "gold", "#49feec",
"deepskyblue", "mediumpurple", "k"
]
self.adjust = [1, 1, 1, 1, 1, 1, 1]
if self.selection == 'mu4j':
self.fakeSF = common.getFakeSF('mu')
self.colorList = ['grey'] + self.colorList
self.adjust = [self.fakeSF] + self.adjust
self.labelList = ['Fakes'] + self.labelList
# e4j
elif "e4j" in self.selection:
self.mcsgQueryList = [
'genCategory in [17,18,19,20,21]',
'genCategory in [1,2,3,4,5,6,7,8,9,13,14,15]',
'genCategory in [16]', 'genCategory in [10,11,12]'
]
self.labelList = [
'Diboson', 'Z+Jets', 'W+Jets', r'$tt/tW \rightarrow$ other',
r'$tt/tW \rightarrow ll$ other', r'$tt/tW \rightarrow e + h$',
r'$tt/tW \rightarrow e + \tau$', 'data'
]
self.pp = pd.read_csv(
baseDirectory + "scripts/plotterItemTables/itemTable_e4j.csv")
self.colorList = [
"#a32020", "#e0301e", "#eb8c00", "gold", "#49feec",
"deepskyblue", "mediumpurple", "k"
]
self.adjust = [1, 1, 1, 1, 1, 1, 1]
if self.selection == 'e4j':
self.fakeSF = common.getFakeSF('e')
self.colorList = ['grey'] + self.colorList
self.adjust = [self.fakeSF] + self.adjust
self.labelList = ['Fakes'] + self.labelList