def test_unsupported_branch_in_branches():
tree = TTree('test', 'test')
vect = TLorentzVector()
double = np.array([0], dtype=float)
tree.Branch('vector', vect)
tree.Branch('double', double, 'double/D')
rnp.tree2array(tree)
assert_raises(TypeError, rnp.tree2array, tree, branches=['vector'])
def test_tree2array():
chain = TChain('tree')
chain.Add(load('single1.root'))
check_single(rnp.tree2array(chain))
f = get_file('single1.root')
tree = f.Get('tree')
check_single(rnp.tree2array(tree))
assert_raises(ValueError, get_file, 'file_does_not_exist.root')
def read_data(input_file_path, tree_name, event=1): input_file = ROOT.TFile(input_file_path, 'read') input_tree = input_file.Get(tree_name) data = tree2array(input_tree) input_file.Close() event_data = data[event-1] # choose event to display return event_data
def test_single_branch():
f = get_file('single1.root')
tree = f.Get('tree')
arr1_1d = rnp.tree2array(tree, branches='n_int')
arr2_1d = rnp.root2array(load('single1.root'), branches='n_int')
assert_equal(arr1_1d.dtype, np.dtype('<i4'))
assert_equal(arr2_1d.dtype, np.dtype('<i4'))
def to_array(self, *args, **kwargs):
"""
Convert this tree into a NumPy structured array
"""
from root_numpy import tree2array
return tree2array(self, *args, **kwargs)
def do_cut(args, did, files, supercuts, weights):
start = clock()
try:
# load up the tree for the files
tree = get_ttree(args.tree_name, files, args.eventWeightBranch)
# if using numpy optimization, load the tree as a numpy array to apply_cuts on
if args.numpy:
# this part is tricky, a user might specify multiple branches
# in their selection string, so we will remove non-alphanumeric characters (underscores are safe)
# and remove anything else that is an empty string (hence the filter)
# and then flatten the entire list, removing duplicate branch names
'''
totalSelections = []
for supercut in supercuts:
selection = supercut['selections']
# filter out non-alphanumeric
selection = p.sub(' ', selection.format("-", "-", "-", "-", "-", "-", "-", "-", "-", "-"))
# split on spaces, since we substituted non alphanumeric with spaces
selections = selection.split(' ')
# remove empty elements
filter(None, selections)
totalSelections.append(selections)
# flatten the thing
totalSelections = itertools.chain.from_iterable(totalSelections)
# remove duplicates
totalSelections = list(set(totalSelections))
'''
alphachars = re.compile('\W+')
branchesSpecified = list(set(itertools.chain.from_iterable(filter(None, alphachars.sub(' ', supercut['selections'].format(*['-']*10)).split(' ')) for supercut in supercuts)))
# get actual list of branches in the file
availableBranches = [i.GetName() for i in tree.GetListOfBranches() if not i.GetName() == args.eventWeightBranch]
# remove anything that doesn't exist
branchesToUse = [branch for branch in branchesSpecified if branch in availableBranches]
branchesSkipped = list(set(branchesSpecified) - set(branchesToUse))
if branchesSkipped:
logger.info("The following branches have been skipped...")
for branch in branchesSkipped:
logger.info("\t{0:s}".format(branch))
tree = rnp.tree2array(tree, branches=[args.eventWeightBranch]+branchesToUse)
# get the scale factor
sample_scaleFactor = get_scaleFactor(weights, did)
# iterate over the cuts available
cuts = {}
for cut in get_cut(copy.deepcopy(supercuts)):
cut_hash = get_cut_hash(cut)
rawEvents, weightedEvents = apply_cuts(tree, cut, args.eventWeightBranch, args.numpy)
scaledEvents = weightedEvents*sample_scaleFactor
cuts[cut_hash] = {'raw': rawEvents, 'weighted': weightedEvents, 'scaled': scaledEvents}
logger.info("Applied {0:d} cuts".format(len(cuts)))
with open('{0:s}/{1:s}.json'.format(args.output_directory, did), 'w+') as f:
f.write(json.dumps(cuts, sort_keys=True, indent=4))
result = True
except:
logger.exception("Caught an error - skipping {0:s}".format(did))
result = False
end = clock()
return (result, end-start)
def i3root2hdf5(infile, force=False):
h5file = infile + '.h5'
bad_keys = ['AntMCTree', 'MasterTree']
rf = root_open(infile, 'r')
keys = [k.name for k in rf.keys()]
if force:
mode = 'w'
else:
mode = 'a'
h5 = h5py.File(h5file, mode)
for key in keys:
if key in bad_keys:
continue
tree = rf[key]
arr = tree2array(tree)
try:
h5.create_dataset(
key,
data=arr,
compression='gzip',
compression_opts=9,
shuffle=True,
fletcher32=True,
)
except TypeError:
continue
h5.flush()
h5.close()
def tree_to_ndarray(trees, branches=None, dtype=np.float32, include_weight=False, weight_dtype="f4"):
"""
Convert a tree or a list of trees into a numpy.ndarray
"""
if isinstance(trees, (list, tuple)):
return np.concatenate(
[
_add_weight_column(
recarray_to_ndarray(tree2array(tree, branches), dtype=dtype), tree, include_weight, weight_dtype
)
for tree in trees
]
)
return _add_weight_column(
recarray_to_ndarray(tree2array(trees, branches), dtype=dtype), trees, include_weight, weight_dtype
)
def main ():
inputdir = '/eos/atlas/user/a/asogaard/Analysis/2016/BoostedJetISR/StatsInputs/2017-06-28/'
outputdir = '/eos/atlas/user/a/asogaard/Analysis/2016/BoostedJetISR/StatsInputs/2017-07-10/'
inputpaths = glob.glob(inputdir + '/ISRgamma_*.root')
outputpaths = [p.replace(inputdir, outputdir).replace('ISRgamma', 'hist_ISRgamma') for p in inputpaths]
for inputpath, outputpath in zip(inputpaths,outputpaths):
print "Processing '%s'" % inputpath
infile = ROOT.TFile(inputpath, 'READ')
outfile = ROOT.TFile(outputpath, 'RECREATE')
categories = [key.GetName() for key in infile.GetListOfKeys()]
for category in categories:
print "-- '%s'" % category
tree = infile.Get(category)
array = tree2array(tree)
#hist = ROOT.TH1F(category, "", 30, 100, 250)
hist = ROOT.TH1F(category, "", 32, 100, 260)
fill_hist(hist, array['mJ'], weights=array['weight'])
# TF shape/norm ...
outfile.cd()
hist.Write()
pass
outfile.Write()
outfile.Close()
infile.Close()
pass
return
def test_chain():
chain = ROOT.TChain('tree')
chain.Add(load('single1.root'))
check_single(rnp.tree2array(chain))
f = load(['single1.root', 'single2.root'])
a = rnp.root2array(f)
check_single(a, 200)
def test_selection():
chain = ROOT.TChain('tree')
chain.Add(load('single1.root'))
chain.Add(load('single2.root'))
a = rnp.tree2array(chain)
assert_equal((a['d_double'] <= 100).any(), True)
a = rnp.tree2array(chain, selection="d_double > 100")
assert_equal((a['d_double'] <= 100).any(), False)
# selection with differing variables in branches and expression
a = rnp.tree2array(chain,
branches=['d_double'],
selection="f_float < 100 && n_int%2 == 1")
# selection with TMath
a = rnp.tree2array(chain,
selection="TMath::Erf(d_double) < 0.5")
def getTreeToArray(tree):
"""
Convert and return a tree into a numpy array
Inputs: TTree object
Return: np array
"""
from root_numpy import tree2array
return tree2array(tree)
def records(self, **kwargs):
""
""
from root_numpy import tree2array
rfile = get_file(self.ntuple_path, self.student)
tree = rfile[self.tree_name]
log.info('Converting tree to record array, sorry if this is long ...')
rec = tree2array(tree, **kwargs).view(np.recarray)
return rec
def test_weights():
f = ROOT.TFile(load('test.root'))
tree = f.Get('tree')
tree.SetWeight(5.)
rec = rnp.tree2array(tree, include_weight=True, weight_name='treeweight')
assert_array_equal(rec['treeweight'], np.ones(100) * 5)
f = load(['single1.root', 'single2.root'])
a = rnp.root2array(f, include_weight=True)
assert_array_equal(
a['weight'],
np.concatenate((np.ones(100) * 2., np.ones(100) * 3.)))
def test_array2tree():
a = np.array([
(12345, 2., 2.1, True),
(3, 4., 4.2, False),],
dtype=[
('x', np.int32),
('y', np.float32),
('z', np.float64),
('w', np.bool)])
with temp() as tmp:
tree = rnp.array2tree(a)
a_conv = rnp.tree2array(tree)
assert_array_equal(a, a_conv)
# extend the tree
tree2 = rnp.array2tree(a, tree=tree)
assert_equal(tree2.GetEntries(), len(a) * 2)
a_conv2 = rnp.tree2array(tree2)
assert_array_equal(np.hstack([a, a]), a_conv2)
assert_raises(TypeError, rnp.array2tree, a, tree=object)
def to_array(self, branches=None,
include_weight=False,
weight_name='weight',
weight_dtype='f4'):
"""
Convert this tree into a NumPy structured array
"""
from root_numpy import tree2array
return tree2array(self, branches,
include_weight=include_weight,
weight_name=weight_name,
weight_dtype=weight_dtype)
def test_array2tree():
a = np.array([
(12345, 2., 2.1, True),
(3, 4., 4.2, False),],
dtype=[
('x', np.int32),
('y', np.float32),
('z', np.float64),
('w', np.bool)])
tmp = ROOT.TFile.Open('test_array2tree_temp_file.root', 'recreate')
tree = rnp.array2tree(a)
a_conv = rnp.tree2array(tree)
assert_array_equal(a, a_conv)
# extend the tree
tree2 = rnp.array2tree(a, tree=tree)
assert_equal(tree2.GetEntries(), len(a) * 2)
a_conv2 = rnp.tree2array(tree2)
assert_array_equal(np.hstack([a, a]), a_conv2)
tmp.Close()
os.remove(tmp.GetName())
assert_raises(TypeError, rnp.array2tree, a, tree=object)
def read_ntuple(path="./", cfg="", selection = 'loose'):
# read the config files the same as for the Zfitter
# to the a chain of the files
config = pd.read_csv(path + "/" + cfg , sep = " ", names = ['id', 'tree', 'file'], comment ="#")
print config
chain = r.TChain('merged')
for index, root in config.iterrows():
print root.id , "\t: ", root.file
chain.Add(root.file+'/'+root.tree)
# transform this chain to an array ment to be used later by matplotlib
data = tree2array( chain, selection = ecal_selections[selection])
return data
def test_single():
f = load('single1.root')
a = rnp.root2array(f)
check_single(a)
# specify tree name
a = rnp.root2array(f, treename='tree')
check_single(a)
# tree2array
f = get_file('single1.root')
tree = f.Get('tree')
check_single(rnp.tree2array(tree))
def KStest( PmissBins , ana_sim , ana_data , var , cut=ROOT.TCut() , debug=2 , Nbins=20):
# [http://docs.scipy.org/doc/scipy-0.15.1/reference/generated/scipy.stats.ks_2samp.html]
KS_distances , Pval_KS = [] , []
figure = plt.figure(figsize=[60,20])
for i in range(len(PmissBins)):
pMiss_min , pMiss_max = PmissBins[i][0] , PmissBins[i][1]
reduced_data = tree2array(ana_data.GetTree(),branches=var , selection = '%f < Pmiss3Mag && Pmiss3Mag < %f'%(pMiss_min , pMiss_max) )
reduced_sim = tree2array(ana_sim.GetTree(),branches=var , selection = '%f < Pmiss3Mag && Pmiss3Mag < %f'%(pMiss_min , pMiss_max))
D , Pvalue = ks_2samp( reduced_sim , reduced_data )
if ( debug > 1 ):
ax = figure.add_subplot(len(PmissBins)/2,3,i+1)
for array,col in zip([reduced_sim , reduced_data],['black','red']):
g=sns.distplot( array, bins=np.linspace(-1, 2 , Nbins), ax=ax, color=col , axlabel=var )
g.axes.set_title(r'%.2f < p$_{miss}$ < %.2f GeV/c'%(pMiss_min , pMiss_max), fontsize=34,color="b")
print_important( "KS test of data vs. simulation for %s in p(miss) bin %d is D = %f, Pvalue = %f"%(var , i , D , Pvalue) )
KS_distances.append(D)
Pval_KS.append(Pvalue)
figure.savefig("/Users/erezcohen/Desktop/cmHistos_%s.pdf"%var)
return KS_distances , Pval_KS
def numpyfy(run):
for f in files_to_convert:
fname = "%s/run%d/%s%d.root" % (root_dir, run,f,run)
print " Processing %s ..." % (fname )
rfile = ROOT.TFile(fname)
for key in rfile.GetListOfKeys():
if key.GetClassName() == "TTree":
name = key.GetName()
print " Converting Tree %s " % name
tree = rfile.Get(name)
#Ok, apparently root_numpy is not smart enough to convert composite objects,
# but we can give it a list of branches instead that we can automatically generate.
# Not sure why it doesn't do it itself...
branch_list = build_branches(tree)
size = tree.GetEntries()
if size == 0:
continue
nchunks = int(math.ceil(size / float(max_output_length)))
outdir = "%s/run%d" % (numpy_dir, run)
if not os.path.exists(outdir):
os.makedirs(outdir)
for chunk in range(nchunks):
arrays = {}
arrays[tree.GetName()] = tree2array(tree, branch_list, start = chunk * max_output_length, stop = (chunk+1) * max_output_length)
outfile = ""
if nchunks > 1:
outfile = "%s/%s%d_%d.npz" % (outdir, tree.GetName(), run,chunk)
else:
outfile = "%s/%s%d.npz" % (outdir, tree.GetName(), run)
print "Creating %s " % outfile
numpy.savez_compressed(outfile, **arrays)
else:
print " No trees found... skipping"
def test_duplicate_branch_name():
from array import array
tree = ROOT.TTree('tree', 'tree')
d = array('d', [0.])
tree.Branch('double', d, 'double/D')
tree.Branch('double', d, 'double/D')
tree.Fill()
# check that a warning was emitted
with warnings.catch_warnings(record=True) as w:
warnings.simplefilter("always")
a = rnp.tree2array(tree)
assert_equal(len(w), 1)
assert_true(issubclass(w[-1].category, RuntimeWarning))
assert_true("ignoring duplicate branch named" in str(w[-1].message))
assert_equal(
a.dtype,
[('double', '<f8')])
def evaluate(config, tree, names, transform=None):
output = []
dtype = []
for name in names:
setup = load(config, name.split("_")[1])
data = rec2array(tree2array(tree.raw(), list(transform(setup["variables"])) if transform else setup["variables"]))
if name.startswith("sklearn"):
fn = os.path.join(config["mvadir"], name + ".pkl")
with open(fn, 'rb') as fd:
bdt, label = pickle.load(fd)
scores = []
if len(data) > 0:
scores = bdt.predict_proba(data)[:, 1]
output += [scores]
dtype += [(name, 'float64')]
fn = os.path.join(config["mvadir"], name + ".xml")
reader = r.TMVA.Reader("Silent")
for var in setup['variables']:
reader.AddVariable(var, array('f', [0.]))
reader.BookMVA("BDT", fn)
scores = evaluate_reader(reader, "BDT", data)
output += [scores]
dtype += [(name.replace("sklearn", "tmvalike"), 'float64')]
f = r.TFile(os.path.join(config.get("mvadir", config.get("indir", config["outdir"])), "mapping.root"), "READ")
if f.IsOpen():
likelihood = f.Get("hTargetBinning")
def lh(values):
return likelihood.GetBinContent(likelihood.FindBin(*values))
indices = dict((v, n) for n, (v, _) in enumerate(dtype))
tt = output[indices['tmvalike_tt']]
ttZ = output[indices['tmvalike_ttZ']]
if len(tt) == 0:
output += [[]]
else:
output += [np.apply_along_axis(lh, 1, np.array([tt, ttZ]).T)]
dtype += [('tmvalike_likelihood', 'float64')]
f.Close()
data = np.array(zip(*output), dtype)
tree.mva(array2tree(data))
def _tree_to_array(schema, to_npy = False):
print 'Loading File...'
f = root_open(schema['sample']['file'])
T = f[schema['sample']['tree']]
if schema['sample'].has_key('selection') == False:
this_sel = None
else:
this_sel = schema['sample']['selection']
if schema['sample'].has_key('step'):
this_step = schema['sample']['step']
else:
this_step = None
print 'Pulling Tree...'
arr = rn.tree2array(T, selection = this_sel, step = this_step)
if to_npy is True:
print 'Writing to *.npy file...'
varlist = "".join(this_sel.split()).replace('(', '').replace(')', '').split('&&')
varlist.sort()
hash_name = os.path.basename(schema['sample']['file']) + schema['sample']['tree'] + ''.join(varlist) + str(this_step)
m = hashlib.sha1()
m.update(hash_name)
np.save(os.path.dirname(schema['sample']['file']) + '/' + m.hexdigest() + '.npy', arr)
print 'Done.'
return arr
import ROOT
import pandas as pd
import numpy as np
from root_numpy import tree2array
f = ROOT.TFile('evetest_CC4GeVmb_110_n50k.root')
tree = f.Get('cbmsim')
tracks_params = tree2array(tree, branches='BmnGemStripHit')
print "Accessed the trees" # get input variable names from branches vars = img.getBoostCandBranchNames(treeHH4W) treeVars = vars print "Variables for jet image creation: ", vars # create selection criteria #sel = "" sel = "jetAK8_pt > 500 && jetAK8_mass > 50" #sel = "tau32 < 9999. && et > 500. && et < 2500. && bDisc1 > -0.05 && SDmass < 400" # make arrays from the trees #start, stop, step = 0, 200000, 1 arrayHH4W = tree2array(treeHH4W, treeVars, sel)#, None, start, stop, step ) arrayHH4W = tools.appendTreeArray(arrayHH4W) print "Number of Jets that will be imaged: ", len(arrayHH4W) imgArrayHH4W = img.makeBoostCandFourVector(arrayHH4W) print "Made candidate 4 vector arrays from the datasets" #================================================================================== # Store BEST Variables //////////////////////////////////////////////////////////// #================================================================================== # get BEST variable names from branches bestVars = tools.getBestBranchNames(treeHH4W) print "Boosted Event Shape Variables: ", bestVars
print 'getting trees...', 'Making training arrays...', 'Making testing arrays...'
seltest = "Tprime2_DeepAK8_Mass >= 0"
#======process=====
#open file
#get tree
#Convert the ntuple branches to numpy arrays
for i in range(1, 7):
fileTTToSemiLepT = TFile.Open(
eosdir +
"TTJets_SingleLeptFromT_TuneCUETP8M1_13TeV-madgraphMLM-pythia8_" +
str(i) + "_hadd.root", "READ")
treeTTToSemiLepT = fileTTToSemiLepT.Get("ljmet")
if i == 1:
arrayTTToSemiLepT = tree2array(treeTTToSemiLepT, treeVars, sel)
testTTToSemiLepT = tree2array(treeTTToSemiLepT, treeVars, seltest)
else:
arrayTTToSemiLepT = np.concatenate(
[arrayTTToSemiLepT,
tree2array(treeTTToSemiLepT, treeVars, sel)])
testTTToSemiLepT = np.concatenate([
testTTToSemiLepT,
tree2array(treeTTToSemiLepT, treeVars, seltest)
])
for i in range(1, 3):
fileTTToSemiLepTb = TFile.Open(
eosdir +
"TTJets_SingleLeptFromTbar_TuneCUETP8M1_13TeV-madgraphMLM-pythia8_" +
str(i) + "_hadd.root", "READ")
]
print branchnames, len(branchnames)
jetbranch = ['jet_pt', 'jet_eta', 'jet_mass', 'jet_phi', 'jet_btag']
mu_branch = ['mu_pt', 'mu_eta', 'mu_mt', 'mu_phi', 'mu_q']
el_branch = ['el_pt', 'el_eta', 'el_mt', 'el_phi', 'el_q']
flat_branch = [
'm_l1j1', 'H_T', 'm_l1j2', 'm_l1l2', 'Nleps', 'H_Tratio', 'Nbtags',
'Nlooseb', 'Ntightb', 'H_Tb', 'Njets', 'MET', 'm_j1j2'
]
truthbranch = ['class']
data_dict = {}
Y = rootnp.tree2array(file_check, branches=truthbranch)
Z_Y = rootnp.rec2array(Y)
flat = rootnp.tree2array(file_check, branches=flat_branch)
Z_flat = rootnp.rec2array(flat)
#Z_Y = np.zeros(Y.shape[0])
#for a in range(0,Y.shape):
# Z_Y[a] = Z_Y[a].tolist()
X_mu = rootnp.tree2array(file_check, branches=mu_branch)
X_mu = rootnp.rec2array(X_mu)
X_el = rootnp.tree2array(file_check, branches=el_branch)
X_el = rootnp.rec2array(X_el)
X_jets = rootnp.tree2array(file_check, branches=jetbranch)
def anaUltraLatency(infilename,
debug=False,
latSigMaskRange=None,
latSigRange=None,
outputDir=None,
outfilename="latencyAna.root",
performFit=False):
"""
Analyzes data taken by ultraLatency.py
infilename - Name of input TFile containing the latTree TTree
debug - If True prints additional debugging statements
latSigMaskRange - Comma separated pair of values defining the region
to be masked when trying to fit the noise, e.g.
lat #notepsilon [40,44] is noise (lat < 40 || lat > 44)")
latSigRange - Comma separated pair of values defining expected
signal range, e.g. lat #epsilon [41,43] is signal")
outfilename - Name of output TFile containing analysis results
performFit - Fit the latency distributions
"""
# Determine output filepath
if outputDir is None:
from gempython.gemplotting.utils.anautilities import getElogPath
outputDir = getElogPath()
pass
# Redirect sys.stdout and sys.stderr if necessary
from gempython.gemplotting.utils.multiprocUtils import redirectStdOutAndErr
redirectStdOutAndErr("anaUltraLatency", outputDir)
# Create the output File and TTree
import ROOT as r
outF = r.TFile(outputDir + "/" + outfilename, "RECREATE")
if not outF.IsOpen():
outF.Close()
raise IOError(
"Unable to open output file {1} check to make sure you have write permissions under {0}"
.format(outputDir, outfilename))
if outF.IsZombie():
outF.Close()
raise IOError(
"Output file {1} is a Zombie, check to make sure you have write permissions under {0}"
.format(outputDir, outfilename))
myT = r.TTree('latFitTree', 'Tree Holding FitData')
# Attempt to open input TFile
inFile = r.TFile(infilename, "read")
if not inFile.IsOpen():
outF.Close()
inFile.Close()
raise IOError(
"Unable to open input file {0} check to make sure you have read permissions"
.format(infilename))
if inFile.IsZombie():
outF.Close()
inFile.Close()
raise IOError(
"Input file {0} is a Zombie, check to make sure you have write permissions and file has expected size"
.format(infilename))
from gempython.tools.hw_constants import vfatsPerGemVariant
# Get ChipID's
import numpy as np
import root_numpy as rp
##### FIXME
from gempython.gemplotting.mapping.chamberInfo import gemTypeMapping
if 'gemType' not in inFile.latTree.GetListOfBranches():
gemType = "ge11"
else:
gemType = gemTypeMapping[rp.tree2array(tree=inFile.latTree,
branches=['gemType'])[0][0]]
print gemType
##### END
from gempython.tools.hw_constants import vfatsPerGemVariant
nVFATS = vfatsPerGemVariant[gemType]
from gempython.gemplotting.mapping.chamberInfo import CHANNELS_PER_VFAT as maxChans
listOfBranches = inFile.latTree.GetListOfBranches()
if 'vfatID' in listOfBranches:
array_chipID = np.unique(
rp.tree2array(inFile.latTree, branches=['vfatID', 'vfatN']))
dict_chipID = {}
for entry in array_chipID:
dict_chipID[entry['vfatN']] = entry['vfatID']
else:
dict_chipID = {vfat: 0 for vfat in range(nVFATS)}
if debug:
print("VFAT Position to ChipID Mapping")
for vfat, vfatID in dict_chipID.iteritems():
print(vfat, vfatID)
# Set default histogram behavior
r.TH1.SetDefaultSumw2(False)
r.gROOT.SetBatch(True)
r.gStyle.SetOptStat(1111111)
#Initializing Histograms
print('Initializing Histograms')
from gempython.utils.gemlogger import printYellow
from gempython.utils.nesteddict import nesteddict as ndict
dict_hVFATHitsVsLat = ndict()
for vfat in range(0, nVFATS):
try:
chipID = dict_chipID[vfat]
except KeyError as err:
chipID = 0
if debug:
printYellow(
"No CHIP_ID for VFAT{0}, If you don't expect data from this VFAT there's no problem"
.format(vfat))
dict_hVFATHitsVsLat[vfat] = r.TH1F(
"vfat{0}HitsVsLat".format(vfat),
"VFAT {0}: chipID {1}".format(vfat, chipID), 1024, -0.5, 1023.5)
pass
#Filling Histograms
print('Filling Histograms')
latMin = 1000
latMax = -1
nTrig = -1
for event in inFile.latTree:
dict_hVFATHitsVsLat[int(event.vfatN)].Fill(event.latency, event.Nhits)
if event.latency < latMin and event.Nhits > 0:
latMin = event.latency
pass
elif event.latency > latMax:
latMax = event.latency
pass
if nTrig < 0:
nTrig = event.Nev
pass
pass
from math import sqrt
for vfat in range(0, nVFATS):
for binX in range(1, dict_hVFATHitsVsLat[vfat].GetNbinsX() + 1):
dict_hVFATHitsVsLat[vfat].SetBinError(
binX, sqrt(dict_hVFATHitsVsLat[vfat].GetBinContent(binX)))
hHitsVsLat_AllVFATs = dict_hVFATHitsVsLat[0].Clone("hHitsVsLat_AllVFATs")
hHitsVsLat_AllVFATs.SetTitle("Sum over all VFATs")
for vfat in range(1, nVFATS):
hHitsVsLat_AllVFATs.Add(dict_hVFATHitsVsLat[vfat])
# Set Latency Fitting Bounds - Signal
latFitMin_Sig = latMin
latFitMax_Sig = latMax
if latSigRange is not None:
listLatValues = map(lambda val: float(val), latSigRange.split(","))
if len(listLatValues) != 2:
raise IndexError(
"You must specify exactly two values for determining latency signal range; values given: {0} do not meet this criterion"
.format(listLatValues))
else:
latFitMin_Sig = min(listLatValues)
latFitMax_Sig = max(listLatValues)
# Set Latency Fitting Bounds - Noise
latFitMin_Noise = latFitMin_Sig - 1
latFitMax_Noise = latFitMax_Sig + 1
if latSigMaskRange is not None:
listLatValues = map(lambda val: float(val), latSigMaskRange.split(","))
if len(listLatValues) != 2:
raise IndexError(
"You must specify exactly two values for determining latency signal range; values given: {0} do not meet this criterion"
.format(listLatValues))
else:
latFitMin_Noise = min(listLatValues)
latFitMax_Noise = max(listLatValues)
# Make output TFile and TTree
from array import array
dirVFATPlots = outF.mkdir("VFAT_Plots")
if 'detName' in listOfBranches:
detName = r.vector('string')()
detName.push_back(
rp.tree2array(inFile.latTree, branches=['detName'])[0][0][0])
myT.Branch('detName', detName)
vfatN = array('i', [0])
myT.Branch('vfatN', vfatN, 'vfatN/I')
vfatID = array('L', [0])
myT.Branch('vfatID', vfatID, 'vfatID/i') #Hex Chip ID of VFAT
hitCountMaxLat = array('f', [0])
myT.Branch('hitCountMaxLat', hitCountMaxLat, 'hitCountMaxLat/F')
hitCountMaxLatErr = array('f', [0])
myT.Branch('hitCountMaxLatErr', hitCountMaxLatErr, 'hitCountMaxLatErr/F')
maxLatBin = array('f', [0])
myT.Branch('maxLatBin', maxLatBin, 'maxLatBin/F')
hitCountBkg = array('f', [0])
hitCountBkgErr = array('f', [0])
hitCountSig = array('f', [0])
hitCountSigErr = array('f', [0])
SigOverBkg = array('f', [0])
SigOverBkgErr = array('f', [0])
if performFit:
myT.Branch('hitCountBkg', hitCountBkg, 'hitCountBkg/F')
myT.Branch('hitCountBkgErr', hitCountBkgErr, 'hitCountBkgErr/F')
myT.Branch('hitCountSig', hitCountSig, 'hitCountSig/F')
myT.Branch('hitCountSigErr', hitCountSigErr, 'hitCountSigErr/F')
myT.Branch('SigOverBkg', SigOverBkg, 'SigOverBkg/F')
myT.Branch('SigOverBkgErr', SigOverBkgErr, 'SigOverBkgErr/F')
# Make output plots
from math import sqrt
dict_grNHitsVFAT = ndict()
dict_fitNHitsVFAT_Sig = ndict()
dict_fitNHitsVFAT_Noise = ndict()
grNMaxLatBinByVFAT = r.TGraphErrors(len(dict_hVFATHitsVsLat))
grMaxLatBinByVFAT = r.TGraphErrors(len(dict_hVFATHitsVsLat))
grVFATSigOverBkg = r.TGraphErrors(len(dict_hVFATHitsVsLat))
grVFATNSignalNoBkg = r.TGraphErrors(len(dict_hVFATHitsVsLat))
r.gStyle.SetOptStat(0)
if debug and performFit:
print("VFAT\tSignalHits\tSignal/Noise")
for vfat in dict_hVFATHitsVsLat:
#if we don't have any data for this VFAT, we just need to initialize the TGraphAsymmErrors since it is drawn later
if vfat not in dict_chipID:
dict_grNHitsVFAT[vfat] = r.TGraphAsymmErrors()
continue
# Store VFAT info
vfatN[0] = vfat
vfatID[0] = dict_chipID[vfat]
# Store Max Info
hitCountMaxLat[0] = dict_hVFATHitsVsLat[vfat].GetBinContent(
dict_hVFATHitsVsLat[vfat].GetMaximumBin())
hitCountMaxLatErr[0] = sqrt(hitCountMaxLat[0])
grNMaxLatBinByVFAT.SetPoint(vfat, vfat, hitCountMaxLat[0])
grNMaxLatBinByVFAT.SetPointError(vfat, 0, hitCountMaxLatErr[0])
maxLatBin[0] = dict_hVFATHitsVsLat[vfat].GetBinCenter(
dict_hVFATHitsVsLat[vfat].GetMaximumBin())
grMaxLatBinByVFAT.SetPoint(vfat, vfat, maxLatBin[0])
grMaxLatBinByVFAT.SetPointError(vfat, 0, 0.5) #could be improved upon
# Initialize
dict_fitNHitsVFAT_Sig[vfat] = r.TF1(
"func_N_vs_Lat_VFAT{0}_Sig".format(vfat), "[0]", latFitMin_Sig,
latFitMax_Sig)
dict_fitNHitsVFAT_Noise[vfat] = r.TF1(
"func_N_vs_Lat_VFAT{0}_Noise".format(vfat), "[0]", latMin, latMax)
dict_grNHitsVFAT[vfat] = r.TGraphAsymmErrors(dict_hVFATHitsVsLat[vfat])
dict_grNHitsVFAT[vfat].SetName("g_N_vs_Lat_VFAT{0}".format(vfat))
# Fitting
if performFit:
# Fit Signal
dict_fitNHitsVFAT_Sig[vfat].SetParameter(0, hitCountMaxLat[0])
dict_fitNHitsVFAT_Sig[vfat].SetLineColor(r.kGreen + 1)
dict_grNHitsVFAT[vfat].Fit(dict_fitNHitsVFAT_Sig[vfat], "QR")
# Remove Signal Region
latVal = r.Double()
hitVal = r.Double()
gTempDist = dict_grNHitsVFAT[vfat].Clone(
"g_N_vs_Lat_VFAT{0}_NoSig".format(vfat))
for idx in range(dict_grNHitsVFAT[vfat].GetN() - 1, 0, -1):
gTempDist.GetPoint(idx, latVal, hitVal)
if latFitMin_Noise < latVal and latVal < latFitMax_Noise:
gTempDist.RemovePoint(idx)
# Fit Noise
dict_fitNHitsVFAT_Noise[vfat].SetParameter(0, 0.)
dict_fitNHitsVFAT_Noise[vfat].SetLineColor(r.kRed + 1)
gTempDist.Fit(dict_fitNHitsVFAT_Noise[vfat], "QR")
# Calc Signal & Signal/Noise
hitCountBkg[0] = dict_fitNHitsVFAT_Noise[vfat].GetParameter(0)
hitCountBkgErr[0] = dict_fitNHitsVFAT_Noise[vfat].GetParError(0)
hitCountSig[0] = dict_fitNHitsVFAT_Sig[vfat].GetParameter(
0) - hitCountBkg[0]
hitCountSigErr[0] = sqrt(
(dict_fitNHitsVFAT_Sig[vfat].GetParError(0))**2 +
hitCountBkgErr[0]**2)
SigOverBkg[0] = hitCountSig[0] / hitCountBkg[0]
SigOverBkgErr[0] = sqrt((hitCountSigErr[0] / hitCountBkg[0])**2 +
(hitCountBkgErr[0]**2 *
(hitCountSig[0] / hitCountBkg[0]**2)**2))
# Add to Plot
grVFATSigOverBkg.SetPoint(vfat, vfat, SigOverBkg[0])
grVFATSigOverBkg.SetPointError(vfat, 0, SigOverBkgErr[0])
grVFATNSignalNoBkg.SetPoint(vfat, vfat, hitCountSig[0])
grVFATNSignalNoBkg.SetPointError(vfat, 0, hitCountSigErr[0])
# Print if requested
if debug:
print("{0}\t{1}\t{2}".format(vfat, hitCountSig[0],
SigOverBkg[0]))
pass
# Format
r.gStyle.SetOptStat(0)
dict_grNHitsVFAT[vfat].SetMarkerStyle(21)
dict_grNHitsVFAT[vfat].SetMarkerSize(0.7)
dict_grNHitsVFAT[vfat].SetLineWidth(2)
dict_grNHitsVFAT[vfat].GetXaxis().SetRangeUser(latMin, latMax)
dict_grNHitsVFAT[vfat].GetXaxis().SetTitle("Lat")
dict_grNHitsVFAT[vfat].GetYaxis().SetRangeUser(0, nTrig)
dict_grNHitsVFAT[vfat].GetYaxis().SetTitle("N")
# Write
dirVFAT = dirVFATPlots.mkdir("VFAT{0}".format(vfat))
dirVFAT.cd()
dict_grNHitsVFAT[vfat].Write()
dict_hVFATHitsVsLat[vfat].Write()
if performFit:
dict_fitNHitsVFAT_Sig[vfat].Write()
dict_fitNHitsVFAT_Noise[vfat].Write()
myT.Fill()
pass
# Store - Summary
from gempython.gemplotting.utils.anautilities import getSummaryCanvas, addPlotToCanvas
if performFit:
canv_Summary = getSummaryCanvas(dict_grNHitsVFAT,
name='canv_Summary',
drawOpt='APE1',
gemType=gemType)
canv_Summary = addPlotToCanvas(canv_Summary, dict_fitNHitsVFAT_Noise,
gemType)
canv_Summary.SaveAs(outputDir + '/Summary.png')
else:
canv_Summary = getSummaryCanvas(dict_grNHitsVFAT,
name='canv_Summary',
drawOpt='APE1',
gemType=gemType)
canv_Summary.SaveAs(outputDir + '/Summary.png')
# Store - Sig Over Bkg
if performFit:
canv_SigOverBkg = r.TCanvas("canv_SigOverBkg", "canv_SigOverBkg", 600,
600)
canv_SigOverBkg.cd()
canv_SigOverBkg.cd().SetLogy()
canv_SigOverBkg.cd().SetGridy()
grVFATSigOverBkg.SetTitle("")
grVFATSigOverBkg.SetMarkerStyle(21)
grVFATSigOverBkg.SetMarkerSize(0.7)
grVFATSigOverBkg.SetLineWidth(2)
grVFATSigOverBkg.GetXaxis().SetTitle("VFAT Pos")
grVFATSigOverBkg.GetYaxis().SetTitle("Sig / Bkg)")
grVFATSigOverBkg.GetYaxis().SetTitleOffset(1.25)
grVFATSigOverBkg.GetYaxis().SetRangeUser(1e-1, 1e2)
grVFATSigOverBkg.GetXaxis().SetRangeUser(-0.5, nVFATS + 0.5)
grVFATSigOverBkg.Draw("APE1")
canv_SigOverBkg.SaveAs(outputDir + '/SignalOverBkg.png')
# Store - Signal
if performFit:
canv_Signal = r.TCanvas("canv_Signal", "canv_Signal", 600, 600)
canv_Signal.cd()
grVFATNSignalNoBkg.SetTitle("")
grVFATNSignalNoBkg.SetMarkerStyle(21)
grVFATNSignalNoBkg.SetMarkerSize(0.7)
grVFATNSignalNoBkg.SetLineWidth(2)
grVFATNSignalNoBkg.GetXaxis().SetTitle("VFAT Pos")
grVFATNSignalNoBkg.GetYaxis().SetTitle("Signal Hits")
grVFATNSignalNoBkg.GetYaxis().SetTitleOffset(1.5)
grVFATNSignalNoBkg.GetYaxis().SetRangeUser(0, nTrig)
grVFATNSignalNoBkg.GetXaxis().SetRangeUser(-0.5, nVFATS + 0.5)
grVFATNSignalNoBkg.Draw("APE1")
canv_Signal.SaveAs(outputDir + '/SignalNoBkg.png')
# Store - Sum over all VFATs
canv_LatSum = r.TCanvas("canv_LatSumOverAllVFATs",
"canv_LatSumOverAllVFATs", 600, 600)
canv_LatSum.cd()
hHitsVsLat_AllVFATs.SetXTitle("Latency")
hHitsVsLat_AllVFATs.SetYTitle("N")
hHitsVsLat_AllVFATs.GetXaxis().SetRangeUser(latMin, latMax)
hHitsVsLat_AllVFATs.Draw("hist")
canv_LatSum.SaveAs(outputDir + '/LatSumOverAllVFATs.png')
# Store - Max Hits By Lat Per VFAT
canv_MaxHitsPerLatByVFAT = r.TCanvas("canv_MaxHitsPerLatByVFAT",
"canv_MaxHitsPerLatByVFAT", 1200, 600)
canv_MaxHitsPerLatByVFAT.Divide(2, 1)
canv_MaxHitsPerLatByVFAT.cd(1)
grNMaxLatBinByVFAT.SetTitle("")
grNMaxLatBinByVFAT.SetMarkerStyle(21)
grNMaxLatBinByVFAT.SetMarkerSize(0.7)
grNMaxLatBinByVFAT.SetLineWidth(2)
grNMaxLatBinByVFAT.GetXaxis().SetRangeUser(-0.5, nVFATS + 0.5)
grNMaxLatBinByVFAT.GetXaxis().SetTitle("VFAT Pos")
grNMaxLatBinByVFAT.GetYaxis().SetRangeUser(0, nTrig)
grNMaxLatBinByVFAT.GetYaxis().SetTitle("Hit Count of Max Lat Bin")
grNMaxLatBinByVFAT.GetYaxis().SetTitleOffset(1.7)
grNMaxLatBinByVFAT.Draw("APE1")
canv_MaxHitsPerLatByVFAT.cd(2)
grMaxLatBinByVFAT.SetTitle("")
grMaxLatBinByVFAT.SetMarkerStyle(21)
grMaxLatBinByVFAT.SetMarkerSize(0.7)
grMaxLatBinByVFAT.SetLineWidth(2)
grMaxLatBinByVFAT.GetXaxis().SetTitle("VFAT Pos")
grMaxLatBinByVFAT.GetYaxis().SetTitle("Max Lat Bin")
grMaxLatBinByVFAT.GetYaxis().SetTitleOffset(1.2)
grMaxLatBinByVFAT.GetXaxis().SetRangeUser(-0.5, nVFATS + 0.5)
grMaxLatBinByVFAT.Draw("APE1")
canv_MaxHitsPerLatByVFAT.SaveAs(outputDir + '/MaxHitsPerLatByVFAT.png')
# Store - TObjects
outF.cd()
hHitsVsLat_AllVFATs.Write()
grNMaxLatBinByVFAT.SetName("grNMaxLatBinByVFAT")
grNMaxLatBinByVFAT.Write()
grMaxLatBinByVFAT.SetName("grMaxLatBinByVFAT")
grMaxLatBinByVFAT.Write()
if performFit:
grVFATSigOverBkg.SetName("grVFATSigOverBkg")
grVFATSigOverBkg.Write()
grVFATNSignalNoBkg.SetName("grVFATNSignalNoBkg")
grVFATNSignalNoBkg.Write()
myT.Write()
outF.Close()
def load_data(inputPathNTuples, treeDirName, variables):
print "In data_manager::load_data()::\n inputPathNTuples: ", inputPathNTuples, "\n treeDirName: ", treeDirName
print " variables: ", variables
my_cols_list = variables + ['proces', 'key', 'target', "totalWeight"]
data = pandas.DataFrame(
columns=my_cols_list
) ## right now an empty dataframe with columns = my_cols_list
print "data: ", data
target = None
for process in keys:
print 'process %s ' % (process)
if 'WZ' in process:
sampleName = "WZ"
target = 0
if 'signal' in process:
sampleName = "signal_ggf_spin0_400_hh_wwww"
target = 1
inputNTuples = glob.glob("%s/%s*_forBDTtraining.root" %
(inputPathNTuples, process))
inputTree = "%s/%s/evtTree" % (treeDirName, sampleName)
print "inputTree", inputTree, ", len(inputNTuples):", len(
inputNTuples), " inputNTuples: ", inputNTuples
for intuple in range(0, len(inputNTuples)):
try:
tfile = ROOT.TFile(inputNTuples[intuple])
except:
print "%s FAIL load root file" % inputNTuples[intuple]
continue
try:
tree = tfile.Get(inputTree)
except:
print(inputTree, "FAIL read inputTree", tfile)
continue
if tree is not None:
print "sampleName: ", sampleName, ", process: ", process, ", inputNTuples[intuple]: ", inputNTuples[
intuple], ", nEvents: ", tree.GetEntries()
try:
chunk_arr = tree2array(tree)
except:
print(inputTree, "FAIL tree2array ", tfile)
tfile.Close()
continue
else:
chunk_df = pandas.DataFrame(chunk_arr, columns=variables)
tfile.Close()
chunk_df['proces'] = sampleName
chunk_df['key'] = process
chunk_df['target'] = target
chunk_df["totalWeight"] = chunk_df["evtWeight"]
#print "chunk_df: ",chunk_df
data = data.append(chunk_df, ignore_index=True)
else:
print("file " + list[ii] + "was empty")
nS = len(data.ix[(data.target.values == 1)
& (data.key.values == process)])
nB = len(data.ix[(data.target.values == 0)
& (data.key.values == process)])
print "%s signal size %g, bk size %g, evtWeight %g, totalWeight %g" % (
process, nS, nB, data.ix[(data.key.values
== process)]["evtWeight"].sum(),
data.ix[(data.key.values == process)]["totalWeight"].sum())
nNW = len(data.ix[(data["totalWeight"].values < 0)
& (data.key.values == process)])
print process, " no. of events with -ve weights", nNW
#print 'data to list = ', (data.columns.values.tolist())
n = len(data)
nS = len(data.ix[data.target.values == 1])
nB = len(data.ix[data.target.values == 0])
print treeDirName, " size of sig, bkg: ", nS, nB
return data
def test_tree2array_wrong_type():
rnp.tree2array(list())
def load_single(tree, start_, stop_, branches_):
X = tree2array(tree, start=start_, stop=stop_, branches=branches_)
X = np.array([x[0] for x in X])
return X
from sklearn import metrics
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt
files = ["BDT.root"]
#files = ["BDT_GlobalOnly.root", "BDT_CandsOnly.root", "BDT_AllFeatures.root", "BDT_GlobalNoIP.root", "BDT_GlobalNoKin.root", "BDT_Global_NoIPRoundedKinematics.root", "BDT_AllFeatures_RoundedKinematics.root"]
branches = ['classID', 'BDT']
for file in files:
f = ROOT.TFile(file)
tree = f.Get("TestTree")
data = root_numpy.tree2array(tree, branches=branches)
tpr, fpr, thresh = metrics.roc_curve(
data['classID'], data['BDT']
) # labels are interpreted backwards (that's why tpr and fpr are reversed)
numpy.savez('ROCs/' + file.replace(".root", ""), tpr=tpr, fpr=fpr)
plt.figure()
plt.plot(tpr, fpr, color='aqua', label='BDT')
plt.xscale('log')
plt.xlim([0.001, 1.0])
plt.ylim([0.0, 1.05])
plt.xlabel('False Positive Rate')
plt.ylabel('True Positive Rate')
plt.legend(loc='lower right')
plt.savefig('plot' + file.replace(".root", "") + '.pdf')
treeHH4B = fileHH4B.Get("run/jetTree")
print "Accessed the trees"
# get input variable names from branches
vars = img.getBoostCandBranchNames(treeJJ)
treeVars = vars
print "Variables for jet image creation: ", vars
# create selection criteria
#sel = ""
sel = "jetAK8_pt > 500 && jetAK8_mass > 50"
#sel = "tau32 < 9999. && et > 500. && et < 2500. && bDisc1 > -0.05 && SDmass < 400"
# make arrays from the trees
arrayJJ = tree2array(treeJJ, treeVars, sel)
arrayJJ = tools.appendTreeArray(arrayJJ)
imgArrayJJ = img.makeBoostCandFourVector(arrayJJ)
arrayHH4W = tree2array(treeHH4W, treeVars, sel)
arrayHH4W = tools.appendTreeArray(arrayHH4W)
imgArrayHH4W = img.makeBoostCandFourVector(arrayHH4W)
arrayHH4B = tree2array(treeHH4B, treeVars, sel)
arrayHH4B = tools.appendTreeArray(arrayHH4B)
imgArrayHH4B = img.makeBoostCandFourVector(arrayHH4B)
print "Made candidate 4 vector arrays from the datasets"
#==================================================================================
# Make Jet Images /////////////////////////////////////////////////////////////////
from ROOT import *
import root_numpy
import matplotlib
matplotlib.use("Agg")
import matplotlib.pyplot as plt
print 1
import numpy
import sys
file = TFile('/gpfs/ddn/cms/user/lgiannini/DeepNtupleRegression/Train_wSV.root'
) #/uscms_data/d3/lgiannin/tree_Reg.root')
tree = file.Get("tree")
tree2 = root_numpy.tree2array(
tree, ["nPVs", "jetpt", "jeteta", "genjetpt", "genjetpt_wNu"],
selection="")
print tree2.shape
print tree2[0].shape
for i in range(len(tree2[0])):
print i, tree2[0][i]
print tree2
tree2 = root_numpy.rec2array(tree2)
print tree2.shape
t2 = root_numpy.tree2array(
tree,
def tvars(rootfile, first, last):
stringa = [
"seed_pt",
"seed_eta",
"seed_phi",
"seed_mass",
"seed_dz",
"seed_dxy",
"seed_3D_ip",
"seed_3D_sip",
"seed_2D_ip",
"seed_2D_sip",
"seed_3D_signedIp",
"seed_3D_signedSip",
"seed_2D_signedIp",
"seed_2D_signedSip",
"seed_chi2reduced",
"seed_nPixelHits",
"seed_nHits",
"seed_jetAxisDistance",
"seed_jetAxisDlength",
#more truth vars
"seed_MC_pt",
"seed_MC_eta",
"seed_MC_phi",
"seed_MC_mass",
"seed_MC_dz",
"seed_MC_dxy",
"seed_MC_MomPdgId",
"seed_MC_MomFlavour",
"seed_MC_BChain",
"seed_MC_DChain", #no B in this case
"seed_MC_vx",
"seed_MC_vy",
"seed_MC_vz",
"seed_MC_pvd"
]
stringa2 = [
"nearTracks_pt",
"nearTracks_eta",
"nearTracks_phi",
"nearTracks_dz",
"nearTracks_dxy",
"nearTracks_mass",
"nearTracks_3D_ip",
"nearTracks_3D_sip",
"nearTracks_2D_ip",
"nearTracks_2D_sip",
"nearTracks_PCAdist",
"nearTracks_PCAdsig",
"nearTracks_PCAonSeed_x",
"nearTracks_PCAonSeed_y",
"nearTracks_PCAonSeed_z",
"nearTracks_PCAonSeed_xerr",
"nearTracks_PCAonSeed_yerr",
"nearTracks_PCAonSeed_zerr",
"nearTracks_PCAonTrack_x",
"nearTracks_PCAonTrack_y",
"nearTracks_PCAonTrack_z",
"nearTracks_PCAonTrack_xerr",
"nearTracks_PCAonTrack_yerr",
"nearTracks_PCAonTrack_zerr",
"nearTracks_dotprodTrack",
"nearTracks_dotprodSeed",
"nearTracks_dotprodTrackSeed2D",
"nearTracks_dotprodTrackSeed3D",
"nearTracks_dotprodTrackSeedVectors2D",
"nearTracks_dotprodTrackSeedVectors3D",
"nearTracks_PCAonSeed_pvd",
"nearTracks_PCAonTrack_pvd",
"nearTracks_PCAjetAxis_dist",
"nearTracks_PCAjetMomenta_dotprod",
"nearTracks_PCAjetDirs_DEta",
"nearTracks_PCAjetDirs_DPhi",
#more MC vars
"nearTracks_MC_pt",
"nearTracks_MC_eta",
"nearTracks_MC_phi",
"nearTracks_MC_dz",
"nearTracks_MC_dxy",
"nearTracks_MC_MomPdgId",
"nearTracks_MC_MomFlavour",
"nearTracks_MC_BChain",
"nearTracks_MC_DChain", #no B in this case
"nearTracks_MC_Track_vx",
"nearTracks_MC_Track_vy",
"nearTracks_MC_Track_vz",
"nearTracks_MC_fromSeedVtx",
"nearTracks_MC_fromSeedChain",
"nearTracks_MC_pvd",
#"nearTracks_MC_fromSeedVtx*(nearTracks_MC_pvd>0)"
]
f = TFile(rootfile)
# tree=f.Get("analyzer1/tree")
tree = root_numpy.tree2array(f.Get('analyzer1/tree'),
branches=stringa2,
selection="(jet_pt>30)&&(abs(jet_eta)<2.4)",
start=first,
stop=last)
print "loaded"
tree2 = root_numpy.rec2array(tree)
print "s"
print tree2.shape
print round(time.time() - starttime, 2), "reshape"
tree3 = tree2.reshape((200, 51, len(tree)))
tree3 = tree3.reshape((10, 51 * 20, len(tree)))
print tree3.shape
tree3 = tree3.swapaxes(0, 2)
t2 = root_numpy.tree2array(f.Get('analyzer1/tree'),
branches=stringa,
selection="(jet_pt>30)&&(abs(jet_eta)<2.4)",
start=first,
stop=last)
t2 = root_numpy.rec2array(t2)
print t2.shape
t2 = t2.reshape((10, len(stringa), len(tree)))
t2 = t2.swapaxes(0, 2)
tree5 = numpy.concatenate((t2, tree3), axis=1)
print tree5.shape
numpy.save(
"tvars_" + str(first) + "_" + str(last) + "_" +
rootfile.split(".")[0] + ".npy", tree5)
print time.time() - starttime
f.Close()
os.system("mv " + "tvars_" + str(first) + "_" + str(last) + "_" +
rootfile.split(".")[0] + ".npy" +
" /gpfs/ddn/users/lgiannini/NN/DataRECO")
def convert_tree_to_np(sources, destination, npy_files=[]):
""" Converts the root files in sources to numpy arrays
Params
sources : list
root file paths/names or path to directory of root files
destination : str
path to directory where the npy files will be saved
npy_files : list
list of already converted files (for recursive functionality)
Returns
list
paths to the converted files
"""
for i in xrange(len(sources)):
if os.path.isdir(sources[i]) and ('failed' not in sources[i]):
# source is a directory -> recurse on all files in directory
new_sources = [
sources[i] + '/' + e for e in os.listdir(sources[i])
]
new_destination = destination + '/' + sources[i].split('/')[-1]
print('new_sources ', len(new_sources), new_sources[-9:])
print('new_destination ', new_destination)
logging.info('new_sources ' + str(len(new_sources)) + ' ' +
str(new_sources[-9:]))
logging.info('new_destination ' + new_destination)
os.mkdir(new_destination)
convert_tree_to_np(new_sources, new_destination, npy_files)
else:
if ".root" in sources[i]:
try:
# print(i, sources[i])
logging.info(str(i) + ' ' + sources[i])
print(str(i) + ' ', end="")
sys.stdout.flush()
tChain = rt.TChain('MyAnalysis/MyTree')
tChain.Add(sources[i])
array = root_numpy.tree2array(tChain)
# print 'Total number of entries: ',tChain.GetEntries()
pkl_file_name = destination + '/' + sources[i].split(
'/')[-1][:-5]
np.save(pkl_file_name, array)
npy_files.append(pkl_file_name + '.npy')
except Exception as e:
if os.path.exists(failed.pkl):
continue
else:
mylist = []
with open('failed.pkl', 'wb') as f:
pickle.dump(mylist, f)
print("")
print(e)
print(sources[i], " ** FAILED ** ")
logging.error(sources[i] + " ** FAILED ** ")
logging.error(e)
f = open('failed.pkl', 'rb')
failed = pickle.load(f)
f.close()
failed.append(sources[i])
f = open('failed.pkl', 'wb')
pickle.dump(failed, f)
f.close()
return npy_files
import pandas as pd
from ROOT import *
from root_numpy import root2array, tree2array
from root_numpy import testdata
from IPython.display import display
import numpy as np
path = '../data/'
# --read signal dataset
sig_file = TFile.Open(path + 'sig.root')
sig_tree = sig_file.Get('ntuple')
sig_arr = tree2array(sig_tree)
sig_df = pd.DataFrame(sig_arr)
# --read background dataset
bkg_file = TFile.Open(path + 'bkg.root')
bkg_tree = bkg_file.Get('ntuple')
bkg_arr = tree2array(bkg_tree)
bkg_df = pd.DataFrame(bkg_arr)
print('sig: ', sig_df.shape[0])
print('bkg: ', bkg_df.shape[0])
# --Normalize
def MinMaxScaler(data):
numerator = data - np.min(data, 0)
denominator = np.max(data, 0) - np.min(data, 0)
denominator = denominator.astype('float')
return numerator / denominator
path = '/beegfs/desy/user/hezhiyua/backed/dustData/crab_folder_v2/'
pathOut = '/beegfs/desy/user/hezhiyua/backed/dustData/crab_folder_v2/test/'
Fname = 'VBFH_HToSSTobbbb_MH-125_MS-40_ctauS-500_TuneCUETP8M1_13TeV-powheg-pythia8_PRIVATE-MC.root'
entries = 200
fin = TFile(path + Fname)
tin = fin.Get('ntuple/tree')
tm1 = tm()
s_cut = 1000#None#100
arr_energy = rnp.tree2array(tin, ['PFCandidates.energy'], stop=s_cut)
arr_phi = rnp.tree2array(tin, ['PFCandidates.phi'], stop=s_cut)
arr_eta = rnp.tree2array(tin, ['PFCandidates.eta'], stop=s_cut)
arr_jetindex = rnp.tree2array(tin, ['PFCandidates.jetIndex'], stop=s_cut)
#e_npar = np.array(arr_energy)
e_df = pd.DataFrame(arr_energy)
phi_df = pd.DataFrame(arr_phi)
eta_df = pd.DataFrame(arr_eta)
#print e_npar
#print arr_energy[3]
print e_df.loc[3,'PFCandidates.energy'][3]
df = pd.DataFrame()
df_o = pd.DataFrame()
df['e'] = e_df
eosdir = "root://cmseos.fnal.gov//store/user/jmanagan/MVAtraining_2017_Jan2021/"
## Choosing valid events with appropriate characteristics and cutting the rest
seltrain = "isValidTTDecayMode_DeepAK8 == 0 && Tprime2_DeepAK8_Mass < 0 && NJetsAK8_JetSubCalc > 2"
seltest = "isValidTTDecayMode_DeepAK8 == 0 && Tprime2_DeepAK8_Mass >= 0"
treeVars = vars
## Getting values from trees for each parent particle and either keeping them in an array or adding them together
fileTTToSemiLepT = TFile.Open(
eosdir +
"TTJets_SingleLeptFromT_TuneCP5_13TeV-madgraphMLM-pythia8_hadd.root",
"READ")
treeTTToSemiLepT = fileTTToSemiLepT.Get("ljmet")
trainTTToSemiLepT = tree2array(treeTTToSemiLepT, treeVars, seltrain)
testTTToSemiLepT = tree2array(treeTTToSemiLepT, treeVars, seltest)
## Selection with Single Lept from TBar
fileTTToSemiLepTb = TFile.Open(
eosdir +
"TTJets_SingleLeptFromTbar_TuneCP5_13TeV-madgraphMLM-pythia8_hadd.root",
"READ")
treeTTToSemiLepTb = fileTTToSemiLepTb.Get("ljmet")
trainTTToSemiLepTb = tree2array(treeTTToSemiLepTb, treeVars, seltrain)
testTTToSemiLepTb = tree2array(treeTTToSemiLepTb, treeVars, seltest)
## Selection with signals
fileTprime = TFile.Open(
eosdir + "TprimeTprime_M-1000_TuneCP5_13TeV-madgraph-pythia8_hadd.root",
"READ")
@author: rupeshdotel
"""
import numpy as np
import LT.box as B
import ROOT as R
from root_numpy import tree2array
import matplotlib.pyplot as plt
#%%
rfile = R.TFile(
"/Users/rupeshdotel/analysis/work/pi0pippimeta/data/qfactor_data/qfactortree/qfactortree_for_may_10_gluexI.root"
)
intree = rfile.Get('qfactortree')
d = tree2array(intree)
#%%
mm2m = d['mm2m']
mpi013 = d['mpi013']
mpi014 = d['mpi014']
mpi023 = d['mpi023']
mpi024 = d['mpi024']
metap = d['metap']
metappi0 = d['metappi0']
cost_etap = d['cos_t']
phi_etap = d['phi_gj']
mpippimpi0 = d['mpippimpi0']
mpi0p = d['mpi0p']
## Open ROOT files
print 'Opening files...'
eosdir = "root://cmseos.fnal.gov//store/user/cholz/Step2MVAtraining_0432020/"
#sel = "Bprime2_DeepAK8_Mass < 0"
sel = "isValidBBDecayMode_DeepAK8 == 0 && Bprime2_DeepAK8_Mass < 0 && NJetsAK8_JetSubCalc > 2"
treeVars = vars
print 'getting trees...','Making training arrays...','Making testing arrays...'
seltest = "isValidBBDecayMode_DeepAK8 == 0 && Bprime2_DeepAK8_Mass >= 0"
for i in range(1,7):
fileTTToSemiLepT = TFile.Open(eosdir + "TTJets_SingleLeptFromT_TuneCUETP8M1_13TeV-madgraphMLM-pythia8_"+ str(i)+"_hadd.root", "READ")
treeTTToSemiLepT = fileTTToSemiLepT.Get("ljmet")
if i == 1:
arrayTTToSemiLepT = tree2array(treeTTToSemiLepT, treeVars, sel)
testTTToSemiLepT = tree2array(treeTTToSemiLepT, treeVars, seltest)
else:
arrayTTToSemiLepT = np.concatenate([arrayTTToSemiLepT,tree2array(treeTTToSemiLepT, treeVars, sel)])
testTTToSemiLepT = np.concatenate([testTTToSemiLepT,tree2array(treeTTToSemiLepT, treeVars, seltest)])
for i in range(1,3):
fileTTToSemiLepTb = TFile.Open(eosdir + "TTJets_SingleLeptFromTbar_TuneCUETP8M1_13TeV-madgraphMLM-pythia8_"+ str(i)+"_hadd.root", "READ")
treeTTToSemiLepTb = fileTTToSemiLepTb.Get("ljmet")
if i == 1:
arrayTTToSemiLepTb = tree2array(treeTTToSemiLepTb, treeVars, sel)
testTTToSemiLepTb = tree2array(treeTTToSemiLepTb, treeVars, seltest)
else:
arrayTTToSemiLepTb = np.concatenate([arrayTTToSemiLepTb,tree2array(treeTTToSemiLepTb, treeVars, sel)])
testTTToSemiLepTb = np.concatenate([testTTToSemiLepTb,tree2array(treeTTToSemiLepTb, treeVars, seltest)])
path_tree = '/home/ucl/cp3/fbury/storage/NNAndELLipseOutputTrees/model_'+str(args.model)+'/'
################################################################################
# Input Trees #
################################################################################
for name in glob.glob(path_tree+'*.root'):
filename = name.replace(path_tree,'')
num = [int(s) for s in re.findall('\d+',filename )]
if num[0]!=mH_select or num[1]!=mA_select:
continue
break
f = ROOT.TFile.Open(name)
t = f.Get("tree")
sig = tree2array(t,branches=['NN_out','Ell_out','weight'],selection='id==0')
DYToLL_0J = tree2array(t,branches=['NN_out','Ell_out','weight'],selection='id==1')
DYToLL_1J = tree2array(t,branches=['NN_out','Ell_out','weight'],selection='id==2')
DYToLL_2J = tree2array(t,branches=['NN_out','Ell_out','weight'],selection='id==3')
TT_Other = tree2array(t,branches=['NN_out','Ell_out','weight'],selection='id==4')
TTTo2L2Nu = tree2array(t,branches=['NN_out','Ell_out','weight'],selection='id==5')
for cn in cut_NN:
print ('NN cut : ',cn)
N_sig = np.sum(sig[sig[:]['NN_out']>cn]['weight'])
N_DYToLL_0J = np.sum(DYToLL_0J[DYToLL_0J[:]['NN_out']>cn]['weight'])
N_DYToLL_1J = np.sum(DYToLL_1J[DYToLL_1J[:]['NN_out']>cn]['weight'])
N_DYToLL_2J = np.sum(DYToLL_2J[DYToLL_2J[:]['NN_out']>cn]['weight'])
N_TT_Other = np.sum(TT_Other[TT_Other[:]['NN_out']>cn]['weight'])
N_TTTo2L2Nu = np.sum(TTTo2L2Nu[TTTo2L2Nu[:]['NN_out']>cn]['weight'])
procP3 = glob.glob(inputPath + "/" + folderName + "_fastsim_p3/" +
folderName +
"_fastsim_p3_forBDTtraining_OS_central_*.root")
list = procP1 + procP2 + procP3
else:
procP1 = glob.glob(inputPath + "/" + folderName + "_fastsim/" +
folderName +
"_fastsim_forBDTtraining_OS_central_*.root")
list = procP1
print("Date: ", time.asctime(time.localtime(time.time())))
for ii in range(0, len(list)): #
#print (list[ii],inputTree)
tfile = ROOT.TFile(list[ii])
tree = tfile.Get(inputTree)
if tree is not None:
chunk_arr = tree2array(tree) #, start=start, stop = stop)
chunk_df = pandas.DataFrame(chunk_arr) #
chunk_df['key'] = folderName
chunk_df['target'] = target
#chunk_df['file']=list[ii].split("_")[10]
if channel == "2lss_1tau":
data[
"totalWeight"] = data.evtWeight * data.tau_frWeight * data.lep1_frWeight * data.lep2_frWeight
if channel == "1l_2tau": data["totalWeight"] = data.evtWeight
data = data.append(chunk_df, ignore_index=True)
else:
print("file " + list[ii] + "was empty")
tfile.Close()
print(data.columns.values.tolist())
n = len(data)
nS = len(data.ix[data.target.values == 0])
data_chf1 = root2array(directory, what_tree, branch_names_chfonly1)
data_chfp1 = root2array(directory, what_tree, branch_names_chfp1)
data_tot1 = root2array(directory, what_tree, branch_names_tot1)
data_noreg1 = root2array(directory, what_tree, branch_names_noreg1)
data_xgb2 = root2array(directory, what_tree, branch_names_xgb2)
data_chf2 = root2array(directory, what_tree, branch_names_chfonly2)
data_chfp2 = root2array(directory, what_tree, branch_names_chfp2)
data_tot2 = root2array(directory, what_tree, branch_names_tot2)
data_noreg2 = root2array(directory, what_tree, branch_names_noreg2)
data_cw = root2array(directory_cw, what_tree_cw, branch_names_cw)
'''
data_xgb1 = tree2array(what_tree, branch_names_xgb1)
data_chf1 = tree2array(what_tree, branch_names_chfonly1)
data_chfp1 = tree2array(what_tree, branch_names_chfp1)
data_tot1 = tree2array(what_tree, branch_names_tot1)
data_cw = tree2array(what_tree_cw, branch_names_cw1)
data_noreg1 = tree2array(what_tree, branch_names_noreg1)
data_xgb2 = tree2array(what_tree, branch_names_xgb2)
data_chf2 = tree2array(what_tree, branch_names_chfonly2)
data_chfp2 = tree2array(what_tree, branch_names_chfp2)
data_tot2 = tree2array(what_tree, branch_names_tot2)
data_cw = tree2array(what_tree_cw, branch_names_cw2)
data_noreg2 = tree2array(what_tree, branch_names_noreg2)
#================LOAD WEIGHT FILES========================
def main():
ROOT.gROOT.SetBatch(1)
ROOT.gStyle.SetOptStat(0)
njet = "all"
#process = {"sig1lnotTwB1wt1000tminfixed_5K": "sig", "bkg1lnotTwB1wt1000tminfixed_5K": "bkg"}
#process = {"sig1lnotTwB1wt100_50K": "sig", "bkg1lnotTwB1wt100_50K": "bkg"}
process = {
"sig1lnotTwB1wt1000_50K": "sig",
"bkg1lnotTwB1wt1000_50K": "bkg"
}
#process = "sig1lnotTwB1wt1000_5K"
#process = "bkg1lnotTwB1wt1000_5K"
#process = "sig1lnotTwB1wt1000_1K"
#process = "bkg1lnotTwB1wt1000_1K"
#process = "sig1lnotTwB1wt100_1K"
#process = "bkg1lnotTwB1wt100_1K"
#process = "sig"
#process = "bkg"
#process = "sigOld"
#process = "bkgOld"
#njet = "ge10"
#njet = "9"
#njet = "8"
#njet = "7"
#njet = "6"
#njet = "5"
tlist = {}
for p in process.keys():
if (njet == "all"):
t = ROOT.TChain("nominal_Loose")
jet_cat = ["ge10", "9", "8", "7", "6", "5"]
for jetc in jet_cat:
fj = ROOT.TFile(
"/afs/cern.ch/work/s/sosen/ChongbinTop/common-framework/run-offline/test_%s/ljets%sj/ttbar_powpyt8.root"
% (p, jetc), 'READ')
t.Add(
"/afs/cern.ch/work/s/sosen/ChongbinTop/common-framework/run-offline/test_%s/ljets%sj/ttbar_powpyt8.root"
% (p, jetc))
else:
f = ROOT.TFile(
"/afs/cern.ch/work/s/sosen/ChongbinTop/common-framework/run-offline/test_%s/ljets%sj/ttbar_powpyt8.root"
% (p, njet), 'READ')
f.ls()
t = f.Get("nominal_Loose")
t.ls()
tlist[process[p] + "tree"] = t
#bucket type count
Stwcount = rnp.tree2array(tlist["sigtree"], branches="twcount")
Stmincount = rnp.tree2array(tlist["sigtree"], branches="tmincount")
St0count = rnp.tree2array(tlist["sigtree"], branches="t0count")
##
SmW0 = rnp.tree2array(tlist["sigtree"], branches="mW0")
SmW1 = rnp.tree2array(tlist["sigtree"], branches="mW1")
SmBucketPrim0 = rnp.tree2array(tlist["sigtree"], branches="mBucketPrim0")
SmBucketPrim1 = rnp.tree2array(tlist["sigtree"], branches="mBucketPrim1")
SmBucketPrim = np.concatenate((SmBucketPrim0, SmBucketPrim1), axis=None)
Stwmass0 = rnp.tree2array(tlist["sigtree"], branches="twmass0")
StwPt0 = rnp.tree2array(tlist["sigtree"], branches="twPt0")
SNaddjets = rnp.tree2array(tlist["sigtree"], branches="Naddjets")
#bucket type count
Btwcount = rnp.tree2array(tlist["bkgtree"], branches="twcount")
Btmincount = rnp.tree2array(tlist["bkgtree"], branches="tmincount")
Bt0count = rnp.tree2array(tlist["bkgtree"], branches="t0count")
##
BmW0 = rnp.tree2array(tlist["bkgtree"], branches="mW0")
BmW1 = rnp.tree2array(tlist["bkgtree"], branches="mW1")
BmBucketPrim0 = rnp.tree2array(tlist["bkgtree"], branches="mBucketPrim0")
BmBucketPrim1 = rnp.tree2array(tlist["bkgtree"], branches="mBucketPrim1")
BmBucketPrim = np.concatenate((BmBucketPrim0, BmBucketPrim1), axis=None)
Btwmass0 = rnp.tree2array(tlist["bkgtree"], branches="twmass0")
BtwPt0 = rnp.tree2array(tlist["bkgtree"], branches="twPt0")
BNaddjets = rnp.tree2array(tlist["bkgtree"], branches="Naddjets")
c0 = ROOT.TCanvas('c', 'c', 800, 600)
leg0 = ROOT.TLegend(0.65, 0.75, 0.88, 0.88)
leg0.SetFillColor(0)
leg0.SetLineColor(0)
#bucket type count
ShNaddjets = ROOT.TH1F("signal hNaddjets", "", 21, -0.5, 20.5)
fill_hist(ShNaddjets, SNaddjets, "additional jets per event", "a.u.")
ShNaddjets.Scale(1. / (ShNaddjets.Integral()))
ShNaddjets.SetLineColor(ROOT.kRed)
BhNaddjets = ROOT.TH1F("bkg hNaddjets", "", 21, -0.5, 20.5)
fill_hist(BhNaddjets, BNaddjets, "additional jets per event", "a.u.")
BhNaddjets.Scale(1. / (BhNaddjets.Integral()))
BhNaddjets.SetLineColor(ROOT.kBlack)
ShNaddjets.SetMaximum(
max(ShNaddjets.GetMaximum(), BhNaddjets.GetMaximum()) * 1.1)
ShNaddjets.Draw("hist")
leg0.AddEntry(ShNaddjets, 'signal', "L")
BhNaddjets.Draw("hist same")
leg0.AddEntry(BhNaddjets, 'ttbar+jets (bkg)', "L")
leg0.Draw()
c0.Print("OverlayhNaddjets_%sjetregion.eps" % njet)
c1 = ROOT.TCanvas('c', 'c', 800, 600)
leg1 = ROOT.TLegend(0.65, 0.75, 0.88, 0.88)
leg1.SetFillColor(0)
leg1.SetLineColor(0)
#bucket type count
Shtwcount = ROOT.TH1F("signal htwcount", "", 4, -0.5, 3.5)
fill_hist(Shtwcount, Stwcount, "tw buckets per event", "a.u.")
Shtwcount.Scale(1. / (Shtwcount.Integral()))
Shtwcount.SetLineColor(ROOT.kRed)
Bhtwcount = ROOT.TH1F("bkg htwcount", "", 4, -0.5, 3.5)
fill_hist(Bhtwcount, Btwcount, "tw buckets per event", "a.u.")
Bhtwcount.Scale(1. / (Bhtwcount.Integral()))
Bhtwcount.SetLineColor(ROOT.kBlack)
Shtwcount.SetMaximum(
max(Shtwcount.GetMaximum(), Bhtwcount.GetMaximum()) * 1.1)
Shtwcount.Draw("hist")
leg1.AddEntry(Shtwcount, 'signal', "L")
Bhtwcount.Draw("hist same")
leg1.AddEntry(Bhtwcount, 'ttbar+jets (bkg)', "L")
leg1.Draw()
c1.Print("Overlayhtwcount_%sjetregion.eps" % njet)
c2 = ROOT.TCanvas('c', 'c', 800, 600)
leg2 = ROOT.TLegend(0.65, 0.75, 0.88, 0.88)
leg2.SetFillColor(0)
leg2.SetLineColor(0)
#bucket type count
ShmW0 = ROOT.TH1F("signal hmW0",
"Mass of the (possible) W candidate in B1", 150, 0.0001,
300)
fill_hist(ShmW0, SmW0, "Mass (GeV)", "")
ShmW0.Scale(1. / (ShmW0.Integral()))
ShmW0.SetLineColor(ROOT.kRed)
BhmW0 = ROOT.TH1F("bkg hmW0", "Mass of the (possible) W candidate in B1",
150, 0.0001, 300)
fill_hist(BhmW0, BmW0, "Mass (GeV)", "")
BhmW0.Scale(1. / (BhmW0.Integral()))
BhmW0.SetLineColor(ROOT.kBlack)
ShmW0.SetMaximum(max(ShmW0.GetMaximum(), BhmW0.GetMaximum()) * 1.1)
ShmW0.Draw("hist")
leg2.AddEntry(ShmW0, 'signal', "L")
BhmW0.Draw("hist same")
leg2.AddEntry(BhmW0, 'ttbar+jets (bkg)', "L")
leg2.Draw()
c2.Print("OverlayhmW0_%sjetregion.eps" % njet)
c3 = ROOT.TCanvas('c', 'c', 800, 600)
leg3 = ROOT.TLegend(0.65, 0.75, 0.88, 0.88)
leg3.SetFillColor(0)
leg3.SetLineColor(0)
#bucket type count
ShmBucketPrim0 = ROOT.TH1F("signal hmBucketPrim0", "Mass of B1", 150, 0,
300)
fill_hist(ShmBucketPrim0, SmBucketPrim0, "Mass (GeV)", "")
ShmBucketPrim0.Scale(1. / (ShmBucketPrim0.Integral()))
ShmBucketPrim0.SetLineColor(ROOT.kRed)
BhmBucketPrim0 = ROOT.TH1F("bkg hmBucketPrim0", "Mass of B1", 150, 0, 300)
fill_hist(BhmBucketPrim0, BmBucketPrim0, "Mass (GeV)", "")
BhmBucketPrim0.Scale(1. / (BhmBucketPrim0.Integral()))
BhmBucketPrim0.SetLineColor(ROOT.kBlack)
ShmBucketPrim0.SetMaximum(
max(ShmBucketPrim0.GetMaximum(), BhmBucketPrim0.GetMaximum()) * 1.1)
ShmBucketPrim0.Draw("hist")
leg3.AddEntry(ShmBucketPrim0, 'signal', "L")
BhmBucketPrim0.Draw("hist same")
leg3.AddEntry(BhmBucketPrim0, 'ttbar+jets (bkg)', "L")
leg3.Draw()
c3.Print("OverlayhmBucketPrim0_%sjetregion.eps" % njet)
c4 = ROOT.TCanvas('c', 'c', 800, 600)
leg4 = ROOT.TLegend(0.65, 0.75, 0.88, 0.88)
leg4.SetFillColor(0)
leg4.SetLineColor(0)
#bucket type count
ShmBucketPrim1 = ROOT.TH1F("signal hmBucketPrim1", "Mass of B2", 150, 0,
300)
fill_hist(ShmBucketPrim1, SmBucketPrim1, "Mass (GeV)", "")
ShmBucketPrim1.Scale(1. / (ShmBucketPrim1.Integral()))
ShmBucketPrim1.SetLineColor(ROOT.kRed)
BhmBucketPrim1 = ROOT.TH1F("bkg hmBucketPrim1", "Mass of B2", 150, 0, 300)
fill_hist(BhmBucketPrim1, BmBucketPrim1, "Mass (GeV)", "")
BhmBucketPrim1.Scale(1. / (BhmBucketPrim1.Integral()))
BhmBucketPrim1.SetLineColor(ROOT.kBlack)
ShmBucketPrim1.SetMaximum(
max(ShmBucketPrim1.GetMaximum(), BhmBucketPrim1.GetMaximum()) * 1.1)
ShmBucketPrim1.Draw("hist")
leg4.AddEntry(ShmBucketPrim1, 'signal', "L")
BhmBucketPrim1.Draw("hist same")
leg4.AddEntry(BhmBucketPrim1, 'ttbar+jets (bkg)', "L")
leg4.Draw()
c4.Print("OverlayhmBucketPrim1_%sjetregion.eps" % njet)
def do_cut(did, files, supercuts, weights, tree_name, output_directory,
eventWeightBranch, doNumpy, pids):
position = -1
if pids is not None:
# handle pid registration
if os.getpid() not in pids: pids[np.argmax(pids == 0)] = os.getpid()
# this gives us the position of this particular process in our list of processes
position = np.where(pids == os.getpid())[0][0]
start = clock()
try:
# load up the tree for the files
tree = get_ttree(tree_name, files, eventWeightBranch)
# if using numpy optimization, load the tree as a numpy array to apply_cuts on
if doNumpy:
# this part is tricky, a user might specify multiple branches
# in their selection string, so we will remove non-alphanumeric characters (underscores are safe)
# and remove anything else that is an empty string (hence the filter)
# and then flatten the entire list, removing duplicate branch names
'''
totalSelections = []
for supercut in supercuts:
selection = supercut['selections']
# filter out non-alphanumeric
selection = p.sub(' ', selection.format("-", "-", "-", "-", "-", "-", "-", "-", "-", "-"))
# split on spaces, since we substituted non alphanumeric with spaces
selections = selection.split(' ')
# remove empty elements
filter(None, selections)
totalSelections.append(selections)
# flatten the thing
totalSelections = itertools.chain.from_iterable(totalSelections)
# remove duplicates
totalSelections = list(set(totalSelections))
'''
branchesSpecified = list(
set(
itertools.chain.from_iterable(
selection_to_branches(supercut['selections'], tree)
for supercut in supercuts)))
eventWeightBranchesSpecified = list(
set(selection_to_branches(eventWeightBranch, tree)))
# get actual list of branches in the file
availableBranches = tree_get_branches(
tree, eventWeightBranchesSpecified)
# remove anything that doesn't exist
branchesToUse = [
branch for branch in branchesSpecified
if branch in availableBranches
]
branchesSkipped = list(set(branchesSpecified) - set(branchesToUse))
if branchesSkipped:
logger.info("The following branches have been skipped...")
for branch in branchesSkipped:
logger.info("\t{0:s}".format(branch))
tree = rnp.tree2array(tree,
branches=eventWeightBranchesSpecified +
branchesToUse)
# get the scale factor
sample_scaleFactor = get_scaleFactor(weights, did)
# build the containing canvas for all histograms drawn in `apply_selection`
canvas = ROOT.TCanvas('test{0:s}'.format(did), 'test{0:s}'.format(did),
200, 10, 100, 100)
# iterate over the cuts available
cuts = {}
for cut in tqdm(get_cut(copy.deepcopy(supercuts)),
desc='Working on DID {0:s}'.format(did),
total=get_n_cuts(supercuts),
disable=(position == -1),
position=position + 1,
leave=True,
mininterval=5,
maxinterval=10,
unit='cuts',
dynamic_ncols=True):
cut_hash = get_cut_hash(cut)
rawEvents, weightedEvents = apply_cuts(tree,
cut,
eventWeightBranch,
doNumpy,
canvas=canvas)
scaledEvents = weightedEvents * sample_scaleFactor
cuts[cut_hash] = {
'raw': rawEvents,
'weighted': weightedEvents,
'scaled': scaledEvents
}
logger.info("Applied {0:d} cuts".format(len(cuts)))
with open('{0:s}/{1:s}.json'.format(output_directory, did), 'w+') as f:
f.write(json.dumps(cuts, sort_keys=True, indent=4))
result = True
del canvas
except:
logger.exception("Caught an error - skipping {0:s}".format(did))
result = False
end = clock()
return (result, end - start)
def main():
treeFolder = "analysis/allEvents/"
treeName = "massData"
wrMassBranch = "WRMass"
SRmassBranch = "superResolvedNNMass"
RmassBranch = "resolvedNNMass"
correctMassBranch = "correctNMass"
incorrectMassBranch = "incorrectNMass"
leadMassBranch = "leadNMass"
subleadMassBranch = "subNMass"
weightBranch = "weight"
#LOADING THE TTREE
fileNames = ["TTTo2L2Nu.root"]
crossSections = [88.29]
counts2 = [79140880]
#numpy arrays
# make new root file with new tree
file = ROOT.TFile("fullttbar.root", 'recreate')
tree = ROOT.TTree("fullttbar", "fullttbar")
# create 1 dimensional float arrays as fill variables, in this way the float
# array serves as a pointer which can be passed to the branch
# create some random numbers, assign them into the fill variables and call Fill()
WRMass = np.zeros(1, dtype=float)
resolvedNNMass = np.zeros(1, dtype=float)
superResolvedNNMass = np.zeros(1, dtype=float)
correctNMass = np.zeros(1, dtype=float)
treeWeight = np.zeros(1, dtype=float)
treeWeight2 = np.zeros(1, dtype=float)
incorrectNMass = np.zeros(1, dtype=float)
leadNMass = np.zeros(1, dtype=float)
subNMass = np.zeros(1, dtype=float)
tree.Branch("WRMass", WRMass, "WRMass/D")
tree.Branch("resolvedNNMass", resolvedNNMass, "resolvedNNMass/D")
tree.Branch("superResolvedNNMass", superResolvedNNMass,
"superResolvedNNMass/D")
tree.Branch("correctNMass", correctNMass, "correctNMass/D")
tree.Branch("incorrectNMass", incorrectNMass, "incorrectNMass/D")
tree.Branch("leadNMass", leadNMass, "leadNMass/D")
tree.Branch("subNMass", subNMass, "subNMass/D")
tree.Branch("weight", treeWeight, "weight/D")
tree.Branch("weight2", treeWeight2, "weight2/D")
for fileName, xSec, count2 in zip(fileNames, crossSections, counts2):
rootfile = ROOT.TFile.Open(fileName, "read")
massTree = rootfile.Get(treeFolder + treeName)
countHisto = rootfile.Get(treeFolder + "countHisto")
counts = countHisto.GetBinContent(1)
print(counts)
WRmassArray = tree2array(massTree, branches=wrMassBranch)
SRmassArray = tree2array(massTree, branches=SRmassBranch)
RmassArray = tree2array(massTree, branches=RmassBranch)
correctMassArray = tree2array(massTree, branches=correctMassBranch)
incorrectMassArray = tree2array(massTree, branches=incorrectMassBranch)
leadMassArray = tree2array(massTree, branches=leadMassBranch)
subleadMassArray = tree2array(massTree, branches=subleadMassBranch)
weightArray = tree2array(massTree, branches=weightBranch)
weightArray2 = weightArray * xSec / count2
weightArray = weightArray * xSec / counts
print(WRmassArray.shape)
print(WRmassArray.shape[0])
for i in range(WRmassArray.shape[0]):
WRMass[0] = WRmassArray[i]
resolvedNNMass[0] = RmassArray[i]
superResolvedNNMass[0] = SRmassArray[i]
correctNMass[0] = correctMassArray[i]
treeWeight[0] = weightArray[i]
treeWeight2[0] = weightArray2[i]
incorrectNMass[0] = incorrectMassArray[i]
leadNMass[0] = leadMassArray[i]
subNMass[0] = subleadMassArray[i]
tree.Fill()
# write the tree into the output file and close the file
file.Write()
file.Close()
def test_tree2rec():
chain = TChain('tree')
chain.Add(load('single1.root'))
check_single(rnp.tree2array(chain))
for name in glob.glob(INPUT_FOLDER + '*.root'):
filename = name.replace(INPUT_FOLDER, '')
print('Opening file : ', filename)
if filename.startswith('HToZATo2L2B'): # Signal
print('\t-> Signal')
Sig = True #Signal case
else: # Background
print('\t-> Background')
Sig = False #Background case
f = ROOT.TFile.Open(name)
t = f.Get("tree")
selection = 'met_pt<80 && ll_M>70 && ll_M<110'
jj_M = np.asarray(tree2array(t, branches='jj_M', selection=selection))
lljj_M = np.asarray(tree2array(t, branches='lljj_M', selection=selection))
MEM_TT = np.asarray(
tree2array(t, branches='weight_TT', selection=selection))
MEM_DY = np.asarray(
tree2array(t, branches='weight_DY', selection=selection))
total_weight = np.asarray(
tree2array(t, branches='total_weight', selection=selection))
N = jj_M.shape[0]
if Sig: #Signal
# Extract mA, mH generated from file title
num = [int(s) for s in re.findall('\d+', filename)]
print('\tmH = ', num[2], ', mA = ', num[3])
mH = np.ones(N) * num[2]
mA = np.ones(N) * num[3]
def test_empty_tree():
from array import array
tree = TTree('tree', 'tree')
d = array('d', [0.])
tree.Branch('double', d, 'double/D')
rnp.tree2array(tree)
from __future__ import print_function
from root_numpy import root2array, tree2array
import ROOT
#Get Data
rfile = ROOT.TFile("TreeFile.root")
intree = rfile.Get('Tree')
intree.Print()
# and convert the TTree into an array
array = tree2array(intree, branches=["Gen", "Reco", "data"])
array.dtype.names = ('reco', "gen", "data")
#Plot Data
import matplotlib.pyplot as plt
from matplotlib.colors import LogNorm
import numpy as np
import pandas as pd
df = pd.DataFrame(array)
NBins = 10
xmin = 0
xmax = 1000
content_reco, bin, patches = plt.hist(df['reco'], bins=NBins)
plt.ylabel('# events')
plt.xlabel("reco")
# plt.show()
plt.savefig("plots/reco.pdf")
content_gen, bin, patches = plt.hist(df['gen'], bins=NBins)
print(content_gen)
def efficiency(year):
import numpy as np
from root_numpy import tree2array, fill_hist
from aliases import AK8veto, electronVeto, muonVeto
genPoints = [
1800, 2000, 2500, 3000, 3500, 4000, 4500, 5000, 5500, 6000, 7000, 8000
]
eff = {}
vetoes = {"AK8": AK8veto, "electron": electronVeto, "muon": muonVeto}
VETO = "AK8" ##could change the veto to investigate here
if SEPARATE: eff_add = {}
#channels = ['none', 'qq', 'bq', 'bb', 'mumu']
channels = ['qq', 'bq', 'bb', 'mumu']
for channel in channels:
treeSign = {}
ngenSign = {}
nevtSign = {}
eff[channel] = TGraphErrors()
if SEPARATE:
nevtSign_add = {}
eff_add[channel] = TGraphErrors()
for i, m in enumerate(genPoints):
signName = "ZpBB_M" + str(m)
ngenSign[m] = 0.
nevtSign[m] = 0.
if SEPARATE: nevtSign_add[m] = 0.
for j, ss in enumerate(sample[signName]['files']):
if year == "run2" or year in ss:
sfile = TFile(NTUPLEDIR + ss + ".root", "READ")
ngenSign[m] += sfile.Get("Events").GetBinContent(1)
treeSign[m] = sfile.Get("tree")
if BTAGGING == 'semimedium':
#if SEPARATE:
# temp_array = tree2array(treeSign[m], branches='BTagAK4Weight_deepJet', selection=aliasSM[channel].replace(vetoes[VETO], ""))
#else:
temp_array = tree2array(
treeSign[m],
branches='BTagAK4Weight_deepJet',
selection=aliasSM[channel])
temp_hist = TH1F('pass', 'pass', 1, 0, 1)
fill_hist(temp_hist,
np.zeros(len(temp_array)),
weights=temp_array)
nevtSign[m] += temp_hist.GetBinContent(1)
temp_array = None
temp_hist.Reset()
if SEPARATE:
temp_array = tree2array(
treeSign[m],
branches='BTagAK4Weight_deepJet',
selection=aliasSM[channel].replace(
vetoes[VETO], ""))
temp_hist = TH1F('pass', 'pass', 1, 0, 1)
fill_hist(temp_hist,
np.zeros(len(temp_array)),
weights=temp_array)
nevtSign[m] += temp_hist.GetBinContent(1)
temp_array = None
temp_hist.Reset()
else:
#if SEPARATE:
# temp_array = tree2array(treeSign[m], branches='BTagAK4Weight_deepJet', selection=alias[channel].format(WP=working_points[BTAGGING]).replace(vetoes[VETO], ""))
#else:
temp_array = tree2array(
treeSign[m],
branches='BTagAK4Weight_deepJet',
selection=alias[channel].format(
WP=working_points[BTAGGING]))
temp_hist = TH1F('pass', 'pass', 1, 0, 1)
fill_hist(temp_hist,
np.zeros(len(temp_array)),
weights=temp_array)
nevtSign[m] += temp_hist.GetBinContent(1)
temp_array = None
temp_hist.Reset()
if SEPARATE:
temp_array = tree2array(
treeSign[m],
branches='BTagAK4Weight_deepJet',
selection=alias[channel].format(
WP=working_points[BTAGGING]).replace(
vetoes[VETO], ""))
temp_hist = TH1F('pass', 'pass', 1, 0, 1)
fill_hist(temp_hist,
np.zeros(len(temp_array)),
weights=temp_array)
nevtSign_add[m] += temp_hist.GetBinContent(1)
temp_array = None
temp_hist.Reset()
sfile.Close()
print channel, ss, ":", nevtSign[m], "/", ngenSign[
m], "=", nevtSign[m] / ngenSign[m]
if nevtSign[m] == 0 or ngenSign[m] < 0: continue
n = eff[channel].GetN()
eff[channel].SetPoint(n, m, nevtSign[m] / ngenSign[m])
eff[channel].SetPointError(n, 0,
math.sqrt(nevtSign[m]) / ngenSign[m])
if SEPARATE:
eff_add[channel].SetPoint(n, m, nevtSign_add[m] / ngenSign[m])
eff_add[channel].SetPointError(
n, 0,
math.sqrt(nevtSign_add[m]) / ngenSign[m])
eff[channel].SetMarkerColor(color[channel])
eff[channel].SetMarkerStyle(20)
eff[channel].SetLineColor(color[channel])
eff[channel].SetLineWidth(2)
if SEPARATE:
eff_add[channel].SetMarkerColor(color[channel] +
color_shift[channel])
eff_add[channel].SetMarkerStyle(21)
eff_add[channel].SetLineColor(color[channel] +
color_shift[channel])
eff_add[channel].SetLineWidth(2)
eff_add[channel].SetLineStyle(7)
if channel == 'qq' or channel == 'none': eff[channel].SetLineStyle(3)
n = max([eff[x].GetN() for x in channels])
maxEff = 0.
# Total efficiency
eff["sum"] = TGraphErrors(n)
eff["sum"].SetMarkerStyle(24)
eff["sum"].SetMarkerColor(1)
eff["sum"].SetLineWidth(2)
if SEPARATE:
eff_add["sum"] = TGraphErrors(n)
eff_add["sum"].SetMarkerStyle(25)
eff_add["sum"].SetMarkerColor(1)
eff_add["sum"].SetLineWidth(2)
eff_add["sum"].SetLineStyle(7)
for i in range(n):
tot, mass = 0., 0.
if SEPARATE: tot_add = 0.
for channel in channels:
if channel == 'qq' or channel == 'none':
continue #not sure if I should include 2mu category in sum
if eff[channel].GetN() > i:
tot += eff[channel].GetY()[i]
if SEPARATE: tot_add += eff_add[channel].GetY()[i]
mass = eff[channel].GetX()[i]
if tot > maxEff: maxEff = tot
eff["sum"].SetPoint(i, mass, tot)
if SEPARATE: eff_add["sum"].SetPoint(i, mass, tot_add)
if SEPARATE:
leg = TLegend(0.15, 0.50, 0.95, 0.8)
else:
leg = TLegend(0.15, 0.60, 0.95, 0.8)
leg.SetBorderSize(0)
leg.SetFillStyle(0) #1001
leg.SetFillColor(0)
leg.SetNColumns(len(channels) / 4)
for i, channel in enumerate(channels):
if eff[channel].GetN() > 0:
leg.AddEntry(eff[channel], getChannel(channel), "pl")
if SEPARATE:
leg.AddEntry(eff_add[channel],
getChannel(channel) + " no " + VETO + "-veto",
"pl")
if SEPARATE:
leg.SetY1(leg.GetY2() -
len([x for x in channels if eff[x].GetN() > 0]) * 0.045)
else:
leg.SetY1(leg.GetY2() -
len([x for x in channels if eff[x].GetN() > 0]) / 2. * 0.045)
if SEPARATE:
legS = TLegend(0.5, 0.8 - 0.045, 0.9, 0.85)
else:
legS = TLegend(0.5, 0.85 - 0.045, 0.9, 0.85)
legS.SetBorderSize(0)
legS.SetFillStyle(0) #1001
legS.SetFillColor(0)
legS.AddEntry(eff['sum'],
"Total b tag efficiency (1 b tag + 2 b tag + 2 #mu)", "pl")
if SEPARATE:
legS.AddEntry(eff_add['sum'],
"Total b tag efficiency, no " + VETO + "-veto", "pl")
c1 = TCanvas("c1", "Signal Efficiency", 1200, 800)
c1.cd(1)
eff['sum'].Draw("APL")
if SEPARATE: eff_add['sum'].Draw("SAME, PL")
for i, channel in enumerate(channels):
eff[channel].Draw("SAME, PL")
if SEPARATE: eff_add[channel].Draw("SAME, PL")
leg.Draw()
legS.Draw()
setHistStyle(eff["sum"], 1.1)
eff["sum"].SetTitle(";m_{Z'} (GeV);Acceptance #times efficiency")
eff["sum"].SetMinimum(0.)
eff["sum"].SetMaximum(max(1., maxEff * 1.5)) #0.65
if SEPARATE:
eff_add["sum"].SetTitle(";m_{Z'} (GeV);Acceptance #times efficiency")
eff_add["sum"].SetMinimum(0.)
eff_add["sum"].SetMaximum(1.)
eff["sum"].GetXaxis().SetTitleSize(0.045)
eff["sum"].GetYaxis().SetTitleSize(0.045)
eff["sum"].GetYaxis().SetTitleOffset(1.1)
eff["sum"].GetXaxis().SetTitleOffset(1.05)
eff["sum"].GetXaxis().SetRangeUser(1500, 8000)
c1.SetTopMargin(0.05)
#drawCMS(-1, "Simulation Preliminary", year=year) #Preliminary
#drawCMS(-1, "Work in Progress", year=year, suppressCMS=True)
drawCMS(-1, "", year=year, suppressCMS=True)
drawAnalysis("")
if SEPARATE:
c1.Print("plots/Efficiency/" + year + "_" + BTAGGING + "_no" + VETO +
"veto.pdf")
c1.Print("plots/Efficiency/" + year + "_" + BTAGGING + "_no" + VETO +
"veto.png")
else:
c1.Print("plots/Efficiency/" + year + "_" + BTAGGING + ".pdf")
c1.Print("plots/Efficiency/" + year + "_" + BTAGGING + ".png")
# print
print "category",
for m in range(0, eff["sum"].GetN()):
print " & %d" % int(eff["sum"].GetX()[m]),
print "\\\\", "\n\\hline"
for i, channel in enumerate(channels + ["sum"]):
if channel == 'sum': print "\\hline"
print getChannel(channel).replace("high ", "H").replace(
"low ", "L").replace("purity", "P").replace("b-tag", ""),
for m in range(0, eff[channel].GetN()):
print "& %.1f" % (100. * eff[channel].GetY()[m]),
print "\\\\"
from matplotlib.colors import LogNorm
from random import randint
seed = 10
np.random.seed(seed) #init for reproducibilty
##Input
maxtracks_read = 100 # max number of tracks to read
maxtracks_train = 25 # max number of tracks to use in the training
filename = 'ntuHevjin.root'
file=TFile(filename, 'R')
tree=file.Get('PDsecondTree')
evtcuts = '((evtNumber % 10) < 8)' # leave out 20% of events for testing
vInput=root_numpy.tree2array(tree, branches=['trkPt', 'trkEta', 'trkPhi', 'trkDxy', 'trkDz', 'trkIsInJet', 'trkIsHighPurity', 'trkCharge'], selection=evtcuts)
vInput=root_numpy.rec2array(vInput)
nspec = 2 # numebr of feature not used in the training
nfeat = len(vInput[0]) - nspec
vPt = vInput[:,0]
vEta = vInput[:,1]
vPhi = vInput[:,2]
vDxy = vInput[:,3]
vDz = vInput[:,4]
vtrkIsInJet = vInput[:,-3]
vtrkIsHighPurity = vInput[:,-2]
vQ = vInput[:,-1]
##Shape formatting and zero padding
def test_branch_DNE():
chain = TChain('tree')
chain.Add(load('single1.root'))
rnp.tree2array(chain, branches=['my_net_worth'])
def Tree2Pandas(input_file,
variables,
weight=None,
cut=None,
xsec=None,
event_weight_sum=None,
luminosity=None,
paramFun=None,
tree_name='tree',
start=None,
stop=None,
additional_columns={}):
"""
Convert a ROOT TTree to a pandas DF
"""
variables = copy.copy([
var for var in variables if not var.startswith("$")
]) # Otherwise will add the weight and have a duplicate branch
# Check for repetitions in variables -> makes root_numpy crash #
repeated_var = [
item for item, count in collections.Counter(variables).items()
if count > 1
]
if len(repeated_var) != 0:
logging.critical('There are repeated variables')
for var in repeated_var:
logging.critical('... %s' % var)
raise RuntimeError("Repeated arguments for importing data")
# Get root tree, check if exists first #
if not os.path.exists(input_file):
logging.warning("File %s does not exist" % input_file)
print("File %s does not exist" % input_file)
return None
file_handle = TFile.Open(input_file)
if not file_handle.GetListOfKeys().Contains(tree_name):
#logging.warning("Could not find tree %s in %s"%(tree_name,input_file))
logging.debug("Could not find tree %s in %s" % (tree_name, input_file))
return None
tree = file_handle.Get(tree_name)
N = tree.GetEntries()
logging.debug('\tNumber of events : %d' % N)
# Read the tree and convert it to a numpy structured array
if weight is not None:
variables += [weight]
try:
data = tree2array(tree,
branches=variables,
selection=cut,
start=start,
stop=stop)
except ValueError as e:
logging.error("Issue with file {}".format(input_file))
raise e
# Convert to pandas dataframe #
df = pd.DataFrame(data)
# Reweighting #
relative_weight = 1
if weight is not None and xsec is not None and event_weight_sum is not None:
if luminosity is None:
luminosity = 1
relative_weight = xsec * luminosity / event_weight_sum
logging.debug('\t\tReweighting requested')
logging.debug('\t\t\tCross section : %0.5f' % xsec)
logging.debug('\t\t\tEvent weight sum : %0.2f' % event_weight_sum)
logging.debug('\t\t\tLuminosity : %0.2f' % luminosity)
logging.debug('\t\tRelative weight %0.3e' % relative_weight)
df['cross_section'] = np.ones(df.shape[0]) * xsec
df['luminosity'] = np.ones(df.shape[0]) * luminosity
df['event_weight_sum'] = np.ones(df.shape[0]) * event_weight_sum
else:
df['cross_section'] = np.ones(df.shape[0])
df['luminosity'] = np.ones(df.shape[0])
df['event_weight_sum'] = np.ones(df.shape[0])
if df.shape[0] != 0:
relative_weight /= df.shape[0]
if weight is not None:
df['event_weight'] = df[weight] * relative_weight
else:
df['event_weight'] = np.ones(df.shape[0])
if paramFun is not None:
assert callable(paramFun)
param = paramFun(os.path.basename(input_file))
if param is None:
param = 0
df['param'] = np.ones(df.shape[0]) * param
# Register additional columns #
if len(additional_columns.keys()) != 0:
for key, val in additional_columns.items():
df[key] = pd.Series([val] * df.shape[0])
# Slice printout #
if start is not None or stop is not None:
ni = start if start is not None else 0
nf = stop if stop is not None else N
logging.debug(
"Reading from {} to {} in input tree (over {} entries)".format(
ni, nf, N))
file_handle.Close()
return df
def test_PyROOT():
f = TFile(load('single1.root'))
tree = f.Get('tree')
rnp.tree2array(tree)
weight__background = []
array__signal = []
array__background = []
name__signal = Options['SignalTree'] #wt_DR_nominal wt_DS
name__background = Options['BackgroundTree'] #tt_nominal tt_radHi
#Need to add lines to make sure I can use may files
print name__signal
#for filename in input:
for name in name__signal:
if file.Get(name) != None:
print 'name', name
tree__signal.append(file.Get(name))
event__signal.append(
tree2array(tree__signal[-1], branches=variableList, selection='1'))
weight__signal.append(
tree2array(tree__signal[-1],
branches=[Options['EventWeight']],
selection='1'))
# weight__signal.append(tree2array(tree__signal[-1], branches="EventWeight", selection='1'))
array__signal.append([
list(elem) for elem in zip(event__signal[-1], weight__signal[-1])
])
for name in name__background:
if file.Get(name) != None:
tree__background.append(file.Get(name))
event__background.append(
tree2array(tree__background[-1],
branches=variableList,
selection='1'))
def reweight( sample, puType = 0 ):
if sample.path is None:
print '[puReweighter]: Need to know the MC tree (option --mcTree or sample.path)'
sys.exit(1)
### create a tree with only weights that will be used as friend tree for reweighting different lumi periods
print 'Opening mc file: ', sample.path[0]
fmc = rt.TFile(sample.path[0],'read')
tmc = None
if sample.tnpTree is None:
dirs = fmc.GetListOfKeys()
for d in dirs:
if (d.GetName() == "sampleInfo"): continue
tmc = fmc.Get("%s/fitter_tree" % d.GetName())
else:
tmc = fmc.Get(sample.tnpTree)
#### can reweight vs nVtx but better to reweight v truePU
puMCnVtx = []
puMCrho = []
if puType == 1 :
hmc = rt.TH1F('hMC_nPV' ,'MC nPV' , 75,-0.5,74.5)
tmc.Draw('event_nPV>>hMC_nPV','','goff')
hmc.Scale(1/hmc.Integral())
for ib in range(1,hmc.GetNbinsX()+1):
puMCnVtx.append( hmc.GetBinContent(ib) )
print 'len nvtxMC = ',len(puMCnVtx)
elif puType == 2 :
hmc = rt.TH1F('hMC_rho' ,'MC #rho' , 75,-0.5,74.5)
tmc.Draw('rho>>hMC_rho','','goff')
hmc.Scale(1/hmc.Integral())
for ib in range(1,hmc.GetNbinsX()+1):
puMCrho.append( hmc.GetBinContent(ib) )
print 'len rhoMC = ',len(puMCrho)
puDataDist = {}
puDataArray= {}
weights = {}
epochKeys = puDataEpoch.keys()
if puType == 1 : epochKeys = nVtxDataEpoch.keys()
if puType == 2 : epochKeys = rhoDataEpoch.keys()
for pu in epochKeys:
fpu = None
if puType == 1 : fpu = rt.TFile(nVtxDataEpoch[pu],'read')
elif puType == 2 : fpu = rt.TFile(rhoDataEpoch[pu],'read')
else : fpu = rt.TFile(puDataEpoch[pu],'read')
puDataDist[pu] = fpu.Get('pileup').Clone('puHist_%s' % pu)
puDataDist[pu].Scale(1./puDataDist[pu].Integral())
puDataDist[pu].SetDirectory(0)
puDataArray[pu] = []
for ipu in range(len(puMC[puMCscenario])):
ibin_pu = puDataDist[pu].GetXaxis().FindBin(ipu+0.00001)
puDataArray[pu].append(puDataDist[pu].GetBinContent(ibin_pu))
print 'puData[%s] length = %d' % (pu,len(puDataArray[pu]))
fpu.Close()
weights[pu] = []
mcEvts = tree2array( tmc, branches = ['weight','truePU','event_nPV','rho'] )
pumc = puMC[puMCscenario]
if puType == 1: pumc = puMCnVtx
elif puType == 2: pumc = puMCrho
else : pumc = puMC[puMCscenario]
puMax = len(pumc)
print '-> nEvtsTot ', len(mcEvts)
for ievt in xrange(len(mcEvts)):
if ievt%1000000 == 0 : print 'iEvt:',ievt
evt = mcEvts[ievt]
for pu in epochKeys:
pum = -1
pud = -1
if puType == 1 and evt['event_nPV'] < puMax:
pud = puDataArray[pu][evt['event_nPV']]
pum = pumc[evt['event_nPV']]
if puType == 2 and int(evt['rho']) < puMax:
pud = puDataArray[pu][int(evt['rho'])]
pum = pumc[int(evt['rho'])]
elif puType == 0:
pud = puDataArray[pu][evt['truePU']]
pum = pumc[evt['truePU']]
puw = 1
if pum > 0:
puw = pud/pum
if evt['weight'] > 0 : totw = +puw
else : totw = -puw
weights[pu].append( ( puw,totw) )
newFile = rt.TFile( sample.puTree, 'recreate')
for pu in epochKeys:
treeWeight = rt.TTree('weights_%s'%pu,'tree with weights')
wpuarray = np.array(weights[pu],dtype=[('PUweight',float),('totWeight',float)])
array2tree( wpuarray, tree = treeWeight )
treeWeight.Write()
newFile.Close()
fmc.Close()
from rootpy.io import root_open
from root_numpy import tree2array
import numpy as np
import math as m
import pprint
from shapely.geometry import Polygon
import waferGeometry
f = root_open("test_triggergeom.root")
cells_tree = f.Get("hgcaltriggergeomtester/TreeCells")
TC_tree = f.Get("hgcaltriggergeomtester/TreeTriggerCells")
cells_wafer_info = tree2array(cells_tree,
branches=[
'id', 'wafertype', 'wafer', 'layer',
'subdet', 'zside', 'x', 'y', 'cell',
'waferrow', 'wafercolumn'
])
cells_tc_info = tree2array(TC_tree,
branches=[
'triggercell', 'c_id', 'c_cell', 'wafer',
'layer', 'subdet', 'zside'
])
f.close()
#########################################################
## FUNCTIONS ##
def ExtractMappingCoordinates(koordx, koordy, d, x0, y0):
distx = koordx - x0
