def plotPulls(optunf="Bayes",
ntest=10,
leff=True,
loufl=False,
optfun="exp",
opttfun="",
gmean=-1.0,
nrebin=4):
if opttfun == "":
opttfun = optfun
if optfun == "blobel":
gmean = 0.0
funttxt, funtxt = funtxts(opttfun, optfun, leff, loufl)
global histos, canv, canv2
histos = []
canv = TCanvas("canv", "thruth vs reco pulls", 600, 800)
canv.Divide(1, 3)
canv2 = TCanvas("canv2", "P(chi^2)", 600, 800)
canv2.Divide(2, 3)
if optunf == "BasisSplines":
bininfo = BinInfo(nrebin)
unfoldtester = UnfoldTester(optunf, nrebin)
else:
bininfo = BinInfo()
unfoldtester = UnfoldTester(optunf)
trainer = Trainer(bininfo, opttfun, optfun)
tester = Tester(bininfo, optfun)
hbininfo = bininfo.create(optfun)
dx = hbininfo["mhi"]
for sigma, ipad in [[0.01 * dx, 1], [0.03 * dx, 2], [0.1 * dx, 3]]:
txt = optunf + ", smear mu, s.d.= " + str(gmean) + ", " + str(
sigma) + ", train: " + funttxt + ", test: " + funtxt + ", " + str(
ntest) + " tests"
hPulls = TProfile("pulls", txt, hbininfo["tbins"], hbininfo["tlo"],
hbininfo["thi"])
hPulls.SetErrorOption("s")
hPulls.SetYTitle("Thruth reco pull")
histos.append(hPulls)
hChisq = TH1D("chisq", "P(chi^2) rec " + txt, 10, 0.0, 1.0)
hChisqm = TH1D("chisqm", "P(chi^2) mea " + txt, 10, 0.0, 1.0)
histos.append(hChisq)
histos.append(hChisqm)
measurement = createMeasurement(gmean, sigma, leff, optfun)
response = trainer.train(measurement, loufl=loufl)
for itest in range(ntest):
print "Test", itest
unfold, hTrue, hMeas = unfoldtester.rununfoldtest(
tester, measurement, response)
unfold.PrintTable(cout, hTrue, 2)
hReco = unfold.Hreco(2)
nbin = hReco.GetNbinsX()
if hbininfo["nrebin"] > 1:
hTrue = hTrue.Rebin(nrebin)
for ibin in range(nbin + 1):
truevalue = hTrue.GetBinContent(ibin)
recvalue = hReco.GetBinContent(ibin)
error = hReco.GetBinError(ibin)
if error > 0.0:
pull = (recvalue - truevalue) / error
hPulls.Fill(hReco.GetBinCenter(ibin), pull)
chisq = unfold.Chi2(hTrue, 2)
hChisq.Fill(TMath.Prob(chisq, hTrue.GetNbinsX()))
chisqm = unfold.Chi2measured()
pchisqm = TMath.Prob(chisqm, hMeas.GetNbinsX() - hReco.GetNbinsX())
print "Chisq measured=", chisqm, "P(chi^2)=", pchisqm
hChisqm.Fill(pchisqm)
canv.cd(ipad)
gStyle.SetErrorX(0)
hPulls.SetMinimum(-3.0)
hPulls.SetMaximum(3.0)
hPulls.SetMarkerSize(1.0)
hPulls.SetMarkerStyle(20)
hPulls.SetStats(False)
hPulls.Draw()
canv2.cd(ipad * 2 - 1)
hChisq.Draw()
canv2.cd(ipad * 2)
hChisqm.Draw()
fname = "RooUnfoldTestPulls_" + optunf + "_" + opttfun + "_" + optfun
if loufl:
fname += "_oufl"
canv.Print(fname + ".pdf")
fname = "RooUnfoldTestChisq_" + optunf + "_" + opttfun + "_" + optfun
if loufl:
fname += "_oufl"
canv2.Print(fname + ".pdf")
return
def dqm_getSingleHist_json(server, run, dataset, hist, rootContent=False):
postfix = "?rootcontent=1" if rootContent else ""
datareq = urllib2.Request(('%s/jsonfairy/archive/%s/%s/%s%s') % (server, run, dataset, hist, postfix))
datareq.add_header('User-agent', ident)
# Get data
data = eval(re.sub(r"\bnan\b", "0", urllib2.build_opener(X509CertOpen()).open(datareq).read()),
{ "__builtins__": None }, {})
histo = data['hist']
# Now convert into real ROOT histogram object
if 'TH1' in histo['type']:
# The following assumes a TH1F object
contents = histo['bins']['content']
nbins = len(contents)
xmin = histo['xaxis']['first']['value']
xmax = histo['xaxis']['last']['value']
roothist = TH1F(histo['stats']['name'],histo['title'],nbins,xmin,xmax)
for xx in range(1,nbins+1):
roothist.SetBinContent(xx, contents[xx-1])
roothist.SetBinError(xx, histo['bins']['error'][xx-1])
roothist.SetEntries(histo['stats']['entries'])
stats=array('d')
stats.append(histo['stats']['entries'])
stats.append(histo['stats']['entries'])
stats.append(histo['stats']['entries']*histo['stats']['mean']['X']['value'])
stats.append((histo['stats']['rms']['X']['value']*histo['stats']['rms']['X']['value']+histo['stats']['mean']['X']['value']*histo['stats']['mean']['X']['value'])*histo['stats']['entries'])
roothist.PutStats(stats)
elif(histo['type']=='TProfile'):
contents = histo['bins']['content']
nbins = len(contents)
xmin = histo['xaxis']['first']['value']
xmax = histo['xaxis']['last']['value']
roothist = TProfile(histo['stats']['name'],histo['title'],nbins,xmin,xmax)
roothist.SetErrorOption("g")
for xx in range(0,nbins):
if(histo['bins']['error'][xx]!=0):
ww=1./(histo['bins']['error'][xx]*histo['bins']['error'][xx])
else:
ww=0.
roothist.Fill(xmin+(2*xx+1)*((xmax-xmin)/(nbins*2.0)), contents[xx],ww)
# roothist.SetBinContent(xx, contents[xx-1])
# roothist.SetBinError(xx, histo['bins']['error'][xx-1])
roothist.SetEntries(histo['stats']['entries'])
stats=array('d')
for i in range(0,6):
stats.append(i)
roothist.GetStats(stats)
stats[0]=(histo['stats']['entries'])
stats[1]=(histo['stats']['entries'])
stats[2]=(histo['stats']['entries']*histo['stats']['mean']['X']['value'])
stats[3]=((histo['stats']['rms']['X']['value']*histo['stats']['rms']['X']['value']+histo['stats']['mean']['X']['value']*histo['stats']['mean']['X']['value'])*histo['stats']['entries'])
roothist.PutStats(stats)
elif 'TH2' in histo['type']:
contents = histo['bins']['content']
nbinsx = histo['xaxis']['last']['id']
xmin = histo['xaxis']['first']['value']
xmax = histo['xaxis']['last']['value']
nbinsy = histo['yaxis']['last']['id']
ymin = histo['yaxis']['first']['value']
ymax = histo['yaxis']['last']['value']
roothist = TH2F(histo['stats']['name'],histo['title'],nbinsx,xmin,xmax,nbinsy,ymin,ymax)
for xx in range(1,nbinsx+1):
for yy in range(1,nbinsy+1):
roothist.SetBinContent(xx,yy, contents[yy-1][xx-1])
roothist.SetEntries(histo['stats']['entries'])
stats=array('d')
stats.append(histo['stats']['entries'])
stats.append(histo['stats']['entries'])
stats.append(histo['stats']['entries']*histo['stats']['mean']['X']['value'])
stats.append((histo['stats']['rms']['X']['value']*histo['stats']['rms']['X']['value']+histo['stats']['mean']['X']['value']*histo['stats']['mean']['X']['value'])*histo['stats']['entries'])
stats.append(histo['stats']['entries']*histo['stats']['mean']['Y']['value'])
stats.append((histo['stats']['rms']['Y']['value']*histo['stats']['rms']['Y']['value']+histo['stats']['mean']['Y']['value']*histo['stats']['mean']['Y']['value'])*histo['stats']['entries'])
roothist.PutStats(stats)
elif(histo['type']=='TProfile2D'):
contents = histo['bins']['content']
nbinsx = histo['xaxis']['last']['id']
xmin = histo['xaxis']['first']['value']
xmax = histo['xaxis']['last']['value']
nbinsy = histo['yaxis']['last']['id']
ymin = histo['yaxis']['first']['value']
ymax = histo['yaxis']['last']['value']
roothist = TProfile2D(histo['stats']['name'],histo['title'],nbinsx,xmin,xmax,nbinsy,ymin,ymax)
for xx in range(0,nbinsx):
for yy in range(0,nbinsy):
roothist.Fill(xmin+(2*xx+1)*((xmax-xmin)/(nbinsx*2.0)),ymin+(2*yy+1)*((ymax-ymin)/(nbinsy*2.0)),0,1)
for xx in range(1,nbinsx+1):
for yy in range(1,nbinsy+1):
roothist.SetBinContent(xx,yy, contents[yy-1][xx-1])
roothist.SetEntries(histo['stats']['entries'])
return roothist
def main(network_name, do_weights, general_weights, electron_weights,
muon_weights, data_files, no_true_comp, nbins, veto_all_jets,
require_jets, WorkingP_List, Independant_Variable_List,
Histogram_Variable_List, Profile_List, TH2D_List):
TH1.SetDefaultSumw2()
## First we check that you havent vetoed and requested jets
if veto_all_jets and require_jets:
print("\n\n\nYOU CANT ASK FOR NONE YET SOME JETS\n\n\n")
return 0
for data_dir, data_set_name in data_files:
OUTPUT_dir = os.path.join(os.environ["HOME_DIRECTORY"], "Output")
DATA_dir = os.path.join(os.environ["HOME_DIRECTORY"], "Data")
data_base_dir = os.path.join(DATA_dir, data_dir)
print("\n\n" + data_set_name)
"""
____ ____ _____ ____ _ ____ _ _____ ___ ___ _ _
| _ \ | _ \ | ____| | _ \ / \ | _ \ / \ |_ _| |_ _| / _ \ | \ | |
| |_) | | |_) | | _| | |_) | / _ \ | |_) | / _ \ | | | | | | | | | \| |
| __/ | _ < | |___ | __/ / ___ \ | _ < / ___ \ | | | | | |_| | | |\ |
|_| |_| \_\ |_____| |_| /_/ \_\ |_| \_\ /_/ \_\ |_| |___| \___/ |_| \_|
"""
## The list of matrices
Indp_output = [None for var in Independant_Variable_List]
TH1D_output = [[None for wp in WorkingP_List]
for var in Histogram_Variable_List]
Tail_output = [[None for wp in WorkingP_List]
for var in Histogram_Variable_List]
Prof_output = [[None for wp in WorkingP_List] for var in Profile_List]
TH2D_output = [[None for wp in WorkingP_List] for var in TH2D_List]
## Open the root file
root_file = TFile.Open(os.path.join(data_base_dir, data_set_name))
## Loading the tree containing the working point information
wpt_tree = root_file.Get("wpt_tree")
## Making the wpt_tree and the other trees friends so that they can be compared
wpt_tree.AddFriend(network_name, root_file)
wpt_tree.AddFriend("alt_tree", root_file)
wpt_tree.AddFriend("var_tree", root_file)
## Creating a string of all the weights to be applied
event_weights = []
if "SR" in data_dir: event_weights += general_weights
if "ee" in data_dir: event_weights += electron_weights
if "mumu" in data_dir: event_weights += muon_weights
if veto_all_jets:
event_weights += ["(var_tree.Jets_Loose_SumET==0)"]
if require_jets:
event_weights += ["(var_tree.Jets_Loose_SumET>0)"]
if do_weights: weight_string = " * ".join(event_weights)
else: weight_string = ""
if len(weight_string):
print("Events are weighted using: {}".format(weight_string))
## The strings to call up the Truth Values
if no_true_comp:
True_Et = "0"
True_Ex = "0"
True_Ey = "0"
True_Phi = "0"
else:
True_Et = "WP_Truth_ET"
True_Ex = "WP_Truth_X"
True_Ey = "WP_Truth_Y"
True_Phi = "WP_Truth_Phi"
"""
____ ____ _ __ __ ___ _ _ ____
| _ \ | _ \ / \ \ \ / / |_ _| | \ | | / ___|
| | | | | |_) | / _ \ \ \ /\ / / | | | \| | | | _
| |_| | | _ < / ___ \ \ V V / | | | |\ | | |_| |
|____/ |_| \_\ /_/ \_\ \_/\_/ |___| |_| \_| \____|
"""
## Before the workingpoint loop, the independant variables are drawn
for v, var in enumerate(Independant_Variable_List):
print(" -- {}".format(var.name))
## Creating the histogram which will be filled
hist_name = var.name
myhist = TH1D(hist_name, hist_name, var.nbins, var.xmin, var.xmax)
myhist.SetStats(True)
myhist.StatOverflows(True)
## Get Maths Function from variable
maths_string = var.tree + "." + var.branch
## Drawing the tree and saving the hist to the matrix
execution = "{}>>{}".format(maths_string, hist_name)
wpt_tree.Draw(execution, weight_string, "goff")
## Saving the Histogram to the Matrix
myhist.SetDirectory(0)
Indp_output[v] = myhist
## First we select the working point and create the correct strings
for w, wp in enumerate(WorkingP_List):
print(" -- {}:".format(wp.name))
Rec_Et = wp.Et
Rec_Ex = wp.Ex
Rec_Ey = wp.Ey
Rec_Phi = wp.Phi
if wp.tree == "ann_tree":
Rec_Et = network_name + "." + Rec_Et
Rec_Ex = network_name + "." + Rec_Ex
Rec_Ey = network_name + "." + Rec_Ey
rec_and_truth_vars = [
Rec_Et, Rec_Ex, Rec_Ey, Rec_Phi, True_Et, True_Ex, True_Ey,
True_Phi
]
## Drawing the 1D histograms
for v, var in enumerate(Histogram_Variable_List):
print(" -- -- {}".format(var.name))
## Creating the histogram which will be filled
hist_name = "{}_{}".format(var.name, wp.name)
myhist = TH1D(hist_name, hist_name, var.nbins, var.xmin,
var.xmax)
myhist.SetStats(True)
myhist.StatOverflows(True)
## Individual special plots
if var.name == "XY":
## Plot the X histogram
maths_string = Evaluation.TH1D_Maths_String(
"X", *rec_and_truth_vars)
execution = "{} >> {}".format(maths_string, hist_name)
wpt_tree.Draw(execution, weight_string, "goff")
## Add the y histogram
maths_string = Evaluation.TH1D_Maths_String(
"Y", *rec_and_truth_vars)
execution = "{} >> +{}".format(maths_string, hist_name)
wpt_tree.Draw(execution, weight_string, "goff")
else:
## Get Maths Function from variable
maths_string = Evaluation.TH1D_Maths_String(
var.name, *rec_and_truth_vars)
## Drawing the tree and saving the hist to the matrix
execution = "{} >> {}".format(maths_string, hist_name)
wpt_tree.Draw(execution, weight_string, "goff")
## Saving the Histogram to the Matrix
myhist.SetDirectory(0)
TH1D_output[v][w] = myhist
## Drawing the Profiles
for v, (vx, vy) in enumerate(Profile_List):
print(" -- -- {} vs {}".format(vx.name, vy.name))
## Creating the profile which will be filled
hist_name = "{}_vs_{}_{}".format(vx.name, vy.name, wp.name)
myhist = TProfile(hist_name, hist_name, vx.nbins, vx.xmin,
vx.xmax)
if vy.reso: myhist.SetErrorOption('s')
## The x variable is called from its branch in the correct tree
x_string = vx.tree + "." + vx.branch
## Individual special plots
if vy.name == "XY":
y_string = Evaluation.TH1D_Maths_String(
"X", *rec_and_truth_vars)
execution = "{}:{} >> {}".format(y_string, x_string,
hist_name)
wpt_tree.Draw(execution, weight_string, "goff prof")
y_string = Evaluation.TH1D_Maths_String(
"Y", *rec_and_truth_vars)
execution = "{}:{} >>+{}".format(y_string, x_string,
hist_name)
wpt_tree.Draw(execution, weight_string, "goff prof")
elif vy.name == "AZ":
z_x = "(alt_tree.ll_px)"
z_y = "(alt_tree.ll_py)"
z_pt = "(alt_tree.ll_pt)"
x_string = z_pt
y_string = "({rx}*{zx} + {ry}*{zy})/{zpt}".format(
rx=Rec_Ex, ry=Rec_Ey, zx=z_x, zy=z_y, zpt=z_pt)
execution = "{}:{} >> {}".format(y_string, x_string,
hist_name)
wpt_tree.Draw(execution, weight_string, "goff prof")
else:
y_string = Evaluation.TH1D_Maths_String(
vy.name, *rec_and_truth_vars)
execution = "{}:{} >> {}".format(y_string, x_string,
hist_name)
wpt_tree.Draw(execution, weight_string, "goff prof")
## Saving the Histogram to the Matrix
myhist.SetDirectory(0)
Prof_output[v][w] = myhist
if wp.name not in ["Network", "Tight"]:
continue
## Drawing the TH2Ds
for v, (vx, vy) in enumerate(TH2D_List):
print(" -- -- {} vs {}".format(vx.name, vy.name))
## Creating the profile which will be filled
hist_name = "2D_{}_vs_{}_{}".format(vx.name, vy.name, wp.name)
myhist = TH2D(hist_name, hist_name, vx.nbins2d, vx.xmin2d,
vx.xmax2d, vy.nbins2d, vy.xmin2d, vy.xmax2d)
x_string = Evaluation.TH1D_Maths_String(
vx.name, *rec_and_truth_vars)
y_string = Evaluation.TH1D_Maths_String(
vy.name, *rec_and_truth_vars)
## The x variable is called from its branch in the correct tree
if x_string is None:
x_string = vx.tree + "." + vx.branch
execution = "{}:{} >> {}".format(y_string, x_string, hist_name)
wpt_tree.Draw(execution, weight_string, "goff")
## Saving the Histogram to the Matrix
myhist.SetDirectory(0)
TH2D_output[v][w] = myhist
root_file.Close()
"""
_____ ____ ___ _____ ___ _ _ ____
| ____| | _ \ |_ _| |_ _| |_ _| | \ | | / ___|
| _| | | | | | | | | | | | \| | | | _
| |___ | |_| | | | | | | | | |\ | | |_| |
|_____| |____/ |___| |_| |___| |_| \_| \____|
"""
## We now go through the 1D histograms and make them include overflow and underflow
for v, var in enumerate(Histogram_Variable_List):
for w, wp in enumerate(WorkingP_List):
## Include the overflow
last_bin = TH1D_output[v][w].GetBinContent(nbins)
overflow = TH1D_output[v][w].GetBinContent(nbins + 1)
TH1D_output[v][w].SetBinContent(nbins, last_bin + overflow)
TH1D_output[v][w].SetBinContent(nbins + 1, 0)
## Include the underflow
first_bin = TH1D_output[v][w].GetBinContent(1)
underflow = TH1D_output[v][w].GetBinContent(0)
TH1D_output[v][w].SetBinContent(1, first_bin + underflow)
TH1D_output[v][w].SetBinContent(0, 0)
## We create a tail distrobution if it is requested
if var.tail:
tail_temp = TH1D_output[v][w].GetCumulative(False)
tail_temp.Scale(1 / tail_temp.GetBinContent(1))
Tail_output[v][w] = tail_temp
## We go through the resolution profiles and replace their entries with the RMSE for each bin
for v, (vx, vy) in enumerate(Profile_List):
if not vy.reso:
continue
for w, wp in enumerate(WorkingP_List):
old_prof = Prof_output[v][w]
bin_width = old_prof.GetBinWidth(1)
name = "{}_vs_{}_{}_res".format(vx.name, vy.name, wp.name)
new_prof = TGraphErrors(vx.nbins)
new_prof.SetName(name)
new_prof.SetTitle(name)
new_prof.SetLineWidth(2)
for b_idx in range(vx.nbins):
new_prof.SetPoint(b_idx, old_prof.GetBinCenter(b_idx + 1),
old_prof.GetBinError(b_idx + 1))
new_prof.SetPointError(b_idx, bin_width / 2, 0)
Prof_output[v][w] = new_prof
"""
____ _ __ __ ___ _ _ ____
/ ___| / \ \ \ / / |_ _| | \ | | / ___|
\___ \ / _ \ \ \ / / | | | \| | | | _
___) | / ___ \ \ V / | | | |\ | | |_| |
|____/ /_/ \_\ \_/ |___| |_| \_| \____|
"""
## Creating the output directory
output_dir = os.path.join(OUTPUT_dir, network_name, data_set_name[:-5])
if veto_all_jets:
output_dir = output_dir + "_NOJETS"
if require_jets:
output_dir = output_dir + "_SOMEJETS"
## Check that the file can be saved
if not os.path.exists(output_dir):
os.system("mkdir -p " + output_dir)
## We create an output file for the histograms
output_file = TFile(os.path.join(output_dir, "histograms.root"),
"update")
gFile = output_file
## We save the independants
for v, var in enumerate(Independant_Variable_List):
Indp_output[v].Write("", TObject.kOverwrite)
## We save the TH1Ds and tails
for v, var in enumerate(Histogram_Variable_List):
for w in range(len(WorkingP_List)):
TH1D_output[v][w].Write("", TObject.kOverwrite)
if var.tail:
Tail_output[v][w].Write("", TObject.kOverwrite)
## We save the profiles
for v in range(len(Profile_List)):
for w in range(len(WorkingP_List)):
Prof_output[v][w].Write("", TObject.kOverwrite)
## We save the TH2Ds
for v in range(len(TH2D_List)):
for w in range(len(WorkingP_List[:2])):
TH2D_output[v][w].Write("", TObject.kOverwrite)
output_file.Close()
return 0