def analyze_data_function(data, parameters): ret = Results() num_events = data["num_events"] muons = data["Muon"] mu_pt = nplib.sqrt(muons.Px**2 + muons.Py**2) muons.attrs_data["pt"] = mu_pt mask_events = nplib.ones(muons.numevents(), dtype=nplib.bool) mask_muons_passing_pt = muons.pt > parameters["muons_ptcut"] num_muons_event = kernels.sum_in_offsets(backend, muons.offsets, mask_muons_passing_pt, mask_events, muons.masks["all"], nplib.int8) mask_events_dimuon = num_muons_event == 2 #get the leading muon pt in events that have exactly two muons inds = nplib.zeros(num_events, dtype=nplib.int32) leading_muon_pt = kernels.get_in_offsets(backend, muons.offsets, muons.pt, inds, mask_events_dimuon, mask_muons_passing_pt) #compute a weighted histogram weights = nplib.ones(num_events, dtype=nplib.float32) bins = nplib.linspace(0, 300, 101, dtype=nplib.float32) hist_muons_pt = Histogram(*kernels.histogram_from_vector( backend, leading_muon_pt[mask_events_dimuon], weights[mask_events_dimuon], bins)) #save it to the output ret["hist_leading_muon_pt"] = hist_muons_pt return ret
def load_hist(hist_dict): return Histogram.from_dict({ "edges": np.array(hist_dict["edges"]), "contents": np.array(hist_dict["contents"]), "contents_w2": np.array(hist_dict["contents_w2"]), })
def fill_histograms_several(hists, systematic_name, histname_prefix, variables, mask, weights, use_cuda): this_worker = get_worker_wrapper() NUMPY_LIB = this_worker.NUMPY_LIB backend = this_worker.backend all_arrays = [] all_bins = [] num_histograms = len(variables) for array, varname, bins in variables: if (len(array) != len(variables[0][0]) or len(array) != len(mask) or len(array) != len(weights["nominal"])): raise Exception( "Data array {0} is of incompatible size".format(varname)) all_arrays += [array] all_bins += [bins] max_bins = max([b.shape[0] for b in all_bins]) stacked_array = NUMPY_LIB.stack(all_arrays, axis=0) stacked_bins = np.concatenate(all_bins) nbins = np.array([len(b) for b in all_bins]) nbins_sum = np.cumsum(nbins) nbins_sum = np.insert(nbins_sum, 0, [0]) for weight_name, weight_array in weights.items(): if use_cuda: nblocks = 32 out_w = NUMPY_LIB.zeros((len(variables), nblocks, max_bins), dtype=NUMPY_LIB.float32) out_w2 = NUMPY_LIB.zeros((len(variables), nblocks, max_bins), dtype=NUMPY_LIB.float32) backend.fill_histogram_several[nblocks, 1024]( stacked_array, weight_array, mask, stacked_bins, NUMPY_LIB.array(nbins), NUMPY_LIB.array(nbins_sum), out_w, out_w2, ) cuda.synchronize() out_w = out_w.sum(axis=1) out_w2 = out_w2.sum(axis=1) out_w = NUMPY_LIB.asnumpy(out_w) out_w2 = NUMPY_LIB.asnumpy(out_w2) else: out_w = NUMPY_LIB.zeros((len(variables), max_bins), dtype=NUMPY_LIB.float32) out_w2 = NUMPY_LIB.zeros((len(variables), max_bins), dtype=NUMPY_LIB.float32) backend.fill_histogram_several( stacked_array, weight_array, mask, stacked_bins, nbins, nbins_sum, out_w, out_w2, ) out_w_separated = [ out_w[i, 0:nbins[i] - 1] for i in range(num_histograms) ] out_w2_separated = [ out_w2[i, 0:nbins[i] - 1] for i in range(num_histograms) ] for ihist in range(num_histograms): hist_name = histname_prefix + variables[ihist][1] bins = variables[ihist][2] target_histogram = Histogram(out_w_separated[ihist], out_w2_separated[ihist], bins) target = {weight_name: target_histogram} update_histograms_systematic(hists, hist_name, systematic_name, target)
def analyze_data(data, sample, NUMPY_LIB=None, parameters={}, samples_info={}, is_mc=True, lumimask=None, cat=False, DNN=False, DNN_model=None, jets_met_corrected=True): #Output structure that will be returned and added up among the files. #Should be relatively small. ret = Results() muons = data["Muon"] electrons = data["Electron"] scalars = data["eventvars"] jets = data["Jet"] nEvents = muons.numevents() indices = {} indices["leading"] = NUMPY_LIB.zeros(nEvents, dtype=NUMPY_LIB.int32) indices["subleading"] = NUMPY_LIB.ones(nEvents, dtype=NUMPY_LIB.int32) mask_events = NUMPY_LIB.ones(nEvents, dtype=NUMPY_LIB.bool) # apply event cleaning and PV selection flags = [ "Flag_goodVertices", "Flag_globalSuperTightHalo2016Filter", "Flag_HBHENoiseFilter", "Flag_HBHENoiseIsoFilter", "Flag_EcalDeadCellTriggerPrimitiveFilter", "Flag_BadPFMuonFilter", "Flag_BadChargedCandidateFilter", "Flag_ecalBadCalibFilter" ] if not is_mc: flags.append("Flag_eeBadScFilter") for flag in flags: mask_events = mask_events & scalars[flag] mask_events = mask_events & (scalars["PV_npvsGood"] > 0) #mask_events = vertex_selection(scalars, mask_events) # apply object selection for muons, electrons, jets good_muons, veto_muons = lepton_selection(muons, parameters["muons"]) good_electrons, veto_electrons = lepton_selection(electrons, parameters["electrons"]) good_jets = jet_selection(jets, muons, (veto_muons | good_muons), parameters["jets"], jets_met_corrected) & jet_selection( jets, electrons, (veto_electrons | good_electrons), parameters["jets"], jets_met_corrected) bjets = good_jets & ( getattr(jets, parameters["btagging algorithm"]) > parameters["btagging WP"][parameters["btagging algorithm"]]) # apply basic event selection -> individual categories cut later nleps = NUMPY_LIB.add( ha.sum_in_offsets(muons, good_muons, mask_events, muons.masks["all"], NUMPY_LIB.int8), ha.sum_in_offsets(electrons, good_electrons, mask_events, electrons.masks["all"], NUMPY_LIB.int8)) nMuons = ha.sum_in_offsets(muons, good_muons, mask_events, muons.masks["all"], NUMPY_LIB.int8) nElectrons = ha.sum_in_offsets(electrons, good_electrons, mask_events, electrons.masks["all"], NUMPY_LIB.int8) lepton_veto = NUMPY_LIB.add( ha.sum_in_offsets(muons, veto_muons, mask_events, muons.masks["all"], NUMPY_LIB.int8), ha.sum_in_offsets(electrons, veto_electrons, mask_events, electrons.masks["all"], NUMPY_LIB.int8)) njets = ha.sum_in_offsets(jets, good_jets, mask_events, jets.masks["all"], NUMPY_LIB.int8) btags = ha.sum_in_offsets(jets, bjets, mask_events, jets.masks["all"], NUMPY_LIB.int8) if jets_met_corrected: #met = (scalars["MET_pt_nom"] > 20) met = (scalars["METFixEE2017_pt_nom"] > 20) else: met = (scalars["MET_pt"] > 20) # trigger logic # needs update for different years! trigger_el = (scalars["HLT_Ele35_WPTight_Gsf"] | scalars["HLT_Ele28_eta2p1_WPTight_Gsf_HT150"]) & ( nleps == 1) & (nElectrons == 1) trigger_mu = (scalars["HLT_IsoMu27"]) & (nleps == 1) & (nMuons == 1) if not is_mc: if "SingleMuon" in sample: trigger_el = NUMPY_LIB.zeros(nEvents, dtype=NUMPY_LIB.bool) if "SingleElectron" in sample: trigger_mu = NUMPY_LIB.zeros(nEvents, dtype=NUMPY_LIB.bool) mask_events = mask_events & (trigger_el | trigger_mu) mask_events = mask_events & (nleps == 1) & (lepton_veto == 0) & ( njets >= 4) & (btags >= 2) & met ### calculation of all needed variables var = {} var["njets"] = njets var["btags"] = btags var["nleps"] = nleps if jets_met_corrected: pt_label = "pt_nom" else: pt_label = "pt" variables = [ ("jet", jets, good_jets, "leading", [pt_label, "eta"]), ("bjet", jets, bjets, "leading", [pt_label, "eta"]), ] # special role of lepton var["leading_lepton_pt"] = NUMPY_LIB.maximum( ha.get_in_offsets(muons.pt, muons.offsets, indices["leading"], mask_events, good_muons), ha.get_in_offsets(electrons.pt, electrons.offsets, indices["leading"], mask_events, good_electrons)) var["leading_lepton_eta"] = NUMPY_LIB.maximum( ha.get_in_offsets(muons.eta, muons.offsets, indices["leading"], mask_events, good_muons), ha.get_in_offsets(electrons.eta, electrons.offsets, indices["leading"], mask_events, good_electrons)) # all other variables for v in variables: calculate_variable_features(v, mask_events, indices, var) #synch #mask = (scalars["event"] == 2895765) # calculate weights for MC samples weights = {} weights["nominal"] = NUMPY_LIB.ones(nEvents, dtype=NUMPY_LIB.float32) if is_mc: weights["nominal"] = weights["nominal"] * scalars[ "genWeight"] * parameters["lumi"] * samples_info[sample][ "XS"] / samples_info[sample]["ngen_weight"] # pu corrections #pu_weights = compute_pu_weights(parameters["pu_corrections_target"], weights["nominal"], scalars["Pileup_nTrueInt"], scalars["PV_npvsGood"]) pu_weights = compute_pu_weights(parameters["pu_corrections_target"], weights["nominal"], scalars["Pileup_nTrueInt"], scalars["Pileup_nTrueInt"]) weights["nominal"] = weights["nominal"] * pu_weights var["pu_weights"] = pu_weights # lepton SF corrections electron_weights = compute_lepton_weights( electrons, (electrons.deltaEtaSC + electrons.eta), electrons.pt, mask_events, good_electrons, evaluator, ["el_triggerSF", "el_recoSF", "el_idSF"]) muon_weights = compute_lepton_weights( muons, muons.pt, NUMPY_LIB.abs(muons.eta), mask_events, good_muons, evaluator, ["mu_triggerSF", "mu_isoSF", "mu_idSF"]) weights[ "nominal"] = weights["nominal"] * muon_weights * electron_weights # btag SF corrections btag_weights = compute_btag_weights(jets, mask_events, good_jets, parameters["btag_SF_target"], jets_met_corrected, parameters["btagging algorithm"]) var["btag_weights"] = btag_weights weights["nominal"] = weights["nominal"] * btag_weights #in case of data: check if event is in golden lumi file if not is_mc and not (lumimask is None): mask_lumi = lumimask(scalars["run"], scalars["luminosityBlock"]) mask_events = mask_events & mask_lumi #evaluate DNN if DNN: DNN_pred = evaluate_DNN(jets, good_jets, electrons, good_electrons, muons, good_muons, scalars, mask_events, nEvents, DNN, DNN_model) # in case of tt+jets -> split in ttbb, tt2b, ttb, ttcc, ttlf processes = {} if sample.startswith("TTTo"): #Changed for TTV samples ttCls = scalars["genTtbarId"] % 100 processes["ttbb"] = mask_events & (ttCls >= 53) & (ttCls <= 56) processes["tt2b"] = mask_events & (ttCls == 52) processes["ttb"] = mask_events & (ttCls == 51) processes["ttcc"] = mask_events & (ttCls >= 41) & (ttCls <= 45) ttHF = ((ttCls >= 53) & (ttCls <= 56)) | (ttCls == 52) | (ttCls == 51) | ( (ttCls >= 41) & (ttCls <= 45)) processes["ttlf"] = mask_events & NUMPY_LIB.invert(ttHF) else: processes["unsplit"] = mask_events for p in processes.keys(): mask_events_split = processes[p] # Categories categories = {} categories["sl_jge4_tge2"] = mask_events_split categories["sl_jge4_tge3"] = mask_events_split & (btags >= 3) categories["sl_jge4_tge4"] = mask_events_split & (btags >= 4) categories["sl_j4_tge3"] = mask_events_split & (njets == 4) & (btags >= 3) categories["sl_j5_tge3"] = mask_events_split & (njets == 5) & (btags >= 3) categories["sl_jge6_tge3"] = mask_events_split & (njets >= 6) & (btags >= 3) categories["sl_j4_t3"] = mask_events_split & (njets == 4) & (btags == 3) categories["sl_j4_tge4"] = mask_events_split & (njets == 4) & (btags >= 4) categories["sl_j5_t3"] = mask_events_split & (njets == 5) & (btags == 3) categories["sl_j5_tge4"] = mask_events_split & (njets == 5) & (btags >= 4) categories["sl_jge6_t3"] = mask_events_split & (njets >= 6) & (btags == 3) categories["sl_jge6_tge4"] = mask_events_split & (njets >= 6) & (btags >= 4) #print("sl_j4_t3", scalars["event"][categories["sl_j4_t3"]], len(scalars["event"][categories["sl_j4_t3"]])) #print("sl_j5_t3", scalars["event"][categories["sl_j5_t3"]], len(scalars["event"][categories["sl_j5_t3"]])) #print("sl_jge6_t3", scalars["event"][categories["sl_jge6_t3"]], len(scalars["event"][categories["sl_jge6_t3"]])) #print("sl_j4_tge4", scalars["event"][categories["sl_j4_tge4"]], len(scalars["event"][categories["sl_j4_tge4"]])) #print("sl_j5_tge4", scalars["event"][categories["sl_j5_tge4"]], len(scalars["event"][categories["sl_j5_tge4"]])) #print("sl_jge6_tge4", scalars["event"][categories["sl_jge6_tge4"]], len(scalars["event"][categories["sl_jge6_tge4"]])) if not isinstance(cat, list): cat = [cat] for c in cat: cut = categories[c] cut_name = c if p == "unsplit": if "Run" in sample: name = "data" + "_" + cut_name else: name = samples_info[sample]["process"] + "_" + cut_name else: name = p + "_" + cut_name # create histograms filled with weighted events for k in var.keys(): if not k in histogram_settings.keys(): raise Exception( "please add variable {0} to definitions_analysis.py". format(k)) hist = Histogram(*ha.histogram_from_vector( var[k][cut], weights["nominal"][cut], NUMPY_LIB.linspace(histogram_settings[k][0], histogram_settings[k][1], histogram_settings[k][2]))) ret["hist_{0}_{1}".format(name, k)] = hist if DNN: if DNN == "mass_fit": hist_DNN = Histogram(*ha.histogram_from_vector( DNN_pred[cut], weights["nominal"][cut], NUMPY_LIB.linspace(0., 300., 30))) hist_DNN_zoom = Histogram(*ha.histogram_from_vector( DNN_pred[cut], weights["nominal"][cut], NUMPY_LIB.linspace(0., 170., 30))) else: hist_DNN = Histogram(*ha.histogram_from_vector( DNN_pred[cut], weights["nominal"][cut], NUMPY_LIB.linspace(0., 1., 16))) ret["hist_{0}_DNN".format(name)] = hist_DNN ret["hist_{0}_DNN_zoom".format(name)] = hist_DNN_zoom #TODO: implement JECs ## To display properties of a single event #evts = [5991859] #mask = NUMPY_LIB.zeros_like(mask_events) #for iev in evts: # mask |= (scalars["event"] == iev) ##import pdb ##pdb.set_trace() #print("mask", mask) #print('nevt', scalars["event"][mask]) #print('pass sel', mask_events[mask]) #print('nleps', nleps[mask]) #print('njets', njets[mask]) ##print('met', scalars['MET_pt_nom'][mask]) ##print('lep_pt', leading_lepton_pt[mask]) ##print('jet_pt', leading_jet_pt[mask]) ##print('lep_eta', leading_lepton_eta[mask]) #print('pu_weight', pu_weights[mask]) #print('btag_weight', btag_weights[mask]) #print('lep_weight', muon_weights[mask] * electron_weights[mask]) #print('nevents', np.count_nonzero(mask_events)) #np.set_printoptions(formatter={'float': lambda x: "{0:0.3f}".format(x)}) #for evt in evts: # evt_idx = NUMPY_LIB.where( scalars["event"] == evt )[0][0] # start = jets.offsets[evt_idx] # stop = jets.offsets[evt_idx+1] # print(f'!!! EVENT {evt} !!!') # print(f'njets good {njets[evt_idx]}, total {stop-start}') # #print('jets mask', nonbjets[start:stop]) # print('jets pt', jets.pt_nom[start:stop]) # print('jets eta', jets.eta[start:stop]) # print('jets btag', getattr(jets, parameters["btagging algorithm"])[start:stop]) # print('jet Id', jets.jetId[start:stop]), # print('jet puId', jets.puId[start:stop]) return ret
def get_histogram(data, weights, bins): return Histogram(*ha.histogram_from_vector(data, weights, bins))
def analyze_data_function(data, parameters): ret = Results() ha = parameters["ha"] num_events = data["num_events"] lep = data["Lep"] lep.hepaccelerate_backend = ha lep.attrs_data["pt"] = lep.lep_pt lep.attrs_data["eta"] = lep.lep_eta lep.attrs_data["phi"] = lep.lep_phi lep.attrs_data["charge"] = lep.lep_charge lep.attrs_data["type"] = lep.lep_type lep_mass = np.zeros_like(lep["pt"], dtype=nplib.float32) lep_mass = np.where(lep["type"] == 11, 0.511, lep_mass) lep_mass = np.where(lep["type"] == 13, 105.65837, lep_mass) lep.attrs_data["mass"] = lep_mass mask_events = nplib.ones(lep.numevents(), dtype=nplib.bool) lep_ele = lep["type"] == 11 lep_muon = lep["type"] == 13 ele_Iso = np.logical_and( lep_ele, np.logical_and(lep.lep_ptcone30 / lep.pt < 0.15, lep.lep_etcone20 / lep.pt < 0.20)) muon_Iso = np.logical_and( lep_muon, np.logical_and(lep.lep_ptcone30 / lep.pt < 0.15, lep.lep_etcone20 / lep.pt < 0.30)) pass_iso = np.logical_or(ele_Iso, muon_Iso) lep.attrs_data["pass_iso"] = pass_iso num_lep_event = kernels.sum_in_offsets( backend, lep.offsets, lep.masks["all"], mask_events, lep.masks["all"], nplib.int8, ) mask_events_4lep = num_lep_event == 4 lep_attrs = ["pt", "eta", "phi", "charge", "type", "mass", "pass_iso"] #, "ptcone30", "etcone20"] lep0 = lep.select_nth(0, mask_events_4lep, lep.masks["all"], attributes=lep_attrs) lep1 = lep.select_nth(1, mask_events_4lep, lep.masks["all"], attributes=lep_attrs) lep2 = lep.select_nth(2, mask_events_4lep, lep.masks["all"], attributes=lep_attrs) lep3 = lep.select_nth(3, mask_events_4lep, lep.masks["all"], attributes=lep_attrs) mask_event_sumchg_zero = (lep0["charge"] + lep1["charge"] + lep2["charge"] + lep3["charge"] == 0) sum_lep_type = lep0["type"] + lep1["type"] + lep2["type"] + lep3["type"] all_pass_iso = (lep0["pass_iso"] & lep1["pass_iso"] & lep2["pass_iso"] & lep3["pass_iso"]) mask_event_sum_lep_type = np.logical_or( (sum_lep_type == 44), np.logical_or((sum_lep_type == 48), (sum_lep_type == 52))) mask_events = mask_events & mask_event_sumchg_zero & mask_events_4lep & mask_event_sum_lep_type & all_pass_iso mask_lep1_passing_pt = lep1["pt"] > parameters["leading_lep_ptcut"] mask_lep2_passing_pt = lep2["pt"] > parameters["lep_ptcut"] mask_events = mask_events & mask_lep1_passing_pt & mask_lep2_passing_pt l0 = to_cartesian(lep0) l1 = to_cartesian(lep1) l2 = to_cartesian(lep2) l3 = to_cartesian(lep3) llll = {k: l0[k] + l1[k] + l2[k] + l3[k] for k in ["px", "py", "pz", "e"]} llll_sph = to_spherical(llll) llll_sph["mass"] = llll_sph["mass"] / 1000. # Convert to GeV #import pdb;pdb.set_trace(); # compute a weighted histogram weights = nplib.ones(num_events, dtype=nplib.float32) ## Add xsec weights based on sample name if parameters["is_mc"]: weights = data['eventvars']['mcWeight'] * data['eventvars'][ 'scaleFactor_PILEUP'] * data['eventvars']['scaleFactor_ELE'] * data[ 'eventvars']['scaleFactor_MUON'] * data['eventvars'][ 'scaleFactor_LepTRIGGER'] info = infofile.infos[parameters["sample"]] weights *= (lumi * 1000 * info["xsec"]) / (info["sumw"] * info["red_eff"]) bins = nplib.linspace(110, 150, 11, dtype=nplib.float32) hist_m4lep = Histogram(*kernels.histogram_from_vector( backend, llll_sph["mass"][mask_events], weights[mask_events], bins, )) # save it to the output ret["hist_m4lep"] = hist_m4lep return ret
def analyze_data(data, sample, NUMPY_LIB=None, parameters={}, samples_info={}, is_mc=True, lumimask=None, cat=False, boosted=False, DNN=False, DNN_model=None): #Output structure that will be returned and added up among the files. #Should be relatively small. ret = Results() muons = data["Muon"] electrons = data["Electron"] scalars = data["eventvars"] jets = data["Jet"] nEvents = muons.numevents() mask_events = NUMPY_LIB.ones(nEvents, dtype=NUMPY_LIB.bool) # apply event cleaning, PV selection and trigger selection flags = [ "Flag_goodVertices", "Flag_globalSuperTightHalo2016Filter", "Flag_HBHENoiseFilter", "Flag_HBHENoiseIsoFilter", "Flag_EcalDeadCellTriggerPrimitiveFilter", "Flag_BadPFMuonFilter", "Flag_BadChargedCandidateFilter", "Flag_ecalBadCalibFilter"] if not is_mc: flags.append("Flag_eeBadScFilter") for flag in flags: mask_events = mask_events & scalars[flag] if args.year.startswith('2016'): trigger = (scalars["HLT_Ele27_WPTight_Gsf"] | scalars["HLT_IsoMu24"] | scalars["HLT_IsoTkMu24"]) else: trigger = (scalars["HLT_Ele35_WPTight_Gsf"] | scalars["HLT_Ele28_eta2p1_WPTight_Gsf_HT150"] | scalars["HLT_IsoMu27"]) mask_events = mask_events & trigger mask_events = mask_events & (scalars["PV_npvsGood"]>0) #mask_events = vertex_selection(scalars, mask_events) # apply object selection for muons, electrons, jets good_muons, veto_muons = lepton_selection(muons, parameters["muons"]) good_electrons, veto_electrons = lepton_selection(electrons, parameters["electrons"]) good_jets = jet_selection(jets, muons, (veto_muons | good_muons), parameters["jets"]) & jet_selection(jets, electrons, (veto_electrons | good_electrons) , parameters["jets"]) bjets = good_jets & (getattr(jets, parameters["btagging algorithm"]) > parameters["btagging WP"]) # apply basic event selection -> individual categories cut later nleps = NUMPY_LIB.add(ha.sum_in_offsets(muons, good_muons, mask_events, muons.masks["all"], NUMPY_LIB.int8), ha.sum_in_offsets(electrons, good_electrons, mask_events, electrons.masks["all"], NUMPY_LIB.int8)) lepton_veto = NUMPY_LIB.add(ha.sum_in_offsets(muons, veto_muons, mask_events, muons.masks["all"], NUMPY_LIB.int8), ha.sum_in_offsets(electrons, veto_electrons, mask_events, electrons.masks["all"], NUMPY_LIB.int8)) njets = ha.sum_in_offsets(jets, good_jets, mask_events, jets.masks["all"], NUMPY_LIB.int8) btags = ha.sum_in_offsets(jets, bjets, mask_events, jets.masks["all"], NUMPY_LIB.int8) met = (scalars["MET_pt"] > 20) # apply basic event definition (inverted for boosted analysis) if boosted: mask_events = mask_events & (nleps == 1) & (lepton_veto == 0) & NUMPY_LIB.invert( (njets >= 4) & (btags >=2) ) & met else: mask_events = mask_events & (nleps == 1) & (lepton_veto == 0) & (njets >= 4) & (btags >=2) & met ### check overlap between AK4 and AK8 jets: if (based on tau32 and tau21) the AK8 jet is a t/H/W candidate remove the AK4 jet, otherwise remove the AK8 jet if boosted: fatjets = data["FatJet"] genparts = data["GenPart"] # get fatjets good_fatjets = jet_selection(fatjets, muons, (veto_muons | good_muons), parameters["fatjets"]) & jet_selection(fatjets, electrons, (veto_electrons | good_electrons), parameters["fatjets"]) bfatjets = good_fatjets & (fatjets.btagHbb > parameters["bbtagging WP"]) fatjets.tau32 = NUMPY_LIB.divide(fatjets.tau3, fatjets.tau2) fatjets.tau21 = NUMPY_LIB.divide(fatjets.tau2, fatjets.tau1) jets_to_keep = ha.mask_overlappingAK4(jets, good_jets, fatjets, good_fatjets, 1.2, tau32cut=parameters["fatjets"]["tau32cut"], tau21cut=parameters["fatjets"]["tau21cut"]) non_overlapping_fatjets = ha.mask_deltar_first(fatjets, good_fatjets, jets, good_jets, 1.2) good_jets &= jets_to_keep good_fatjets &= non_overlapping_fatjets | (fatjets.tau32 < parameters["fatjets"]["tau32cut"]) | (fatjets.tau21 < parameters["fatjets"]["tau21cut"]) #we keep fat jets which are not overlapping, or if they are either a top or W/H candidate top_candidates = (fatjets.tau32 < parameters["fatjets"]["tau32cut"]) WH_candidates = (fatjets.tau32 > tau32cut) & (fatjets.tau21 < parameters["fatjets"]["tau21cut"]) bjets = good_jets & (jets.btagDeepB > parameters["btagging WP"]) njets = ha.sum_in_offsets(jets, good_jets, mask_events, jets.masks["all"], NUMPY_LIB.int8) btags = ha.sum_in_offsets(jets, bjets, mask_events, jets.masks["all"], NUMPY_LIB.int8) bbtags = ha.sum_in_offsets(fatjets, bfatjets, mask_events, fatjets.masks["all"], NUMPY_LIB.int8) ntop_candidates = ha.sum_in_offsets(fatjets, top_candidates, mask_events, fatjets.masks["all"], NUMPY_LIB.int8) nWH_candidates = ha.sum_in_offsets(fatjets, WH_candidates, mask_events, fatjets.masks["all"], NUMPY_LIB.int8) ### 2 fat jets from H and W, 2 b jets from the tops #mask_events &= (nWH_candidates > 1) & (btags > 1) ### 1 top candidate and 1 H candidate, and 1 b jet from the leptonic top mask_events &= (ntop_candidates > 0) & (nWH_candidates > 0) & (btags > 0) ### calculation of all needed variables var = {} var["njets"] = njets var["btags"] = btags var["nleps"] = nleps if boosted: higgs = (genparts.pdgId == 25) & (genparts.status==62) tops = ( (genparts.pdgId == 6) | (genparts.pdgId == -6) ) & (genparts.status==62) var["nfatjets"] = ha.sum_in_offsets(fatjets, good_fatjets, mask_events, fatjets.masks["all"], NUMPY_LIB.int8) var["ntop_candidates"] = ha.sum_in_offsets(fatjets, tops, mask_events, fatjets.masks["all"], NUMPY_LIB.int8) indices = {} indices["leading"] = NUMPY_LIB.zeros(nEvents, dtype=NUMPY_LIB.int32) indices["subleading"] = NUMPY_LIB.ones(nEvents, dtype=NUMPY_LIB.int32) if boosted: indices["inds_WHcandidates"] = ha.index_in_offsets(fatjets.btagHbb, fatjets.offsets, 1, mask_events, WH_candidates) variables = [ ("jet", jets, good_jets, "leading", ["pt", "eta"]), ("bjet", jets, bjets, "leading", ["pt", "eta"]), ] if boosted: variables += [ ("fatjet", fatjets, good_fatjets, "leading",["pt", "eta", "mass", "msoftdrop", "tau32", "tau21"]), ("fatjet", fatjets, good_fatjets, "subleading",["pt", "eta", "mass", "msoftdrop", "tau32", "tau21"]), ("top_candidate", fatjets, top_candidates, "leading", ["pt", "eta", "mass", "msoftdrop", "tau32", "tau21"]), ("WH_candidate", fatjets, WH_candidates, "inds_WHcandidates", ["pt", "eta", "mass", "msoftdrop", "tau32", "tau21"]), ("higgs", genparts, higgs, "leading", ["pt", "eta"]), ("tops", genparts, tops, "leading", ["pt", "eta"]) ] # special role of lepton var["leading_lepton_pt"] = NUMPY_LIB.maximum(ha.get_in_offsets(muons.pt, muons.offsets, indices["leading"], mask_events, good_muons), ha.get_in_offsets(electrons.pt, electrons.offsets, indices["leading"], mask_events, good_electrons)) var["leading_lepton_eta"] = NUMPY_LIB.maximum(ha.get_in_offsets(muons.eta, muons.offsets, indices["leading"], mask_events, good_muons), ha.get_in_offsets(electrons.eta, electrons.offsets, indices["leading"], mask_events, good_electrons)) # all other variables for v in variables: calculate_variable_features(v, mask_events, indices, var) # calculate weights for MC samples weights = {} weights["nominal"] = NUMPY_LIB.ones(nEvents, dtype=NUMPY_LIB.float32) if is_mc: weights["nominal"] = weights["nominal"] * scalars["genWeight"] * parameters["lumi"] * samples_info[sample]["XS"] / samples_info[sample]["ngen_weight"] # pu corrections pu_weights = compute_pu_weights(parameters["pu_corrections_target"], weights["nominal"], scalars["Pileup_nTrueInt"], scalars["PV_npvsGood"]) weights["nominal"] = weights["nominal"] * pu_weights # lepton SF corrections electron_weights = compute_lepton_weights(electrons, electrons.pt, (electrons.deltaEtaSC + electrons.eta), mask_events, good_electrons, evaluator, ["el_triggerSF", "el_recoSF", "el_idSF"]) muon_weights = compute_lepton_weights(muons, muons.pt, NUMPY_LIB.abs(muons.eta), mask_events, good_muons, evaluator, ["mu_triggerSF", "mu_isoSF", "mu_idSF"]) weights["nominal"] = weights["nominal"] * muon_weights * electron_weights # btag SF corrections btag_weights = compute_btag_weights(jets, mask_events, good_jets, evaluator) weights["nominal"] = weights["nominal"] * btag_weights #in case of data: check if event is in golden lumi file if not is_mc and not (lumimask is None): mask_lumi = lumimask(scalars["run"], scalars["luminosityBlock"]) mask_events = mask_events & mask_lumi #evaluate DNN if DNN: DNN_pred = evaluate_DNN(jets, good_jets, electrons, good_electrons, muons, good_muons, scalars, mask_events, DNN, DNN_model) # in case of tt+jets -> split in ttbb, tt2b, ttb, ttcc, ttlf processes = {} if sample.startswith("TT"): ttCls = scalars["genTtbarId"]%100 processes["ttbb"] = mask_events & (ttCls >=53) & (ttCls <=56) processes["tt2b"] = mask_events & (ttCls ==52) processes["ttb"] = mask_events & (ttCls ==51) processes["ttcc"] = mask_events & (ttCls >=41) & (ttCls <=45) ttHF = ((ttCls >=53) & (ttCls <=56)) | (ttCls ==52) | (ttCls ==51) | ((ttCls >=41) & (ttCls <=45)) processes["ttlf"] = mask_events & NUMPY_LIB.invert(ttHF) else: processes["unsplit"] = mask_events for p in processes.keys(): mask_events_split = processes[p] # Categories categories = {} if not boosted: categories["sl_jge4_tge2"] = mask_events_split categories["sl_jge4_tge3"] = mask_events_split & (btags >=3) categories["sl_j4_tge3"] = mask_events_split & (njets ==4) & (btags >=3) categories["sl_j5_tge3"] = mask_events_split & (njets ==5) & (btags >=3) categories["sl_jge6_tge3"] = mask_events_split & (njets >=6) & (btags >=3) categories["sl_j4_t3"] = mask_events_split & (njets ==4) & (btags ==3) categories["sl_j4_tge4"] = mask_events_split & (njets ==4) & (btags >=4) categories["sl_j5_t3"] = mask_events_split & (njets ==5) & (btags ==3) categories["sl_j5_tge4"] = mask_events_split & (njets ==5) & (btags >=4) categories["sl_jge6_t3"] = mask_events_split & (njets >=6) & (btags ==3) categories["sl_jge6_tge4"] = mask_events_split & (njets >=6) & (btags >=4) if not isinstance(cat, list): cat = [cat] for c in cat: cut = categories[c] cut_name = c if p=="unsplit": if "Run" in sample: name = "data" + "_" + cut_name else: name = samples_info[sample]["process"] + "_" + cut_name else: name = p + "_" + cut_name # create histograms filled with weighted events for k in var.keys(): if not k in histogram_settings.keys(): raise Exception("please add variable {0} to config_analysis.py".format(k)) hist = Histogram(*ha.histogram_from_vector(var[k][cut], weights["nominal"][cut], NUMPY_LIB.linspace(histogram_settings[k][0], histogram_settings[k][1], histogram_settings[k][2]))) ret["hist_{0}_{1}".format(name, k)] = hist if DNN: if DNN.endswith("multiclass"): class_pred = NUMPY_LIB.argmax(DNN_pred, axis=1) for n, n_name in zip([0,1,2,3,4,5], ["ttH", "ttbb", "tt2b", "ttb", "ttcc", "ttlf"]): node = (class_pred == n) DNN_node = DNN_pred[:,n] hist_DNN = Histogram(*ha.histogram_from_vector(DNN_node[(cut & node)], weights["nominal"][(cut & node)], NUMPY_LIB.linspace(0.,1.,16))) ret["hist_{0}_DNN_{1}".format(name, n_name)] = hist_DNN hist_DNN_ROC = Histogram(*ha.histogram_from_vector(DNN_node[(cut & node)], weights["nominal"][(cut & node)], NUMPY_LIB.linspace(0.,1.,1000))) ret["hist_{0}_DNN_ROC_{1}".format(name, n_name)] = hist_DNN_ROC else: hist_DNN = Histogram(*ha.histogram_from_vector(DNN_pred[cut], weights["nominal"][cut], NUMPY_LIB.linspace(0.,1.,16))) ret["hist_{0}_DNN".format(name)] = hist_DNN hist_DNN_ROC = Histogram(*ha.histogram_from_vector(DNN_pred[cut], weights["nominal"][cut], NUMPY_LIB.linspace(0.,1.,1000))) ret["hist_{0}_DNN_ROC".format(name)] = hist_DNN_ROC #TODO: implement JECs return ret
def divide(h1,h2): contents = h1.contents/h2.contents contents_w2 = h1.contents_w2/h2.contents_w2 edges = h1.edges return Histogram(contents, contents_w2, edges)
import numpy as np from hepaccelerate.utils import Histogram, Results from glob import glob import json,os,argparse from pdb import set_trace flist = glob('results/201*/v12/met20_btagDDBvL086/nominal/btagEfficiencyMaps/out_btagEfficiencyMaps_*json') def divide(h1,h2): contents = h1.contents/h2.contents contents_w2 = h1.contents_w2/h2.contents_w2 edges = h1.edges return Histogram(contents, contents_w2, edges) for fn in flist: with open(fn) as f: data = json.load(f) for h in data: data[h] = Histogram( *data[h].values() ) for flav in ['b','l','lc']: for var in ['central','updown']: data[f'eff_flav{flav}_{var}'] = divide( data[f'btags_flav{flav}_{var}'], data[f'total_flav{flav}_{var}'] ) ret = Results(data) ret.save_json(fn)