-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathdet_unfd_strength.py
171 lines (148 loc) · 10.4 KB
/
det_unfd_strength.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
from hadana.packages import *
import hadana.slicing_method as slicing
import hadana.multiD_mapping as multiD
import hadana.parameters as parameters
from scipy import stats
beamPDG = 211
datafilename = "processed_files/procVars_pidata.pkl"
MCfilename = "processed_files/procVars_piMC.pkl"
# types of systematic uncertainties to include
bkg_scale = [1, 1, 1, 1, 1, 1, 1] # should be imported from sideband fit pionp [0.93, 1, 1.72, 1.43, 0.93, 1, 1] proton [1, 1, 1, 1, 1, 1, 1]
bkg_scale_err = [0, 0, 0, 0, 0, 0, 0] # pionp [0.12, 0, 0.13, 0.11, 0.12, 0, 0] proton [0, 0, 0, 0, 0, 0, 0]
inc_sys_bkg = True
if beamPDG == 211:
true_bins = parameters.true_bins_pionp
meas_bins = parameters.meas_bins_pionp
elif beamPDG == 2212:
true_bins = parameters.true_bins_proton
meas_bins = parameters.meas_bins_proton
### load data histograms
with open(datafilename, 'rb') as datafile: # either fake data or real data
processedVars = pickle.load(datafile)
mask_SelectedPart = processedVars["mask_SelectedPart"]
mask_FullSelection = processedVars["mask_FullSelection"]
combined_mask = mask_SelectedPart & mask_FullSelection
reco_initial_energy = processedVars["reco_initial_energy"]
reco_end_energy = processedVars["reco_end_energy"]
reco_sigflag = processedVars["reco_sigflag"]
reco_containing = processedVars["reco_containing"]
particle_type = processedVars["particle_type"]
weight_dt = processedVars["reweight"]
### selection
print("### selection")
divided_recoEini, divided_weights = utils.divide_vars_by_partype(reco_initial_energy, particle_type, mask=combined_mask, weight=weight_dt)
divided_recoEend, divided_weights = utils.divide_vars_by_partype(reco_end_energy, particle_type, mask=combined_mask, weight=weight_dt)
divided_recoflag, divided_weights = utils.divide_vars_by_partype(reco_sigflag, particle_type, mask=combined_mask, weight=weight_dt)
divided_recoisct, divided_weights = utils.divide_vars_by_partype(reco_containing, particle_type, mask=combined_mask, weight=weight_dt)
data_reco_Eini = divided_recoEini[0]
data_reco_Eend = divided_recoEend[0]
data_reco_flag = divided_recoflag[0]
data_reco_isCt = divided_recoisct[0]
data_reco_weight = divided_weights[0]
Ndata = len(data_reco_Eini)
Ntruebins, Ntruebins_3D, true_cKE, true_wKE = utils.set_bins(true_bins)
Nmeasbins = len(meas_bins)
data_meas_SIDini, data_meas_SIDend, data_meas_SIDint_ex = slicing.get_sliceID_histograms(data_reco_Eini, data_reco_Eend, data_reco_flag, data_reco_isCt, meas_bins)
data_meas_SID3D, data_meas_N3D, data_meas_N3D_Vcov = slicing.get_3D_histogram(data_meas_SIDini, data_meas_SIDend, data_meas_SIDint_ex, Nmeasbins, data_reco_weight)
data_meas_N3D_err = np.sqrt(np.diag(data_meas_N3D_Vcov))
### background subtraction
print("### background subtraction")
with open(MCfilename, 'rb') as mcfile: # truth MC
processedVars_mc = pickle.load(mcfile)
mask_SelectedPart_mc = processedVars_mc["mask_SelectedPart"]
mask_FullSelection_mc = processedVars_mc["mask_FullSelection"]
combined_mask_mc = mask_SelectedPart_mc & mask_FullSelection_mc
reco_initial_energy_mc = processedVars_mc["reco_initial_energy"]
reco_end_energy_mc = processedVars_mc["reco_end_energy"]
reco_sigflag_mc = processedVars_mc["reco_sigflag"]
reco_containing_mc = processedVars_mc["reco_containing"]
particle_type_mc = processedVars_mc["particle_type"]
weight_mc = processedVars_mc["reweight"]
divided_recoEini_mc, divided_weights_mc = utils.divide_vars_by_partype(reco_initial_energy_mc, particle_type_mc, mask=combined_mask_mc, weight=weight_mc)
divided_recoEend_mc, divided_weights_mc = utils.divide_vars_by_partype(reco_end_energy_mc, particle_type_mc, mask=combined_mask_mc, weight=weight_mc)
divided_recoflag_mc, divided_weights_mc = utils.divide_vars_by_partype(reco_sigflag_mc, particle_type_mc, mask=combined_mask_mc, weight=weight_mc)
divided_recoisct_mc, divided_weights_mc = utils.divide_vars_by_partype(reco_containing_mc, particle_type_mc, mask=combined_mask_mc, weight=weight_mc)
Ntruemc = len(reco_initial_energy_mc[combined_mask_mc]) - len(divided_recoEini_mc[0])
bkg_meas_N3D_list = []
bkg_meas_N3D_err_list = []
for ibkg in range(3, len(divided_recoEini_mc)):
bkg_reco_Eini = divided_recoEini_mc[ibkg]
bkg_reco_Eend = divided_recoEend_mc[ibkg]
bkg_reco_flag = divided_recoflag_mc[ibkg]
bkg_reco_isCt = divided_recoisct_mc[ibkg]
bkg_reco_weight = divided_weights_mc[ibkg]
bkg_meas_SIDini, bkg_meas_SIDend, bkg_meas_SIDint_ex = slicing.get_sliceID_histograms(bkg_reco_Eini, bkg_reco_Eend, bkg_reco_flag, bkg_reco_isCt, meas_bins)
bkg_meas_SID3D, bkg_meas_N3D, bkg_meas_N3D_Vcov = slicing.get_3D_histogram(bkg_meas_SIDini, bkg_meas_SIDend, bkg_meas_SIDint_ex, Nmeasbins, bkg_reco_weight)
bkg_meas_N3D_list.append(bkg_meas_N3D)
bkg_meas_N3D_err_list.append(np.sqrt(np.diag(bkg_meas_N3D_Vcov)))
print(f"Bkg scale \t{bkg_scale}\nError \t\t{bkg_scale_err}")
sig_meas_N3D, sig_meas_N3D_err = utils.bkg_subtraction(data_meas_N3D, data_meas_N3D_err, bkg_meas_N3D_list, bkg_meas_N3D_err_list, mc2data_scale=Ndata/Ntruemc, bkg_scale=bkg_scale, bkg_scale_err=bkg_scale_err, include_bkg_err=inc_sys_bkg)
### unfolding
print("### model response")
mask_TrueSignal_mc = processedVars_mc["mask_TrueSignal"]
combined_true_mask_mc = mask_SelectedPart_mc & mask_TrueSignal_mc
beam_matched_mc = processedVars_mc["reco_beam_true_byE_matched"]
combined_reco_mask_mc = mask_FullSelection_mc & np.array(beam_matched_mc, dtype=bool)
true_initial_energy_mc = processedVars_mc["true_initial_energy"]
true_end_energy_mc = processedVars_mc["true_end_energy"]
true_sigflag_mc = processedVars_mc["true_sigflag"]
true_containing_mc = processedVars_mc["true_containing"]
particle_type_bool_mc = np.where(particle_type_mc==0, 0, 1) # 0 for fake data, 1 for truth MC
divided_trueEini_mc, divided_weights_mc = utils.divide_vars_by_partype(true_initial_energy_mc, particle_type_bool_mc, mask=combined_true_mask_mc, weight=weight_mc)
divided_trueEend_mc, divided_weights_mc = utils.divide_vars_by_partype(true_end_energy_mc, particle_type_bool_mc, mask=combined_true_mask_mc, weight=weight_mc)
divided_trueflag_mc, divided_weights_mc = utils.divide_vars_by_partype(true_sigflag_mc, particle_type_bool_mc, mask=combined_true_mask_mc, weight=weight_mc)
divided_trueisct_mc, divided_weights_mc = utils.divide_vars_by_partype(true_containing_mc, particle_type_bool_mc, mask=combined_true_mask_mc, weight=weight_mc)
mc_true_Eini = divided_trueEini_mc[1]
mc_true_Eend = divided_trueEend_mc[1]
mc_true_flag = divided_trueflag_mc[1]
mc_true_isCt = divided_trueisct_mc[1]
mc_true_weight = divided_weights_mc[1]
mc_true_SIDini, mc_true_SIDend, mc_true_SIDint_ex = slicing.get_sliceID_histograms(mc_true_Eini, mc_true_Eend, mc_true_flag, mc_true_isCt, true_bins)
mc_true_Nini, mc_true_Nend, mc_true_Nint_ex, mc_true_Ninc = slicing.derive_energy_histograms(mc_true_SIDini, mc_true_SIDend, mc_true_SIDint_ex, Ntruebins, mc_true_weight)
mc_true_SID3D, mc_true_N3D, mc_true_N3D_Vcov = slicing.get_3D_histogram(mc_true_SIDini, mc_true_SIDend, mc_true_SIDint_ex, Ntruebins, mc_true_weight)
divided_recoEini_mc, divided_weights_mc = utils.divide_vars_by_partype(reco_initial_energy_mc, particle_type_bool_mc, mask=combined_true_mask_mc, weight=weight_mc)
divided_recoEend_mc, divided_weights_mc = utils.divide_vars_by_partype(reco_end_energy_mc, particle_type_bool_mc, mask=combined_true_mask_mc, weight=weight_mc)
divided_recoflag_mc, divided_weights_mc = utils.divide_vars_by_partype(reco_sigflag_mc, particle_type_bool_mc, mask=combined_true_mask_mc, weight=weight_mc)
divided_recoisct_mc, divided_weights_mc = utils.divide_vars_by_partype(reco_containing_mc, particle_type_bool_mc, mask=combined_true_mask_mc, weight=weight_mc)
divided_FullSelection_mc, divided_weights_mc = utils.divide_vars_by_partype(mask_FullSelection_mc, particle_type_bool_mc, mask=combined_true_mask_mc, weight=weight_mc)
pass_selection_mc = divided_FullSelection_mc[1]
mc_reco_Eini = divided_recoEini_mc[1][pass_selection_mc]
mc_reco_Eend = divided_recoEend_mc[1][pass_selection_mc]
mc_reco_flag = divided_recoflag_mc[1][pass_selection_mc]
mc_reco_isCt = divided_recoisct_mc[1][pass_selection_mc]
mc_reco_weight = divided_weights_mc[1][pass_selection_mc]
#print(len(mc_reco_Eini), mc_reco_Eini, mc_reco_Eend, mc_reco_flag, mc_reco_weight, mc_pass_selection, sep='\n')
mc_meas_SIDini, mc_meas_SIDend, mc_meas_SIDint_ex = slicing.get_sliceID_histograms(mc_reco_Eini, mc_reco_Eend, mc_reco_flag, mc_reco_isCt, meas_bins)
mc_meas_SID3D, mc_meas_N3D, mc_meas_N3D_Vcov = slicing.get_3D_histogram(mc_meas_SIDini, mc_meas_SIDend, mc_meas_SIDint_ex, Nmeasbins, mc_reco_weight)
true_3D1D_map, mc_true_N1D, mc_true_N1D_err, Ntruebins_1D = multiD.map_index_to_combined_variable(mc_true_N3D, np.sqrt(np.diag(mc_true_N3D_Vcov)), Ntruebins)
meas_3D1D_map, mc_meas_N1D, mc_meas_N1D_err, Nmeasbins_1D = multiD.map_index_to_combined_variable(mc_meas_N3D, np.sqrt(np.diag(mc_meas_N3D_Vcov)), Nmeasbins)
#print(true_3D1D_map, mc_true_N1D, mc_true_N1D_err, Ntruebins_1D, sep='\n')
#print(meas_3D1D_map, mc_meas_N1D, mc_meas_N1D_err, Nmeasbins_1D, sep='\n')
eff1D, mc_true_SID3D_sel = multiD.get_efficiency(mc_true_N3D, mc_true_N1D, mc_true_SID3D, Ntruebins_3D, pass_selection_mc, mc_reco_weight)
response_matrix, response = multiD.get_response_matrix(Nmeasbins_1D, Ntruebins_1D, meas_3D1D_map[mc_meas_SID3D], true_3D1D_map[mc_true_SID3D_sel], mc_reco_weight)
print(eff1D, response_matrix,sep='\n')
print("### unfolding")
sig_meas_N1D, sig_meas_N1D_err = multiD.map_data_to_MC_bins(sig_meas_N3D, sig_meas_N3D_err, meas_3D1D_map)
sig_meas_V1D = np.diag(sig_meas_N1D_err*sig_meas_N1D_err)
#print(sig_meas_N1D, sig_meas_N1D_err, sep='\n')
N_unreg = 100 # the large Niter to represent the unregularized result
N_iter = 100 # the largest Niter to try
sig_unfold_unreg, sig_unfold_cov_unreg = multiD.unfolding(sig_meas_N1D, sig_meas_V1D, response, niter=N_unreg)
niter_list = np.arange(1, N_iter+1)
pvalue_list = [0]
for ii in niter_list:
sig_unfold, sig_unfold_cov = multiD.unfolding(sig_meas_N1D, sig_meas_V1D, response, niter=int(ii), verbose=False)
diff_v = sig_unfold - sig_unfold_unreg
pooled_cov = sig_unfold_cov + sig_unfold_cov_unreg
tstat = np.einsum('i,ij,j->', diff_v, np.linalg.pinv(pooled_cov), diff_v)
pvalue = 1 - stats.chi2.cdf(tstat, len(sig_unfold))
pvalue_list.append(pvalue)
print("Input hist:", sig_meas_N1D.tolist())
plt.plot(np.concatenate([[0], niter_list]), pvalue_list[:], "*-")
plt.plot([0, N_iter], [0.5]*2, "r:")
plt.xlabel("Number of iterations")
plt.ylabel("p-value")
plt.xlim([0, N_iter])
plt.ylim([0, 1])
plt.savefig(f"plots/Niter_{beamPDG}.pdf")
plt.show()