Skip to content
Snippets Groups Projects

Analysis of truth/reconstruction associations

Merged Analysis of truth/reconstruction associations
truth_reconstruction into master
Merged Tooba Ali requested to merge truth_reconstruction into master
Compare
and
5 files
+ 337
− 0
Compare changes
  • Side-by-side
  • Inline
Files
5
benchmarks/dis/analysis/truth_reconstruction.py 0 → 100644
+ 315
− 0
import os
import numpy as np
import uproot as ur
import awkward as ak
import matplotlib.pyplot as plt
import matplotlib as mpl
import mplhep
import argparse
parser = argparse.ArgumentParser()
parser.add_argument('--rec_file', type=str, help='Reconstructed track file.')
parser.add_argument('--config', type=str, help='Momentum configuration.')
parser.add_argument('--nevents', type=float, help='Number of events to process.')
parser.add_argument('--results_path', type=str, help='Output directory.')
args = parser.parse_args()
kwargs = vars(args)
rec_file = args.rec_file
config = args.config
Nevents = int(args.nevents)
r_path = args.results_path + '/truth_reconstruction/'
for array in ur.iterate(rec_file + ':events',['MCParticles/MCParticles.generatorStatus',
'MCParticles/MCParticles.PDG',
'MCParticles/MCParticles.momentum.x',
'MCParticles/MCParticles.momentum.y',
'MCParticles/MCParticles.momentum.z',
'ReconstructedParticles/ReconstructedParticles.PDG',
'ReconstructedParticles/ReconstructedParticles.momentum.x',
'ReconstructedParticles/ReconstructedParticles.momentum.y',
'ReconstructedParticles/ReconstructedParticles.momentum.z',
'ReconstructedParticlesAssoc/ReconstructedParticlesAssoc.simID',
'ReconstructedParticlesAssoc/ReconstructedParticlesAssoc.recID',],step_size=Nevents):
PDG_mc = array['MCParticles/MCParticles.PDG']
px_mc = array['MCParticles/MCParticles.momentum.x']
py_mc = array['MCParticles/MCParticles.momentum.y']
pz_mc = array['MCParticles/MCParticles.momentum.z']
PDG_rc = array['ReconstructedParticles/ReconstructedParticles.PDG']
px_rc = array['ReconstructedParticles/ReconstructedParticles.momentum.x']
py_rc = array['ReconstructedParticles/ReconstructedParticles.momentum.y']
pz_rc = array['ReconstructedParticles/ReconstructedParticles.momentum.z']
simID = array['ReconstructedParticlesAssoc/ReconstructedParticlesAssoc.simID']
recID = array['ReconstructedParticlesAssoc/ReconstructedParticlesAssoc.recID']
momentum_mc = np.sqrt(((px_mc**2)+(py_mc**2)+(pz_mc**2)))
theta_mc = np.arctan2(np.sqrt(px_mc**2+py_mc**2), pz_mc)
phi_mc = np.arctan2(py_mc, px_mc)
momentum_rc = np.sqrt(((px_rc**2)+(py_rc**2)+(pz_rc**2)))
theta_rc = np.arctan2(np.sqrt(px_rc**2+py_rc**2), pz_rc)
phi_rc = np.arctan2(py_rc, px_rc)
booll = (PDG_mc[simID])==(PDG_rc[recID])
boolean_pion = np.logical_or(ak.Array(PDG_mc[simID][booll])==-211, ak.Array(PDG_mc[simID][booll])==+211)
boolean_proton = np.logical_or(ak.Array(PDG_mc[simID][booll])==-2212, ak.Array(PDG_mc[simID][booll])==+2212)
boolean_electron = ak.Array(PDG_mc[simID][booll])==11
boolean_neutron = ak.Array(PDG_mc[simID][booll])==2112
boolean_photon = ak.Array(PDG_mc[simID][booll])==22
MC_list = [ak.Array(momentum_mc[simID][booll]),
ak.Array(theta_mc[simID][booll]),
ak.Array(phi_mc[simID][booll]),
-np.log(np.tan((ak.Array(theta_mc[simID][booll]))/2))]
RC_list = [ak.Array(momentum_rc[recID][booll]),
ak.Array(theta_rc[recID][booll]),
ak.Array(phi_rc[recID][booll]),
-np.log(np.tan((ak.Array(theta_rc[recID][booll]))/2))]
M_list = [ak.Array(momentum_mc[simID][booll]),
ak.Array(momentum_mc[simID][booll][boolean_pion]),
ak.Array(momentum_mc[simID][booll][boolean_proton]),
ak.Array(momentum_mc[simID][booll][boolean_electron]),
ak.Array(momentum_mc[simID][booll][boolean_neutron]),
ak.Array(momentum_mc[simID][booll][boolean_photon])]
title_list = ['Momentum','Theta','Phi','Eta']
if Nevents == 100:
ssize = 1
else:
ssize = 0.01
particle = config.split('-')[0].strip()
particle_dict = {'e':[boolean_electron,'Electrons'],'pi':[boolean_pion,'Pions']}
###################
for i in range(len(MC_list)):
X1 = MC_list[i]
Y1 = RC_list[i]
X_list = [ak.Array(X1),
ak.Array(X1[boolean_pion]),
ak.Array(X1[boolean_proton]),
ak.Array(X1[boolean_electron]),
ak.Array(X1[boolean_neutron]),
ak.Array(X1[boolean_photon])]
Y_list = [ak.Array(Y1),
ak.Array(Y1[boolean_pion]),
ak.Array(Y1[boolean_proton]),
ak.Array(Y1[boolean_electron]),
ak.Array(Y1[boolean_neutron]),
ak.Array(Y1[boolean_photon])]
X_plot = list(np.zeros(len(X_list)))
Y_plot = list(np.zeros(len(X_list)))
for j in range(len(X_list)):
X = X_list[j]
Y = Y_list[j]
X_len = ak.count(X,axis=None)
Y_len = ak.count(Y,axis=None)
if X_len > Y_len:
F_boolean = np.ones_like(Y) == 1
else:
F_boolean = np.ones_like(X) == 1
X_s = np.array(ak.flatten(X[F_boolean]))
Y_s = np.array(ak.flatten(Y[F_boolean]))
if i == 0: #Momentum
ratio = np.array((ak.Array(Y_s)/ak.Array(X_s)))
else:
ratio = np.array((ak.Array(Y_s)-(ak.Array(X_s))))
X_plot[j] = X_s
Y_plot[j] = ratio
fig = plt.figure()
gs = fig.add_gridspec(3, 2, wspace=0)
(ax1, ax2), (ax3, ax4),(ax5, ax6) = gs.subplots(sharex=True, sharey=True)
# fig.suptitle('')
if i == 1: #theta
X_plot[0],X_plot[1],X_plot[2],X_plot[3],X_plot[4],X_plot[5] = -X_plot[0],-X_plot[1],-X_plot[2],-X_plot[3],-X_plot[4],-X_plot[5]
ax1.scatter(X_plot[0], Y_plot[0], s = ssize)
ax2.scatter(X_plot[1], Y_plot[1], s = ssize)
ax3.scatter(X_plot[2], Y_plot[2], s = ssize)
ax4.scatter(X_plot[3], Y_plot[3], s = ssize)
ax5.scatter(X_plot[4], Y_plot[4], s = ssize)
ax6.scatter(X_plot[5], Y_plot[5], s = ssize)
ax_list = [ax1,ax2,ax3,ax4,ax5]
if i == 0:
ax1.set_ylabel('rc/mc')
ax3.set_ylabel('rc/mc')
ax5.set_ylabel('rc/mc')
title ='ratio'
for ax in ax_list:
ax.set_yscale('log')
ax.set_xscale('log')
else:
ax1.set_ylabel('rc-mc')
ax3.set_ylabel('rc-mc')
ax5.set_ylabel('rc-mc')
title ='difference'
ax2.set_title('Pions')
ax3.set_title('Protons')
ax4.set_title('Electrons')
ax5.set_title('Neutrons')
ax6.set_title('Photons')
if i == 3: #Eta
ax1.set_xlim(-5.5,5.5)
if i == 1: #Theta
ax1.set_xlim(-np.pi,0)
ax5.set_xlabel('- %s mc'%(title_list[i]))
ax6.set_xlabel('- %s mc'%(title_list[i]))
x_range = [0,np.pi]
else:
ax5.set_xlabel('%s mc'%(title_list[i]))
ax6.set_xlabel('%s mc'%(title_list[i]))
x_range = list(ax1.get_xlim())
fig.set_figwidth(20)
fig.set_figheight(10)
ax1.set_title('%s %s %s %s events'%(title_list[i],title,config,Nevents))
plt.savefig(os.path.join(r_path, '%s_%s_%s.png' % (title_list[i],title,config)))
plt.close()
############
#Correlation
X_len = ak.count(X1,axis=None)
Y_len = ak.count(Y1,axis=None)
if X_len > Y_len:
F_boolean = np.ones_like(Y1) == 1
else:
F_boolean = np.ones_like(X1) == 1
X_s = np.array(ak.flatten(X1[F_boolean]))
Y_s = np.array(ak.flatten(Y1[F_boolean]))
if i == 0 and particle in particle_dict.keys(): #Momentum in Single events
h, xedges, yedges, image = plt.hist2d(x=X_s,y= Y_s, bins = 11)
else:
h, xedges, yedges, image = plt.hist2d(x=X_s,y= Y_s, bins = 11, range = [x_range,x_range])
plt.close()
col_sum = ak.sum(h,axis=-1) #number of events in each (verticle) column
norm_h = [] #norm_h is the normalized matrix
norm_h_text = [] #display labels matrix
for j in range(len(col_sum)):
if col_sum[j] != 0:
norm_c = h[j]/col_sum[j] #normalized column = column values divide by sum of the column
else:
norm_c = h[j]
norm_h.append(norm_c)
norm_c_text = [ '%.3f' % elem for elem in norm_c ] #display value to 3 dp
norm_h_text.append(norm_c_text)
fig, axs = plt.subplots(1, 2, figsize=(20, 10))
if i == 0 and particle in particle_dict.keys(): #Momentum in Single events
axs[0].hist2d(x=X_s,y=Y_s, bins = 11)
else:
axs[0].hist2d(x=X_s,y=Y_s, bins = 11,range = [x_range,x_range])
mplhep.hist2dplot(H=norm_h,norm=mpl.colors.LogNorm(vmin= 1e-4, vmax= 1),labels=norm_h_text, xbins = xedges, ybins = yedges, ax=axs[1])
axs[0].set_title('%s Histogram'%(title_list[i]))
axs[0].set_xlabel('%s_mc'%(title_list[i]))
axs[0].set_ylabel('%s_rc'%(title_list[i]))
axs[1].set_xlabel('%s_mc'%(title_list[i]))
axs[1].set_ylabel('%s_rc'%(title_list[i]))
axs[1].set_title('%s Correlation'%(title_list[i]))
fig.suptitle('%s %s events'%(config,Nevents))
plt.savefig(os.path.join(r_path, '%s_correlation_%s.png' % (title_list[i],config)))
###############
if i > 0:
for j in range(len(X_list)):
X = X_list[j]
Y = Y_list[j]
M_mc = M_list[j]
boolean_M = np.ones_like(M_mc) == 1
X_s = np.array(ak.flatten(X[boolean_M]))
Y_s = np.array(ak.flatten(Y[boolean_M]))
M_s = np.array(ak.flatten(M_mc))
ratio = np.array((ak.Array(Y_s)-(ak.Array(X_s))))
X_plot[j] = M_s
Y_plot[j] = ratio
fig = plt.figure()
gs = fig.add_gridspec(3, 2, wspace=0)
(ax1, ax2), (ax3, ax4),(ax5, ax6) = gs.subplots(sharex=True, sharey=True)
# fig.suptitle('')
ax1.scatter(X_plot[0], Y_plot[0], s = ssize)
ax2.scatter(X_plot[1], Y_plot[1], s = ssize)
ax3.scatter(X_plot[2], Y_plot[2], s = ssize)
ax4.scatter(X_plot[3], Y_plot[3], s = ssize)
ax5.scatter(X_plot[4], Y_plot[4], s = ssize)
ax6.scatter(X_plot[5], Y_plot[5], s = ssize)
ax1.set_xscale('log')
ax1.set_ylabel('rc-mc')
ax3.set_ylabel('rc-mc')
ax5.set_ylabel('rc-mc')
ax5.set_xlabel('Momentum mc')
ax6.set_xlabel('Momentum mc')
ax2.set_title('Pions')
ax3.set_title('Protons')
ax4.set_title('Electrons')
ax5.set_title('Neutrons')
ax6.set_title('Photons')
fig.set_figwidth(20)
fig.set_figheight(10)
ax1.set_title('%s Difference Vs Momentum %s %s events'%(title_list[i],config,Nevents))
plt.savefig(os.path.join(r_path, '%s_difference_vs_momentum_%s.png' % (title_list[i],config)))
################
def particle_plots(boolean_particle):
theta_mc_fil = ak.Array(theta_mc[simID][booll])[boolean_particle]
theta_rc_fil = ak.Array(theta_rc[recID][booll])[boolean_particle]
phi_mc_fil = ak.Array(phi_mc[simID][booll])[boolean_particle]
phi_rc_fil = ak.Array(phi_rc[recID][booll])[boolean_particle]
theta_mc_fil_len = ak.count(theta_mc_fil,axis=None)
theta_rc_fil_len = ak.count(theta_rc_fil,axis=None)
if theta_mc_fil_len > theta_rc_fil_len:
F_boolean = np.ones_like(theta_rc_fil) == 1
else:
F_boolean = np.ones_like(theta_mc_fil) == 1
theta_mc_F = np.array(ak.flatten(theta_mc_fil[F_boolean]))
theta_rc_F = np.array(ak.flatten(theta_rc_fil[F_boolean]))
phi_mc_F = np.array(ak.flatten(phi_mc_fil[F_boolean]))
phi_rc_F = np.array(ak.flatten(phi_rc_fil[F_boolean]))
ratio = np.array((ak.Array(theta_rc_F)-(ak.Array(theta_mc_F))))
fig = plt.figure()
gs = fig.add_gridspec(2, 2, wspace=0.01)
(ax1, ax2), (ax3, ax4) = gs.subplots(sharex=True, sharey=True)
ax1.scatter(-theta_mc_F, ratio, s = ssize)
ax2.scatter(-theta_rc_F, ratio, s = ssize)
ax3.scatter(-theta_mc_F, phi_mc_F, s = ssize)
ax4.scatter(-theta_rc_F, phi_rc_F, s = ssize)
ax1.set_ylabel('rc-mc')
ax2.set_ylabel('rc-mc')
ax3.set_ylabel('Phi mc')
ax4.set_ylabel('Phi rc')
ax1.set_xlabel('- Theta mc')
ax2.set_xlabel('- Theta rc')
ax3.set_xlabel('- Theta mc')
ax4.set_xlabel('- Theta rc')
fig.set_figwidth(20)
fig.set_figheight(10)
if particle in particle_dict.keys():
boolean_particle = particle_dict[particle][0]
particle_name = particle_dict[particle][1]
particle_plots(boolean_particle)
plt.suptitle('%s in %s %s events'%(particle_name,config,Nevents))
plt.savefig(os.path.join(r_path, '%s_%s.png' % (particle_name,config)))
else:
for i in [[boolean_photon,'Photons'],[boolean_electron,'Electrons'],[boolean_pion,'Pions']]:
boolean_particle = i[0]
particle_name = i[1]
particle_plots(boolean_particle)
plt.suptitle('%s in %s %s events'%(particle_name,config,Nevents))
plt.savefig(os.path.join(r_path, '%s_%s.png' % (particle_name,config)))