def event_mass_4jet(self):
self.event_mass_4jet = []
for event_4_vector in event_stats_added_4:
event_4_list = list(event_4_vector)
event_4_array = np.array(event_4_list)
event_mass = ef.ms_from_p4s(event_4_array)
self.event_mass_4jet.append(event_mass)
def test_ms_from_ps(dim, nevents, nparticles):
masses = np.random.rand(nevents, nparticles)
events = ef.gen_random_events(nevents, nparticles, mass=masses, dim=dim)
masses = masses.reshape(events.shape[:-1])
results = ef.ms_from_ps(events)
assert epsilon_diff(results, masses, 1e-10)
if dim == 4:
assert epsilon_diff(results, ef.ms_from_p4s(events), 1e-10)
def fit_pfn(self, model_settings):
# convert labels to categorical
Y_PFN = energyflow.utils.to_categorical(self.y, num_classes=2)
# preprocess by centering jets and normalizing pts
X_PFN = self.X_particles
for x_PFN in X_PFN:
mask = x_PFN[:,0] > 0
yphi_avg = np.average(x_PFN[mask,1:3], weights=x_PFN[mask,0], axis=0)
x_PFN[mask,1:3] -= yphi_avg
x_PFN[mask,0] /= x_PFN[:,0].sum()
# handle particle id channel
if model_settings['use_pids']:
self.my_remap_pids(X_PFN)
else:
X_PFN = X_PFN[:,:,:3]
# Split data into train, val and test sets
(X_PFN_train, X_PFN_val, X_PFN_test,Y_PFN_train, Y_PFN_val, Y_PFN_test) = energyflow.utils.data_split(X_PFN, Y_PFN,
val=self.n_val, test=self.n_test)
# build architecture
pfn = energyflow.archs.PFN(input_dim=X_PFN.shape[-1],
Phi_sizes=model_settings['Phi_sizes'],
F_sizes=model_settings['F_sizes'])
# train model
pfn.fit(X_PFN_train, Y_PFN_train,
epochs=model_settings['epochs'],
batch_size=model_settings['batch_size'],
validation_data=(X_PFN_val, Y_PFN_val),
verbose=1)
# get predictions on test data
preds_PFN = pfn.predict(X_PFN_test, batch_size=1000)
# Get AUC and ROC curve + make plot
auc_PFN = sklearn.metrics.roc_auc_score(Y_PFN_test[:,1], preds_PFN[:,1])
print('Particle Flow Networks/Deep Sets: AUC = {} (test set)'.format(auc_PFN))
self.roc_curve_dict['PFN'] = sklearn.metrics.roc_curve(Y_PFN_test[:,1], preds_PFN[:,1])
# # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # #
# Now we compare the PFN ROC curve to single observables
# 1. Jet mass (Note: This takes in (pt,y,phi) and converts it to 4-vectors and computes jet mass)
# (Note: X_PFN_train is centered and normalized .. should be ok)
masses = np.asarray([energyflow.ms_from_p4s(energyflow.p4s_from_ptyphims(x).sum(axis=0)) for x in X_PFN_train])
self.roc_curve_dict['Jet_mass'] = sklearn.metrics.roc_curve(Y_PFN_train[:,1], -masses)
# 2. Multiplicity (Is this a useful observable for pp vs AA?)
mults = np.asarray([np.count_nonzero(x[:,0]) for x in X_PFN_train])
self.roc_curve_dict['Multiplicity'] = sklearn.metrics.roc_curve(Y_PFN_train[:,1], -mults)
def test_ms_from_p4s(nevents, nparticles):
p4s = ef.gen_random_events(nevents, nparticles, dim=4,
mass='random').reshape(nevents, nparticles, 4)
ms = ef.ms_from_p4s(p4s)
slow_ms = []
for i in range(nevents):
event_ms = []
for j in range(nparticles):
event_ms.append(
np.sqrt(p4s[i, j, 0]**2 - p4s[i, j, 1]**2 - p4s[i, j, 2]**2 -
p4s[i, j, 3]**2))
slow_ms.append(event_ms)
slow_ms = np.asarray(slow_ms)
assert epsilon_diff(slow_ms, ms)
def test_ys_from_pts_etas_ms(nevents, nparticles):
p4s = ef.gen_random_events(nevents, nparticles, dim=4,
mass='random').reshape(nevents, nparticles, 4)
ys = ef.ys_from_p4s(p4s)
y_primes = ef.ys_from_pts_etas_ms(ef.pts_from_p4s(p4s),
ef.etas_from_p4s(p4s),
ef.ms_from_p4s(p4s))
assert epsilon_diff(ys, y_primes, 1e-12)
# test cutoff
pts, ms = 1000 * np.random.rand(25), 10 * np.random.rand(25)
etas = np.random.choice([-1., 1.], size=25) * 100 * np.random.rand(25)
for c1, c2 in zip(np.linspace(20, 100, 5), 20 + 80 * np.random.rand(5)):
ys_c1 = ef.ys_from_pts_etas_ms(pts, etas, ms, _cutoff=c1)
ys_c2 = ef.ys_from_pts_etas_ms(pts, etas, ms, _cutoff=c2)
assert epsilon_diff(ys_c1, ys_c2, 1e-12)
auc = roc_auc_score(Y_test[:, 1], preds[:, 1])
print()
print('EFN AUC:', auc)
print()
# some nicer plot settings
plt.rcParams['font.family'] = 'serif'
plt.rcParams['figure.autolayout'] = True
fig, axes = plt.subplots(1, 2, figsize=(8, 4))
######################### ROC Curve Plot #########################
# get multiplicity and mass for comparison
masses = np.asarray(
[ef.ms_from_p4s(ef.p4s_from_ptyphims(x).sum(axis=0)) for x in X])
mults = np.asarray([np.count_nonzero(x[:, 0]) for x in X])
mass_fp, mass_tp, threshs = roc_curve(Y[:, 1], -masses)
mult_fp, mult_tp, threshs = roc_curve(Y[:, 1], -mults)
# plot the ROC curves
axes[0].plot(efn_tp, 1 - efn_fp, '-', color='black', label='EFN')
axes[0].plot(mass_tp, 1 - mass_fp, '-', color='blue', label='Jet Mass')
axes[0].plot(mult_tp, 1 - mult_fp, '-', color='red', label='Multiplicity')
# axes labels
axes[0].set_xlabel('Quark Jet Efficiency')
axes[0].set_ylabel('Gluon Jet Rejection')
# axes limits
axes[0].set_xlim(0, 1)
def test_gen_random_events_mcom(nevents, nparticles, dim):
events = ef.gen_random_events_mcom(nevents, nparticles, dim=dim)
assert events.shape == (nevents, nparticles, dim)
assert epsilon_diff(ef.ms_from_p4s(events)**2 / dim, 0, 10**-12)
assert epsilon_diff(np.sum(events, axis=1), 0, 10**-12)
def test_gen_random_events(nevents, nparticles, dim, mass):
events = ef.gen_random_events(nevents, nparticles, dim=dim, mass=mass)
assert events.shape == (nevents, nparticles, dim)
assert epsilon_diff(ef.ms_from_p4s(events)**2, mass**2, 10**-13)
def test_gen_massless_phase_space(nevents, nparticles):
events = ef.gen_massless_phase_space(nevents, nparticles)
assert events.shape == (nevents, nparticles, 4)
assert epsilon_diff(ef.ms_from_p4s(events)**2, 0, 10**-13)
# get ROC curve if we have sklearn
if roc_curve:
cnn_fp, cnn_tp, threshs = roc_curve(Y_test[:,1], preds[:,1])
# get area under the ROC curve
auc = roc_auc_score(Y_test[:,1], preds[:,1])
print()
print('CNN AUC:', auc)
print()
# make ROC curve plot if we have matplotlib
if plt:
# get multiplicity and mass for comparison
masses = np.asarray([ef.ms_from_p4s(ef.p4s_from_ptyphims(x).sum(axis=0)) for x in X])
mults = np.asarray([np.count_nonzero(x[:,0]) for x in X])
mass_fp, mass_tp, threshs = roc_curve(Y[:,1], -masses)
mult_fp, mult_tp, threshs = roc_curve(Y[:,1], -mults)
# some nicer plot settings
plt.rcParams['figure.figsize'] = (4,4)
plt.rcParams['font.family'] = 'serif'
plt.rcParams['figure.autolayout'] = True
# plot the ROC curves
plt.plot(cnn_tp, 1-cnn_fp, '-', color='black', label='CNN')
plt.plot(mass_tp, 1-mass_fp, '-', color='blue', label='Jet Mass')
plt.plot(mult_tp, 1-mult_fp, '-', color='red', label='Multiplicity')
# axes labels
