def alpha_from_angle(y_reco, y_true):
zep, zet, azp, azt = y_reco[:, 1], y_true[:, 1], y_reco[:, 2], y_true[:, 2]
cosalpha = abs(sin(zep)) * cos(azp) * sin(zet) * cos(azt) + abs(
sin(zep)) * sin(azp) * sin(zet) * sin(azt) + cos(zep) * cos(zet)
cosalpha -= tf.math.sign(cosalpha) * eps
alpha = acos(cosalpha)
return alpha
def energy_angle(y_reco, y_true):
energy_metric = tfp.stats.percentile(tf.math.abs(tf.subtract(y_true[:, 0], y_reco[:, 0])), [50-34, 50, 50+34])
#for comparison
classic_stat=tfp.stats.percentile(tf.subtract(y_true[:, 0], y_reco[:, 0]), [25, 75])
w_energy=tf.subtract(classic_stat[1],classic_stat[0])/1.349
alpha= acos(cos_angle(y_reco, y_true))
angle_resi = 180 / np.pi * alpha #degrees
angle_metric = tfp.stats.percentile(angle_resi, [50-34,50,50+34])
w_angle = tfp.stats.percentile(angle_resi, [68])
return energy_metric.numpy(), angle_metric.numpy(), [float(w_energy), float(w_angle)]
def compute_angle_tensor(pts1, pts2):
""" Compute the angle between pt1 and pt2 with respect to the origin
Input:
pts1: batch_size x 1 x 3 tensor
pts2: batch_size x 1 x 3 tensor
"""
b = tf.constant([0.,0.,0.])
angle_diff = []
for pt1, pt2 in zip(pts1, pts2):
ba = tf.subtract(pt1, b)
bc = tf.subtract(pt2, b)
cosine_angle = tm.divide(tf.tensordot(ba, bc, 1), tm.multiply(tf.norm(ba), tf.norm(bc)))
angle = tm.acos(cosine_angle)
angle_diff.append(tf.cast(angle, tf.float32))
return tf.stack(angle_diff, axis=0)
def angle_loss_fn_batch(y_true, y_pred):
x_p = math.sin(y_pred[:, 0]) * math.cos(y_pred[:, 1])
y_p = math.sin(y_pred[:, 0]) * math.sin(y_pred[:, 1])
z_p = math.cos(y_pred[:, 0])
x_t = math.sin(y_true[:, 0]) * math.cos(y_true[:, 1])
y_t = math.sin(y_true[:, 0]) * math.sin(y_true[:, 1])
z_t = math.cos(y_true[:, 0])
norm_p = math.sqrt(x_p * x_p + y_p * y_p + z_p * z_p)
norm_t = math.sqrt(x_t * x_t + y_t * y_t + z_t * z_t)
dot_pt = x_p * x_t + y_p * y_t + z_p * z_t
angle_value = dot_pt / (norm_p * norm_t)
angle_value = tf.clip_by_value(angle_value, -0.99999, 0.99999)
loss_val = (math.acos(angle_value))
return loss_val
def angle_loss_fn(self, y_true, y_pred):
x_p = math.sin(y_pred[:, 0]) * math.cos(y_pred[:, 1])
y_p = math.sin(y_pred[:, 0]) * math.sin(y_pred[:, 1])
z_p = math.cos(y_pred[:, 0])
x_t = math.sin(y_true[:, 0]) * math.cos(y_true[:, 1])
y_t = math.sin(y_true[:, 0]) * math.sin(y_true[:, 1])
z_t = math.cos(y_true[:, 0])
norm_p = math.sqrt(x_p * x_p + y_p * y_p + z_p * z_p)
norm_t = math.sqrt(x_t * x_t + y_t * y_t + z_t * z_t)
dot_pt = x_p * x_t + y_p * y_t + z_p * z_t
angle_value = dot_pt / (norm_p * norm_t)
angle_value = tf.clip_by_value(angle_value, -0.99999, 0.99999)
loss_val = (math.acos(angle_value))
# tf.debugging.check_numerics(
# loss_val, "Vse propalo", name=None
# )
# print(loss_val.shape)
return loss_val