import scipy.stats as sci
import numpy as np
import warnings
def norm(x, mu, sigma):
return (x - mu) / sigma
def denorm(x, mu, sigma):
return sigma * x + mu
[docs]def generate_ps_par(h):
"""
Simulate PS parameters modelled after the TUM40 phonon detector.
:param h: Array of pulse heights.
:type h: 1D array
:return: (t0, An, At, tau_n, tau_in, tau_t)
:rtype: list of 6 1D arrays
"""
warnings.warn('Watch out, this function is not fully tested!')
SAMPLE_SIZE = len(h)
h_mean = 0.89
h_std = 0.57
tau_n_mean = 112.64
tau_n_std = 38.04
t0_mean = -1.36
t0_std = 0.29
# An_mean = 1.27
# An_std = 0.81
# At_mean = 1.03
# At_std = 0.66
sigma_tn_mean = 0.718
sigma_tn_std = 0.709
tau_in_mean = 2.52
tau_in_std = 0.058
tau_t_mean = 21.45
tau_t_std = 1.46
# these equations are found with genetic algorithms
# independent var
h_ = norm(h, h_mean, h_std)
# semidependent vars
tau_n_ = sci.uniform.rvs(size=SAMPLE_SIZE,
loc=-0.134 - np.sqrt(3) * denorm(-0.312 * h_, sigma_tn_mean, sigma_tn_std),
scale=2 * np.sqrt(3) * denorm(-0.312 * h_, sigma_tn_mean,
sigma_tn_std)) # underscore means the normed variable
# dependent vars
# An_ = np.copy(h_)
# At_ = np.copy(h_)
t0_ = (h_ - 0.455) / (0.504 * h_ + 1.103) + sci.norm.rvs(size=SAMPLE_SIZE, loc=0,
scale=0.644)
tau_in_ = 0.23 * tau_n_ ** 2 - tau_n_ - 0.331 + sci.norm.rvs(size=SAMPLE_SIZE, loc=0, scale=0.5454978 / 2)
tau_t_ = -0.468 * tau_n_ ** 2 + tau_n_ + 0.372 + sci.norm.rvs(size=SAMPLE_SIZE, loc=0, scale=1.0354906 / 2)
# transform back
t0 = denorm(t0_, t0_mean, t0_std)
An = h / 0.705 # denorm(denorm(An_,h_mean,h_std), An_mean, An_std)
At = h / 0.858 # denorm(denorm(At_,h_mean,h_std), At_mean, At_std)
tau_n = denorm(tau_n_, tau_n_mean, tau_n_std)
tau_in = denorm(tau_in_, tau_in_mean, tau_in_std)
tau_t = denorm(tau_t_, tau_t_mean, tau_t_std)
return np.array([t0, An, At, tau_n, tau_in, tau_t])