# likelihood_measure.calc_likelihood(hurst_related_exponent_local,
# characteristic_dist_local,
# fft_well_log,
# list_wave_number,
# squared_energy_of_window,
# variance_gaussian_noise_time_domain)
# return output
if __name__ == '__main__':
# ###############################################
# Choose test to run
test_to_run = 'likelihood'
# ###############################################
for case in math_op.switch(test_to_run):
if case('likelihood'):
# ##################################################################
# USER CONFIG: choose run parameters
# ##################################################################
do_parallel = 0
num_value_well_log = 2 ** 12
no_noise = 0 # if true, no noise in observed well log e.g well log is exactly von karman spectrum
hurst_related_exponent_true = 0.5 # (-0.25, -0.25, 0.25, 0.5, 0.75)
characteristic_dist_true = 5.0 # [m] # (10.0, 5.00, 10.00, 5.00, 3)
mean_gaussian_noise_time_domain = 0.0
variance_gaussian_noise_time_domain_true = 1.0e-10
hurst_hurst_related_exponent_guess_start = max(hurst_related_exponent_true - 0.45, -0.49)
print 'l2_regression_powers_error = %g' % l2_regression_powers_error
print 'l2_median_balance_powers_error / l2_regression_powers_error = %g' % (l2_median_balance_powers_error/l2_regression_powers_error)
return output
if __name__ == '__main__':
##########choose test##########################
test_to_run = 'median_balance'
#################################################3
for case in math_op.switch(test_to_run):
if case('compare_medianbal_vs_regression'):
list_of_power = np.array([2]*10)
output = compareMedianBalanceVsRegressionGivenGroundTruth(
list_of_power,
num_time_sample = 2000,
origin_time_sample = 1,
delta_time_sample = 0.002,
width_parameter_ricker_wavelet = 10,
noise_level = 0)
powers_col_by_col = np.transpose(np.vstack((output.list_median_balance_recovered_power,output.list_regression_recovered_power)))
plt.figure()
plt.hist(powers_col_by_col,
histtype='bar',
label=['median balance', 'regression'])
plt.legend()