radon_conc_chain = A[:, parameters['nhyper']+parameters['nstate']:]
b = sampler.chain[:, :, parameters['nhyper']+parameters['nstate']:]
df_params_chain = pd.DataFrame(data = params_chain, columns=parameters['variable_parameter_names'])
# calculate net eff and a to c ratio for each row in the chain
net_effs = []
a_to_c = []
for ii in range(len(df_params_chain)):
pm = {}
pm.update(parameters)
for k in df_params_chain.columns:
pm[k] = df_params_chain.iloc[ii][k]
ne = tm.calc_detector_efficiency(pm)
ac = tm.calc_detector_activity_a_to_c_ratio(pm)
net_effs.append(ne)
a_to_c.append(ac)
df_params_chain['net_eff'] = net_effs
df_params_chain['a_to_c'] = a_to_c
chains_list.append(df_params_chain)
# how does the optimised solution look?
p_mod = dict()
p_mod.update(parameters)
p_mod.update(df_params_chain.mean())
deconv_radon_conc = radon_conc_chain.mean(axis=0)
from scipy.stats import scoreatpercentile
dfss['lldmod_opt'] = np.r_[np.NaN,
parameters.pop('transform_radon_timeseries')
parameters.pop('total_efficiency')
parameters.pop('total_efficiency_frac_error')
# calibration parameters
parameters['cal_source_strength'] = source_strength
parameters['cal_begin'] = cal_begin
if not site == 'ri':
parameters['cal_duration'] = (dfss.cal_radon_conc > 0).sum() * 1800.0
else:
parameters['cal_duration'] = cal_duration
# check net sensitivity/efficiency
ne = tm.calc_detector_efficiency(parameters)
# set eff so that net eff equals expectation
parameters['eff'] = parameters['eff'] * expected_net_eff / ne
# times relative to injection
times = dfss.index.to_pydatetime()
tzero = times[0]
t = np.array([ (itm-tzero).total_seconds() for itm in times])
tres = t[1] - t[0]
# priors on parameters
variable_parameter_names = 'Q_external', 'Q', 'rs', 'lamp', 't_delay', 'eff'
parameters['variable_parameter_lower_bounds'] = np.array([0.0, 0.0, 0.0, 0.0, -np.inf, 0.0])
parameters['variable_parameter_upper_bounds'] = np.array([np.inf, np.inf, 2.0, np.inf, np.inf, np.inf])
variable_parameters_mu_prior = np.array(
[parameters[k] for k in variable_parameter_names])
variable_parameters_sigma_prior = np.array([parameters['Q_external'] * 0.02,
