def testcw(fname, t_initial, t_final, two_pulse_fit, one_pulse,
sigma0, sum_mu, sum_tau, sum_a, sum_b, diff_tau, diff_a, diff_b,
sampling, burn, thin, height_th, Plot=True, debug=False):
result, mcmc_flag, unequal_edges = fit_two_cw(*hpa.trace_extr(fname, height_th,
t_initial=t_initial, t_final=t_final),
sigma0=sigma0, # signal noise
sum_mu=sum_mu,
sum_tau=sum_tau,
sum_a=sum_a,
sum_b=sum_b,
diff_tau=diff_tau,
diff_a=diff_a,
diff_b=diff_b,
two_pulse_fit=two_pulse_fit,
one_pulse=one_pulse,
height_th=height_th,
sampling=sampling,
burn=burn,
thin=thin,
Plot=Plot,
debug=debug)
# pymc.gelman_rubin() #https://pymc-devs.github.io/pymc/modelchecking.html
# print 'test'
if Plot:
plt.figure(figsize=(10,5))
plt.plot(*hpa.trace_extr(fname,height_th,t_initial,t_final), label='unwindowed', color='grey')
result.plot_fit(data_kws={'alpha':0.5, 'marker':'.'})
# plt.savefig(results_directory+'testmcmccw.pdf')
print result.fit_report()
print ('arrival time difference = {:.2f}ns'.\
format((result.best_values['one_x_offset']-result.best_values['two_x_offset'])*1e9))
return result, mcmc_flag, unequal_edges
# def testcw(i, sampling, burn, thin, height_th, Plot=True):
# result, mcmc_flag, unequal_edges = fit_two_cw(*hpa.trace_extr(filelist_cont[:10000][mask_2ph_cont][i],
# t_initial, t_final),
# height_th=height_th,
# sampling=sampling,
# burn=burn,
# thin=thin,
# debug=True)
# # pymc.gelman_rubin() #https://pymc-devs.github.io/pymc/modelchecking.html
# if Plot:
# plt.figure()
# plt.plot(*hpa.trace_extr(filelist_cont[mask_2ph_cont][i]), label='unwindowed', color='grey', alpha=0.5)
# result.plot_fit(data_kws={'alpha':0.5, 'marker':'.'})
# plt.savefig(results_directory+'testmcmccw.pdf')
# print result.fit_report()
# print ('arrival time difference = {:.2f}ns'.\
# format((result.init_values['one_x_offset']-result.init_values['two_x_offset'])*1e9))
# return result, mcmc_flag, unequal_edges
def trace_simple_ave(filelist, height_th, t_initial=None, t_final=None):
time, _ = hpa.trace_extr(filelist[0], t_initial, t_final)
a = [hpa.trace_extr(file, t_initial, t_final)[1] for file in filelist]
# reduce to trace length to avoid edge effects
idx_0 = int(len(time) / 30)
v_len = len(time) - 2 * idx_0
time = time[idx_0:idx_0 + v_len]
a = [line[idx_0:idx_0 + v_len] for line in a]
amean = np.nanmean(a, 0)
bg = np.median(amean[amean < height_th])
return time, amean - bg, np.std(amean - bg)
def fit_corrected_pulse(filelist, fit_model, t_initial=None, t_final=None):
time, _ = hps.trace_extr(filelist[0], t_initial, t_final)
a = [
fit_shift(*hps.trace_extr(file, t_initial, t_final),
fit_model=fit_model) for file in filelist
]
# reduce to trace length to avoid edge effects
idx_0 = int(len(time) / 30)
v_len = len(time) - 2 * idx_0
time = time[idx_0:idx_0 + v_len]
a = [line[idx_0:idx_0 + v_len] for line in a]
return time, np.nanmean(a, 0)
def trace_ave(filelist, height_th, t_initial=None, t_final=None, smooth=201):
time, _ = hps.trace_extr(filelist[0], height_th, t_initial, t_final)
a = [
time_offset(*hps.trace_extr(file, height_th, t_initial, t_final),
height_th=height_th) for file in filelist
]
# reduce to trace length to avoid edge effects
idx_0 = int(len(time) / 30)
v_len = len(time) - 2 * idx_0
time = time[idx_0:idx_0 + v_len]
a = [line[idx_0:idx_0 + v_len] for line in a]
return time, savgol_filter(np.nanmean(a, 0), smooth, 3)
def fit_corrected_pulse(filelist,
height_th,
fit_model,
t_initial=None,
t_final=None):
time, _ = hps.trace_extr(filelist[0], height_th, t_initial, t_final)
a = [
fit_shift(*hps.trace_extr(file, height_th, t_initial, t_final),
fit_model=fit_model,
height_th=height_th) for file in filelist
]
# reduce to trace length to avoid edge effects
idx_0 = int(len(time) / 30)
v_len = len(time) - 2 * idx_0
time = time[idx_0:idx_0 + v_len]
a = [line[idx_0:idx_0 + v_len] for line in a]
a_avg = np.nanmean(a, 0)
a_std = np.nanstd(a, 0)
# bg = np.median(a[a<0.5*np.max(a)])
# a_fs = savgol_filter(a_avg, 101, 5)
hist_fs = np.histogram(a_avg[a_avg < np.max(a_avg / 2)], 500)
a_avg = a_avg - hist_fs[1][np.argmax(hist_fs[0])]
return time, a_avg, a_std
# def fit_corrected_pulse(filelist, height_th, fit_model, t_initial=None, t_final=None):
# time, _ = hps.trace_extr(filelist[0], height_th, t_initial, t_final)
# a = [fit_shift(*hps.trace_extr(file, height_th, t_initial, t_final),
# fit_model=fit_model)
# for file
# in filelist]
# # reduce to trace length to avoid edge effects
# idx_0 = int(len(time) / 30)
# v_len = len(time) - 2 * idx_0
# time = time[idx_0:idx_0 + v_len]
# a = [line[idx_0:idx_0 + v_len] for line in a]
# a_avg = np.nanmean(a, 0)
# a_std = np.nanstd(a, 0)
# # bg = np.median(a[a<0.5*np.max(a)])
# # a_fs = savgol_filter(a_avg, 101, 5)
# hist_fs = np.histogram(a_avg[a_avg<np.max(a_avg/2)], 500)
# a_avg = a_avg - hist_fs[1][np.argmax(hist_fs[0])]
# return time, a_avg, a_std
def g2(filelist, t_initial=None, t_final=None):
return [time_diff(*hps.trace_extr(file, t_initial, t_final))
for file
in filelist]
params=p,
# weights=1 / yerr,
# method=method
)
return result, result_mcmc
"""Import source traces"""
directory_name_source = '/workspace/projects/TES/data/20170126_TES5_n012_distinguishibility_20MHz_tau_vs_offset/200ns/'
filelist_200ns = np.array(glob.glob(directory_name_source + '*.trc'))
data_200ns = np.array([
hpa.param_extr(f, t_initial=None, t_final=None, h_th=height_th)
for f in tqdm.tqdm(filelist_200ns[:10000])
])
"""Testing mcmc on one trace"""
mdl = fit_two_mcmc(*hpa.trace_extr(filelist_200ns[mask_2ph_200ns][0],
t_initial, t_final),
sampling=1.5e4)
def testmcmc(i, iters, height_th, Plot=True):
result, result_mcmc = fit_two(*trace_extr(
filelist_200ns[mask_2ph_200ns][i],
t_initial,
t_final,
height_th=height_th),
sampling=iters)
# pymc.gelman_rubin() #https://pymc-devs.github.io/pymc/modelchecking.html
if Plot:
plt.figure()
plt.plot(*trace_extr(filelist_200ns[mask_2ph_200ns][i]),
label='unwindowed')