def plot_fmpe_fit(x, y, y_err, fitter, pixel_id=None):
if pixel_id is None:
pixel_id = ''
else:
pixel_id = str(pixel_id)
m = fitter
n_free_parameters = len(m.list_of_vary_param())
n_dof = len(x) - n_free_parameters
n_events = int(np.sum(y))
# m.draw_contour('sigma_e', 'sigma_s', show_sigma=True, bound=5)
x_fit = np.linspace(np.min(x), np.max(x), num=len(x) * 10)
y_fit = fmpe_pdf_10(x_fit, **m.values)
text = '$\chi^2 / ndof : $ {:.01f} / {}\n'.format(m.fval, n_dof)
text += 'Baseline : {:.02f} $\pm$ {:.02f} [LSB]\n'.format(
m.values['baseline'], m.errors['baseline'])
text += 'Gain : {:.02f} $\pm$ {:.02f} [LSB]\n'.format(m.values['gain'],
m.errors['gain'])
text += '$\sigma_e$ : {:.02f} $\pm$ {:.02f} [LSB]\n'.format(
m.values['sigma_e'], m.errors['sigma_e'])
text += '$\sigma_s$ : {:.02f} $\pm$ {:.02f} [LSB]'.format(
m.values['sigma_s'], m.errors['sigma_s'])
data_text = r'$N_{events}$' + ' : {}\nPixel : {}'.format(n_events,
pixel_id)
fig = plt.figure()
axes = fig.add_axes([0.1, 0.3, 0.8, 0.6])
axes_residual = fig.add_axes([0.1, 0.1, 0.8, 0.2])
axes.step(x, y, where='mid', color='k', label=data_text)
axes.errorbar(x, y, y_err, linestyle='None', color='k')
axes.plot(x_fit, y_fit, label=text, color='r')
y_fit = fmpe_pdf_10(x, **m.values)
axes_residual.errorbar(x, ((y - y_fit) / y_err), marker='o', ls='None',
color='k')
axes_residual.set_xlabel('[LSB]')
axes.set_ylabel('count')
axes_residual.set_ylabel('pull')
# axes_residual.set_yscale('log')
axes.legend(loc='best')
return fig
def pdf(self, x, baseline, gain, sigma_e, sigma_s, a_0, a_1):
params = {
'baseline': baseline,
'gain': gain,
'sigma_e': sigma_e,
'sigma_s': sigma_s,
'a_0': a_0,
'a_1': a_1,
'bin_width': 0
}
return fmpe_pdf_10(x, **params)
def pdf(self, x, baseline, gain, sigma_e, sigma_s, a_0, a_1, a_2, a_3, a_4,
a_5, a_6, a_7, a_8, a_9):
params = {
'baseline': baseline,
'gain': gain,
'sigma_e': sigma_e,
'sigma_s': sigma_s,
'a_0': a_0,
'a_1': a_1,
'a_2': a_2,
'a_3': a_3,
'a_4': a_4,
'a_5': a_5,
'a_6': a_6,
'a_7': a_7,
'a_8': a_8,
'a_9': a_9,
'bin_width': 0
}
return fmpe_pdf_10(x, **params)
def compute_init_fmpe(x, y, y_err, n_pe_peaks, snr=3, min_dist=5, debug=False):
y = y.astype(np.float)
cleaned_y = np.convolve(y, np.ones(min_dist), mode='same')
cleaned_y_err = np.convolve(y_err, np.ones(min_dist), mode='same')
bin_width = np.diff(x)
bin_width = np.mean(bin_width)
if (x != np.sort(x)).any():
raise ValueError('x must be sorted !')
d_y = np.diff(cleaned_y)
indices = np.arange(len(y))
peak_mask = np.zeros(y.shape, dtype=bool)
peak_mask[1:-1] = (d_y[:-1] > 0) * (d_y[1:] < 0)
peak_mask[1:-1] *= (cleaned_y[1:-1] / cleaned_y_err[1:-1]) > snr
peak_indices = indices[peak_mask]
peak_indices = peak_indices[:min(len(peak_indices), n_pe_peaks)]
if len(peak_indices) <= 1:
raise PeakNotFound('Not enough peak found for : \n'
'SNR : {} \t '
'Min distance : {} \n '
'Need a least 2 peaks, found {}!!'.
format(snr, min_dist, len(peak_indices)))
x_peak = x[peak_indices]
y_peak = y[peak_indices]
gain = np.diff(x_peak)
gain = np.average(gain, weights=y_peak[:-1])
sigma = np.zeros(len(peak_indices))
mean_peak_x = np.zeros(len(peak_indices))
amplitudes = np.zeros(len(peak_indices))
distance = int(gain / 2)
if distance < bin_width:
raise ValueError('Distance between peaks must be >= {} the bin width'
''.format(bin_width))
n_x = len(x)
for i, peak_index in enumerate(peak_indices):
left = x[peak_index] - distance
left = np.searchsorted(x, left)
left = max(0, left)
right = x[peak_index] + distance + 1
right = np.searchsorted(x, right)
right = min(n_x - 1, right)
amplitudes[i] = np.sum(y[left:right]) * bin_width
mean_peak_x[i] = np.average(x[left:right], weights=y[left:right])
sigma[i] = np.average((x[left:right] - mean_peak_x[i])**2,
weights=y[left:right])
sigma[i] = np.sqrt(sigma[i] - bin_width**2 / 12)
gain = np.diff(mean_peak_x)
gain = np.average(gain)
sigma_e = np.sqrt(sigma[0]**2)
sigma_s = (sigma[1:] ** 2 - sigma_e**2) / np.arange(1, len(sigma), 1)
sigma_s = np.mean(sigma_s)
sigma_s = np.sqrt(sigma_s)
params = {'baseline': mean_peak_x[0], 'sigma_e': sigma_e,
'sigma_s': sigma_s, 'gain': gain}
for i, amplitude in enumerate(amplitudes):
params['a_{}'.format(i)] = amplitude
if debug:
x_fit = np.linspace(np.min(x), np.max(x), num=len(x)*10)
plt.figure()
plt.step(x, y, where='mid', color='k', label='data')
plt.errorbar(x, y, y_err, linestyle='None', color='k')
plt.plot(x[peak_indices], y[peak_indices], linestyle='None',
marker='o', color='r', label='Peak positions')
plt.plot(x_fit, fmpe_pdf_10(x_fit, bin_width=bin_width, **params),
label='init', color='g')
plt.legend(loc='best')
plt.show()
return params