def dar_no_fit(ortho,
normalize,
sigdriv=_sigdriv,
sigdriv_imag=_sigdriv_imag,
**model_params):
dar = DAR(ortho=ortho, normalize=normalize, **model_params)
dar.sigin = _sigin[None, :]
dar.sigdriv = sigdriv[None, :]
dar.sigdriv_imag = sigdriv_imag[None, :]
dar.make_basis()
return dar
def fast_delay(low_sig=signal,
high_sig=None,
mask=None,
fs=fs,
low_fq=low_fq,
*args,
**kwargs):
dar_model = DAR(ordar=10, ordriv=2)
return DelayEstimator(fs=fs,
dar_model=dar_model,
low_fq=low_fq,
low_fq_width=low_fq_width,
random_state=0,
max_delay=0.01 / low_fq,
*args,
**kwargs).fit(low_sig=low_sig,
high_sig=high_sig,
mask=mask)
def fit_dar_and_plot(sig, ax_logl, ax_phase, phi_0, random_state=None):
dar_model = DAR(ordar=10, ordriv=2)
est = DelayEstimator(fs,
dar_model=dar_model,
low_fq=low_fq,
low_fq_width=bandwidth,
random_state=random_state)
est.fit(sig)
est.plot(ax=ax_logl)
# plot the modulation of the best model
est.best_model_.plot(ax=ax_phase, mode='c', frange=plot_fq_range)
ax_phase.set_title('')
ticks = [-np.pi, phi_0, np.pi]
ax_phase.set_xticks(ticks)
ax_phase.set_xticklabels([r'$%s$' % phase_string(d) for d in ticks])
ax_phase.grid('on')
ax_phase.grid(color=(0.5, 0.5, 0.5))
# Prepare the plot for the two figures
fig, axs = plt.subplots(1, 2, figsize=(10, 4))
axs = axs.ravel()
# Extract a low frequency band
sigdriv, sigin, sigdriv_imag = extract_driver(sigs=signal,
fs=fs,
low_fq=low_fq,
bandwidth=low_fq_width,
extract_complex=True,
random_state=0,
fill=2)
# Create a DAR model
# Here we use BIC selection to get optimal hyperparameters (ordar, ordriv)
dar = DAR(ordar=20, ordriv=2, criterion='bic')
# Fit the DAR model
dar.fit(sigin=sigin, sigdriv=sigdriv, sigdriv_imag=sigdriv_imag, fs=fs)
# Plot the BIC selection
bic_array = dar.model_selection_criterions_['bic']
lines = axs[0].plot(bic_array)
axs[0].legend(lines, ['ordriv=%d' % d for d in [0, 1, 2]])
axs[0].set_xlabel('ordar')
axs[0].set_ylabel('BIC / T')
axs[0].set_title('BIC order selection')
axs[0].plot(dar.ordar_, bic_array[dar.ordar_, dar.ordriv_], 'ro')
# Plot the modulation extracted by the optimal model
dar.plot(ax=axs[1])
axs[1].set_title(dar.get_title(name=True))