Function _prep_data_for_plot

mne/time_frequency/tfr.py:4243–4285

(
    data,
    times,
    freqs,
    *,
    tmin=None,
    tmax=None,
    fmin=None,
    fmax=None,
    baseline=None,
    mode=None,
    dB=False,
    taper_weights=None,
    verbose=None,
)

@verbose
def _prep_data_for_plot(
    data,
    times,
    freqs,
    *,
    tmin=None,
    tmax=None,
    fmin=None,
    fmax=None,
    baseline=None,
    mode=None,
    dB=False,
    taper_weights=None,
    verbose=None,
):
    # baseline
    copy = baseline is not None
    data = rescale(data, times, baseline, mode, copy=copy, verbose=verbose)
    # crop times
    time_mask = np.nonzero(_time_mask(times, tmin, tmax))[0]
    times = times[time_mask]
    # crop freqs
    freq_mask = np.nonzero(_time_mask(freqs, fmin, fmax))[0]
    freqs = freqs[freq_mask]
    # crop data
    data = data[..., freq_mask, :][..., time_mask]
    # handle unaggregated multitaper (complex or phase multitaper data)
    if taper_weights is not None:  # assumes a taper dimension
        logger.info("Aggregating multitaper estimates before plotting...")
        if np.iscomplexobj(data):  # complex coefficients → power
            data = _tfr_from_mt(data, taper_weights)
        else:  # tapered phase data → weighted phase data
            # channels, tapers, freqs, time
            assert data.ndim == 4
            # weights as a function of (tapers, freqs)
            assert taper_weights.ndim == 2
            data = (data * taper_weights[np.newaxis, :, :, np.newaxis]).mean(axis=1)
    # handle remaining complex amplitude → real power
    if np.iscomplexobj(data):
        data = (data * data.conj()).real
    if dB:
        data = 10 * np.log10(data)
    return data, times, freqs


def _tfr_from_mt(x_mt, weights):