template_correlate

specutils.analysis.template_correlate(observed_spectrum, template_spectrum, lag_units=Unit('km / s'), apodization_window=0.5, resample=True, method='direct')

Compute cross-correlation of the observed and template spectra.

After re-sampling into log-wavelength, both observed and template spectra are apodized by a Tukey window in order to minimize edge and consequent non-periodicity effects and thus decrease high-frequency power in the correlation function. To turn off the apodization, use alpha=0.

Parameters:

observed_spectrumSpectrum1D

The observed spectrum.

template_spectrumSpectrum1D

The template spectrum, which will be correlated with the observed spectrum.

lag_units: `~astropy.units.Unit`

Must be a unit with velocity physical type for lags in velocity. To output the lags in redshift, use u.dimensionless_unscaled.

apodization_window: float, callable, or None

If a callable, will be treated as a window function for apodization of the cross-correlation (should behave like a windows window function, with sym=True). If a float, will be treated as the alpha parameter for a Tukey window (tukey), in units of pixels. If None, no apodization will be performed

resample: bool or dict

If True or a dictionary, resamples the spectrum and template following the process in template_logwl_resample. If a dictionary, it will be used as the keywords for template_logwl_resample. For example, resample=dict(delta_log_wavelength=.1) would be the same as calling template_logwl_resample(spectrum, template, delta_log_wavelength=.1). If False, no resampling is performed (and the user is responsible for a sensible resampling).

method: str

If you choose “FFT”, the correlation will be done through the use of convolution and will be calculated faster (for small spectral resolutions it is often correct), otherwise the correlation is determined directly from sums (the “direct” method in correlate).

Returns:

(Quantity, Quantity)

Arrays with correlation values and lags in km/s