WidthEstimate1D

turbustat.statistics.WidthEstimate1D(inList, method='walk-down')

Find widths from spectral eigenvectors. These eigenvectors should already be normalized. Widths are defined by the location where 1/e of the maximum occurs.

Note

If the spectral dimension is small in the given eigenvectors (i.e., their length), the 1/e level might not be reached. If this is the case, try padding the initial data cube with zeros in the spectral dimension. The effect on the results should be minimal, as the additional eigenvalues from the padding will be zero. This is especially important when using walk-down.

Warning

Error estimation is not implemented for interpolate.

Parameters:

inList: {list of 1D `~numpy.ndarray`s, 2D `~numpy.ndarray}

List of normalized eigenvectors, or a 2D array with eigenvectors along the 2nd axis.

method{‘walk-down’, ‘fit’, ‘interpolate’}, optional

The width estimation method to use. The options are ‘fit’, ‘interpolate’, or ‘walk-down’. walk-down starts at the peak, and uses a bisector to estimate where the 1/e level lies between the two nearest points. fit fits a Gaussian to data before the first local minimum. interpolate estimates the 1/e level before the first local minimum.

Returns:

scalesarray

The array of estimated scales with length len(inList)

scale_errorsarray

Uncertainty estimations on the scales.