timecorr.laplace_weights

timecorr.laplace_weights(T, params={'scale': 100})

Generate Laplace (exponential) weighting function for dynamic correlations.

This function creates a time-varying weighting matrix where each timepoint receives weights according to a Laplace distribution centered at that timepoint. Provides sharper temporal localization compared to Gaussian weights.

Parameters:

Tint

Number of timepoints in the timeseries

paramsdict, optional

Dictionary containing Laplace parameters. Default: {‘scale’: 100} - ‘scale’ : float, scale parameter of the Laplace distribution

Returns:

numpy.ndarray

T x T matrix of Laplace weights, where weights[i,j] represents the weight given to timepoint j when computing correlations at timepoint i

Examples

>>> import timecorr as tc
>>> weights = tc.laplace_weights(50, {'scale': 5})
>>> print(weights.shape)  # (50, 50)