get_alpha_and_c

• full name: tenpy.tools.prediction.get_alpha_and_c

• parent module: tenpy.tools.prediction

• type: function

tenpy.tools.prediction.get_alpha_and_c(x, lpc, truncation_mode='cutoff', epsilon=1e-06)

Get the eigenvalues and coefficients from a vector of linear prediction coefficients for the time series x.

This follows the approach taken in arXiv:0901.2342. If necessary, the eigenvalues are truncated according to the truncation_mode

Parameters:

• x (ndarray) – one-dimensional time series data

• lpc (ndarray) – 1D array containing the p linear prediction coefficients [a_p, a_{p-1}, ..., a_1] from the correlations of x

• truncation_mode (str) – the truncation mode (default is ‘cutoff’, which means that those eigenvalues will be cut) used for truncating the eigenvalues. Other options are ‘renormalize’ (meaning their absolute value will be set to 1) and ‘conjugate’

• epsilon (float) – Regularization parameter for matrix inversion. However, if the matrix can’t be inverted here, linear prediction is probably not applicable

Returns:

• evals (ndarray)

• c (ndarray)