_spectra

cultivars.engine._core._spectra = <module 'cultivars.engine._core._spectra' from '/home/runner/work/cultivars/cultivars/src/cultivars/engine/_core/_spectra.py'>

grids, transfer functions, spectral matrices.

Stated once so no two frequency-domain consumers can disagree about a convention. The conventions are: frequencies live on [0, pi] in radians per observation (the process is real, so the negative half-circle is the conjugate mirror and reporting it would be decoration); the transfer function uses e**(-i omega l) for lag l; and the spectral density carries the 1 / (2 pi) normalization, so integrating it over [-pi, pi] – twice its real part over [0, pi] – returns the autocovariance at lag zero. Everything here is a closed-form map from a fitted model’s (coefficients, sigma_u); nothing touches data.

Type:

Frequency-domain primitives