Cosine

class tinygp.kernels.quasisep.Cosine(scale: JAXArray | float, sigma: JAXArray | float = <factory>)

Bases: Quasisep

A scalable implementation of tinygp.kernels.stationary.Cosine

This kernel takes the form:

\[k(\tau)=\sigma^2\,\cos(-2\,\pi\,\tau/\ell)\]

for \(\tau = |x_i - x_j|\).

Parameters:

• scale – The parameter \(\ell\).

• sigma (optional) – The parameter \(\sigma\). Defaults to a value of 1. Specifying the explicit value here provides a slight performance boost compared to independently multiplying the kernel with a prefactor.

coord_to_sortable(X: tinygp.helpers.JAXArray) → tinygp.helpers.JAXArray

A helper function used to convert coordinates to sortable 1-D values

By default, this is the identity, but in cases where X is structured (e.g. multivariate inputs), this can be used to appropriately unwrap that structure.

design_matrix() → tinygp.helpers.JAXArray

The design matrix for the process

evaluate(X1: tinygp.helpers.JAXArray, X2: tinygp.helpers.JAXArray) → tinygp.helpers.JAXArray

The kernel evaluated via the quasiseparable representation

evaluate_diag(X: tinygp.helpers.JAXArray) → tinygp.helpers.JAXArray

For quasiseparable kernels, the variance is simple to compute

observation_model(X: tinygp.helpers.JAXArray) → tinygp.helpers.JAXArray

The observation model for the process

stationary_covariance() → tinygp.helpers.JAXArray

The stationary covariance of the process

to_general_qsm(X1: tinygp.helpers.JAXArray, X2: tinygp.helpers.JAXArray) → GeneralQSM

The generalized quasiseparable representation of this kernel

to_symm_qsm(X: tinygp.helpers.JAXArray) → SymmQSM

The symmetric quasiseparable representation of this kernel

transition_matrix(X1: tinygp.helpers.JAXArray, X2: tinygp.helpers.JAXArray) → tinygp.helpers.JAXArray

The adjoint transition matrix between two coordinates.

If a column-state mean propagates from X1 to X2 as m2 = F @ m1, this method must return F.T. Equivalently, tinygp’s Kalman implementation propagates means using transition_matrix(X1, X2).T @ m1.