skretrieval.retrieval.prior.VerticalTikhonov

skretrieval.retrieval.prior.VerticalTikhonov#

class skretrieval.retrieval.prior.VerticalTikhonov(order: int, prior_state: array = None, tikhonov: array = None)[source]#

Bases: VerticalPrior

A prior that is constructed as a Tikhonov constraint.

Parameters:

  • order (int) – Order of the Tikhonov constraint, only 1 and 2 are supported

  • prior_state (np.array, optional) – Prior state. If set to None a zero prior is used, by default None

  • tikhonov (np.array, optional) – Array of factors to multiply the constraint by, by default None

__init__(order: int, prior_state: array = None, tikhonov: array = None)[source]#

A prior that is constructed as a Tikhonov constraint.

Parameters:

  • order (int) – Order of the Tikhonov constraint, only 1 and 2 are supported

  • prior_state (np.array, optional) – Prior state. If set to None a zero prior is used, by default None

  • tikhonov (np.array, optional) – Array of factors to multiply the constraint by, by default None

Methods

__init__(order[, prior_state, tikhonov])

A prior that is constructed as a Tikhonov constraint.

init(sv[, sv_slice])

Attributes

inverse_covariance

The inverse covariance of the prior state.

precision_factor

Return R such that R.T @ R is the prior precision.

state

The prior state $x_a$ of size (n,)
property inverse_covariance#

The inverse covariance of the prior state.

The returned matrix has shape (n, n).

property precision_factor#

Return R such that R.T @ R is the prior precision.

Structured priors override this to retain their sparse residual form. The fallback is intended for small or dense custom priors.

property state#

The prior state $x_a$ of size (n,)