class CovarianceModel

class deeptime.covariance.CovarianceModel(cov_00: Optional[ndarray] = None, cov_0t: Optional[ndarray] = None, cov_tt: Optional[ndarray] = None, mean_0: Optional[ndarray] = None, mean_t: Optional[ndarray] = None, bessels_correction: bool = True, symmetrized: bool = False, lagtime: Optional[int] = None, data_mean_removed: bool = False)

A model which in particular carries the estimated covariances, means from a Covariance.

Parameters:

• cov_00 ((n, n) ndarray, optional, default=None) – The instantaneous covariances if computed (see Covariance.compute_c00).

• cov_0t ((n, n) ndarray, optional, default=None) – The time-lagged covariances if computed (see Covariance.compute_c0t).

• cov_tt ((n, n) ndarray, optional, default=None) – The time-lagged instantaneous covariances if computed (see Covariance.compute_ctt).

• mean_0 ((n,) ndarray, optional, default=None) – The instantaneous means if computed.

• mean_t ((n,) ndarray, optional, default=None) – The time-shifted means if computed.

• bessels_correction (bool, optional, default=True) – Whether Bessel’s correction was used during estimation.

• lagtime (int, default=None) – The lagtime that was used during estimation.

• data_mean_removed (bool, default=False) – Whether the data mean was removed. This can have an influence on the effective VAMP score.

Attributes

bessels_correction	Whether Bessel's correction was applied during estimation.
cov_00	The instantaneous covariances.
cov_0t	The time-shifted covariances.
cov_tt	The time-shifted instantaneous covariances.
data_mean_removed	Whether the data mean was removed.
lagtime	The lagtime at which estimation was performed.
mean_0	The instantaneous means.
mean_t	The time-shifted means.
symmetrized	Whether correlations and second moments are symmetrized in time.

Methods

copy()	Makes a deep copy of this model.
get_params([deep])	Get the parameters.
set_params(**params)	Set the parameters of this estimator.
whiten(data[, epsilon, method])	Whiten a (T, N)-shaped chunk of data by transforming it into the PCA basis.

copy() → Model

Makes a deep copy of this model.

Returns:

A new copy of this model.

Return type:

copy

get_params(deep=False)

Get the parameters.

Returns:

params – Parameter names mapped to their values.

Return type:

mapping of string to any

set_params(**params)

Set the parameters of this estimator.

The method works on simple estimators as well as on nested objects (such as pipelines). The latter have parameters of the form <component>__<parameter> so that it’s possible to update each component of a nested object.

Parameters:

**params (dict) – Estimator parameters.

Returns:

self – Estimator instance.

Return type:

object

whiten(data: ndarray, epsilon=1e-10, method='QR') → ndarray

Whiten a (T, N)-shaped chunk of data by transforming it into the PCA basis. In case of rank deficiency this reduces the dimension.

Parameters:

• data ((T,N) ndarray) – The data to be whitened.

• epsilon (float, optional, default=1e-10) – Truncation parameter. See deeptime.numeric.spd_inv_sqrt().

• method (str, optional, default='QR') – Decomposition method. See deeptime.numeric.spd_inv_sqrt().

Returns:

whitened_data – Whitened data.

Return type:

(T, n) ndarray

property bessels_correction: bool

Whether Bessel’s correction was applied during estimation.

Type:

bool

property cov_00: Optional[ndarray]

The instantaneous covariances.

Type:

(n, n) ndarray or None

property cov_0t: Optional[ndarray]

The time-shifted covariances.

Type:

(n, n) ndarray or None

property cov_tt: Optional[ndarray]

The time-shifted instantaneous covariances.

Type:

(n, n) ndarray or None

property data_mean_removed: bool

Whether the data mean was removed.

Type:

bool

property lagtime: Optional[int]

The lagtime at which estimation was performed.

Type:

int or None

property mean_0: Optional[ndarray]

The instantaneous means.

Type:

(n,) ndarray or None

property mean_t: Optional[ndarray]

The time-shifted means.

Type:

(n,) ndarray or None

property symmetrized: bool

Whether correlations and second moments are symmetrized in time.

Type:

bool