carma_core#

Overview#

Classes#
`CARMAProcess`	Base class for inference with Continuous autoregressive moving average processes

Classes#

class pioran.carma.carma_core.CARMAProcess(p: int, q: int, observation_indexes: jax.Array, observation_values: jax.Array, observation_errors=None, **kwargs)[source]#

Bases: equinox.Module

Base class for inference with Continuous autoregressive moving average processes

Parameters:

pint: Order of the AR polynomial.
qint: Order of the MA polynomial.
observation_indexesjax.Array: Indexes of the observations.
observation_valuesjax.Array: Values of the observations.
observation_errorsjax.Array: Errors of the observations, if None, the errors are set to sqrt(eps).
kwargsdict: Additional arguments to pass to the CARMA model. AR_quad : jax.Array Quadratic coefficients of the AR polynomial. beta : jax.Array Coefficients of the MA polynomial. use_beta : bool If True, uses the beta coefficients otherwise uses the quadratic coefficients of the MA polynomial. scale_errors : bool If True, scales the errors by a factor nu. estimate_mean : bool If True, estimates the mean of the process.

Attributes:

pint: Order of the AR polynomial.
qint: Order of the MA polynomial.
observation_indexesjax.Array: Indexes of the observations.
observation_valuesjax.Array: Values of the observations.
observation_errorsjax.Array: Errors of the observations, if None, the errors are set to sqrt(eps).
prediction_indexesjax.Array: Indexes of the predictions.
modelCARMA_model: CARMA model.
kalmanKalmanFilter: Kalman filter associated to the CARMA model.
use_betabool: If True, uses the beta coefficients otherwise uses the quadratic coefficients of the MA polynomial.
scale_errorsbool: If True, scales the errors by a factor nu.
estimate_meanbool: If True, estimates the mean of the process.
nb_prediction_pointsint: Number of prediction points.

Overview

Attributes#
`p`	-
`q`	-
`observation_indexes`	-
`observation_values`	-
`observation_errors`	-
`prediction_indexes`	-
`model`	-
`kalman`	-
`use_beta`	-
`estimate_mean`	-
`scale_errors`	-
`nb_prediction_points`	-

Methods#
`compute_predictive_distribution`(**kwargs)	-
`compute_log_marginal_likelihood`()	-
`wrapper_log_marginal_likelihood`(params)	Wrapper to compute the log marginal likelihood in function of the (hyper)parameters.
`__str__`()	String representation of the CARMA object.
`__repr__`()	Return repr(self).