LinearGaussianTransition#
- class hybrid_aif.LinearGaussianTransition[source]#
Bases:
Modulex' = F_j x + G_j u + b_j + eps_j: linear map(s) with conditioned innovations.The mode index
jruns over the state factor’s discrete parents (seeinnovations);FandGare shared across modes or given per mode (the switching linear dynamical system pattern).- F: Array#
The state map:
(d, d), shared across modes, or(*parent_shape, d, d)matchinginnovations.parent_shape, one map per mode.
- innovations: ConditionalGaussian#
A
ConditionalGaussianwhose parent axes are the state factor’s discrete parents in declared (B_deps) order: its discrete state parents and any discrete control node, each a mode indexj. A state factor with no discrete parent declares the trivial parent shape(1,)(always index 0). Its location is the offsetb_j.
- G: Array | None = None#
The control input map of the state factor’s continuous control nodes,
utheir values concatenated in declared order:(d, du)shared, or(*parent_shape, d, du)per mode on the ruleFfollows;Nonewhen the state factor reads no continuous node.
- noise_prior: InverseGamma | InverseWishart | Parameter | None = None#
A conjugate prior over the innovations’ covariance
Q(InverseGamma, per-dimension variances, orInverseWishart, a full covariance), shared across the modes or with leading axesinnovations.parent_shape(one prior per mode); orParameter(name), reading a noise held inSystemParams.parameters. When set, every consumer reads the innovations’ Cholesky factor lowered from the prior’s expected-precision point (transition_point()) and the declaredinnovations.cholis ignored (only its shape is a declaration);learnupdates the prior from the pairwise residual scatter, under the rate keytransition:<factor>:noise.None:Qis declared, not learned. Design: process-noise-learning.md D1, D4.
- property learnable_noise: bool#
Capability, per-field: the process noise is a posterior (
noise_priorholds one).
- __init__(F, innovations, G=None, noise_prior=None)#