Filters Directory Reference

Files
	AmmFilter.h
	AmmFilter (decisions.md [autonomous-multiple-model]): the ImmFilter on the identity chain, whose constructor builds that chain from the state's modes, and the AmmFilter alias.
	FilterUpkeep.h
	The upkeep a filter holds (decisions.md [upkeep]): internal::HeldUpkeep, the base that gives every filter its upkeep() and maintain().
	HiddenMarkovFilter.h
	HiddenMarkovFilter (decisions.md [filters-and-recursions], [emissions], [update-outcomes]): a CategoricalParams state moved by a MarkovChain or a ContinuousTimeMarkovChain and updated by emission models, with the log-evidence of the measurements so far, and HiddenMarkovFilterUpdate.
	ImmFilter.h
	ImmRecursion and ImmFilter (decisions.md [switching-models], [imm], [filters-and-recursions]): a Hybrid state mixed through a Markov chain, predicted through each mode's model, and updated per mode, with Markov outlier states when a measurement's noise is by mode axis (ModeNoise) and a measurement per mode through a SwitchingMeasurement; the recursion that does this to a hybrid passed in, keeping the scratch it mixes and updates into, the filter that owns a hybrid and runs the recursion on it.
	InfoFilter.h
	InfoFilter (decisions.md [filters-and-recursions], [gaussian-forms]): the KalmanFilter over an InfoGaussianParams, spelled space first, whose update adds information and whose prediction goes through the moments.
	KalmanFilter.h
	KalmanRecursion and KalmanFilter (decisions.md [templated-measurements] and [filters-and-recursions]): the Kalman steps over one component as a recursion, a process model and a prediction method acting on a state passed in, and the filter that owns a component and runs that recursion on it, with updates by any measurement reduced to one component by a policy.
	MixtureFilter.h
	MixtureFilter (decisions.md [mixture-filter], [filters-and-recursions]): a Mixture state whose components are each predicted and updated by the recursion their type names, a KalmanRecursion for Gaussians and an ImmRecursion for hybrid states, or by one of the user's own (Recursion.h), and reweighted by their likelihoods, with MixtureFilterUpdate; GaussianMixturePredictedJoint, what predictJoint() records for the mixture smoother [smoothing] in GaussianMixtureHistory.h; and GaussianMixtureFilter, the spelling for the mixture of Gaussians.
	ParticleFilter.h
	ParticleFilter (decisions.md [upkeep]): a ParticleSet moved by sampling the process model and its noise, weighted by the measurement likelihood, with upkeep left to the caller, and ParticleFilterUpdate; its steps also run under an execution policy [execution-policies].
	RbParticleFilter.h
	RbParticleFilter (decisions.md [rb-particle-filter]): particles for one block of the state, each carrying a Kalman filter over the rest, or an IMM filter under a switching model, with a coupled prediction that draws the block from the predicted joint and conditions the rest on it.
	RbPointMassFilter.h
	RbPointMassFilter (decisions.md [rb-point-mass] and [rao-blackwellized-core]): fixed points for one block of the state, each carrying a Kalman filter over the rest, or an IMM filter under a switching model, with the PointMassFilter alias for a grid that covers the whole state.
	RbUpdate.h
	The update that the Rao-Blackwellized filters share (decisions.md [rb-point-mass], [rao-blackwellized-core], [rb-particle-filter]): every point's Gaussian conditioned on a measurement given the point, with one transform for all points under a shared covariance or one Kalman update each, into the reused PointLikelihoods.
	Recursion.h
	The Recursion concept (decisions.md [filters-and-recursions], [user-extension-points]): what a KalmanFilter or a MixtureFilter asks of the object that runs its steps, spelled so that a recursion of the user's own, a method the library doesn't ship, runs inside either filter; and internal::UpdateAccess, through which the filters write the log-likelihood their results hold [update-outcomes].
	SqrtInfoFilter.h
	SqrtInfoFilter (decisions.md [filters-and-recursions], [gaussian-forms]): the KalmanFilter over a SqrtInfoGaussianParams, spelled space first, Bierman's filter on the data array.
	SqrtKalmanFilter.h
	SqrtKalmanFilter (decisions.md [filters-and-recursions], [gaussian-forms]): the KalmanFilter over a SqrtGaussianParams, spelled space first, whose steps triangularize stacked factors and never form a covariance.