Module ssm :: Class StochasticSM

Class StochasticSM

sm.SM --+
        |
       StochasticSM

Known Subclasses:

StochasticSMWithStateObservation

Stochastic state machine.

Instance Methods

__init__(self, startDistribution, transitionDistribution, observationDistribution, beliefDisplayFun=None, sensorDisplayFun=None)

startState(self)

getNextValues(self, state, inp)
Default version of this method.

Inherited from sm.SM: check, doTraceTasks, done, getStartState, guaranteeName, isDone, printDebugInfo, run, start, step, transduce, transduceF

Class Variables
Inherited from `sm.SM`: `legalInputs`, `name`

Instance Variables
	startDistribution `dist.DDist` over states.
	transitionDistribution A procedure that takes an action and returns a procedure, which takes an old state and returns a distribution over new states.
	observationDistribution A procedure that takes a state and returns a distribution over observations.
	beliefDisplayFun (Optional) function that is not used here, but that state estimator, for example, might call to display a belief state.
	sensorDisplayFun (Optional) function that is not used here, but that state estimator, for example, might call to display a sensor likelihoods.

Method Details

init(self, startDistribution, transitionDistribution, observationDistribution, beliefDisplayFun=None, sensorDisplayFun=None)
(Constructor)

Parameters:

transitionDistribution - P(S_t+1 | S_t, A_t) represented as a procedure that takes an action and returns a procedure. The returned procedure takes an old state and returns a distribution over new states.
observationDistribution - P(O_t | S_t) represented as a procedure that takes a state and returns a distribution over observations.
startDistribution - distribution on states, represented as a dist.DDist
beliefDisplayFun - optional function that is not used here, but that state estimator, for example, might call to display a belief state. Takes a belief state {dist.DDist} as input.
sensorDisplayFun - optional function that is not used here, but that state estimator, for example, might call to display a sensor likelihoods. Takes an observation as input.

startState(self)

Overrides: startState

getNextValues(self, state, inp)

Default version of this method. If a subclass only defines getNextState, then we assume that the output of the machine is the same as its next state.

Overrides: sm.SM.getNextValues: (inherited documentation)

Instance Variable Details

beliefDisplayFun

(Optional) function that is not used here, but that state estimator, for example, might call to display a belief state. Takes a belief state {dist.DDist} as input.

sensorDisplayFun

(Optional) function that is not used here, but that state estimator, for example, might call to display a sensor likelihoods. Takes an observation as input.

Class StochasticSM

__init__(self, startDistribution, transitionDistribution, observationDistribution, beliefDisplayFun=None, sensorDisplayFun=None) (Constructor)

startState(self)

getNextValues(self, state, inp)

beliefDisplayFun

sensorDisplayFun

init(self, startDistribution, transitionDistribution, observationDistribution, beliefDisplayFun=None, sensorDisplayFun=None)
(Constructor)