Delayed, noisy channels

feedbax.channel.ChannelState (Module)

Type of state PyTree operated on by Channel instances.

Attributes:

Name	Type	Description
output	PyTree[Array, T]	The current output of the channel.
queue	Tuple[Optional[PyTree[Array, T]], ...]	A tuple of previous inputs to the channel, with the most recent appearing last.
noise	PyTree[Array, T]	The noise added to the current output, if any.

feedbax.channel.ChannelSpec (Module, Generic[StateT])

Specifies how to build a Channel, with respect to the state PyTree of its owner.

Attributes:

Name	Type	Description
where	Callable[[StateT], PyTree[Array]]	A function that selects the subtree of feedback states.
delay	int	The number of previous inputs to store in the queue.
noise_std	int	The standard deviation of the noise to add to the output.

feedbax.channel.Channel (AbstractStagedModel[ChannelState])

A noisy queue.

Note

This can be used for modeling an axon, tract, wire, or other delayed and semi-reliable connection between model components.

Attributes:

Name	Type	Description
delay	int	The number of previous inputs stored in the queue. May be zero.
noise_std	int	The standard deviation of the noise added to the output. If None, no noise is added.
input_proto	PyTree[Array]	A PyTree of arrays with the same structure/shapes as the inputs to the channel will have.
intervenors	ModelIntervenors[ChannelState]	Intervenors to add to the model at construction time.

model_spec: OrderedDict[str, ModelStage[Self, ChannelState]] property

Returns an OrderedDict that specifies the stages of the channel model.

Always includes a queue input-output stage. Optionally includes a stage that adds noise to the output, if noise_std was not None at construction time.

__call__ (input: ModelInput,state: StateT,key: PRNGKeyArray) -> StateT

Inherited from feedbax._staged.AbstractStagedModel

state_consistency_update (state: StateT) -> StateT

Inherited from feedbax._model.AbstractModel

init (*,key: PRNGKeyArray) -> ChannelState

Returns an empty ChannelState for the channel.

change_input (input_proto: PyTree[Array]) -> Channel

Returns a similar Channel with a changed input structure.