nengolib.Connection

class nengolib.Connection(pre, post, solver=LstsqL2(reg=0.1, solver=Cholesky(transpose=None), weights=False), **kwargs)[source]

Drop-in replacement for nengo.Connection.

Extends nengo.Connection to improve decoding by using a bias. If the pre object is a nengo.Ensemble, then a bias will be included in its decoders. This automatically improves the decoding of functions with a constant offset when the ensemble has few neurons. This is equivalent to a change in postsynaptic biases (in effect changing the representation to be centered around some constant offset, that is discovered optimally).

Parameters:
pre : nengo.Ensemble or nengo.ensemble.Neurons or nengo.Node

Nengo source object for the connection.

post : nengo.Ensemble or nengo.ensemble.Neurons or nengo.Node or nengo.Probe

Nengo destination object for the connection.

solver : nengo.solvers.Solver, optional

Solver to use for decoded bias and connection. Defaults to nengo.solvers.LstsqL2.

**kwargs : dictionary, optional

Additional keyword arguments passed to nengo.Connection.

Attributes:
eval_points
function
function_info

Connection-specific validation for functions.

is_decoded
label

A parameter where the value is a string.

learning_rule

(LearningRule or iterable) Connectable learning rule object(s).

learning_rule_type

Connection-specific validation for learning rules.

modulatory

A parameter that is no longer supported.

params

Returns a list of parameter names that can be set.

post
post_obj
post_slice
pre
pre_obj
pre_slice
scale_eval_points

A parameter where the value is a boolean.

seed

A parameter where the value is an integer.

size_in

(int) The number of output dimensions of the pre object.

size_mid

(int) The number of output dimensions of the function, if specified.

size_out

(int) The number of input dimensions of the post object.

solver

Connection-specific validation for decoder solvers.

synapse
transform

The transform additionally validates size_out.

Methods

copy