Configurables

Configurables are helper objects used when constructing steps. They are not graph elements and do not run as separate nodes during simulation.

Unlike most steps, configurables take parameter dictionaries.

Configurable	Purpose
Gaussian	Builds Gaussian kernels for inputs, convolutions, and neural fields.
LateralKernel	Combines multiple kernel objects into one lateral interaction kernel.
Sigmoid	Selects one of JUNIPER's built-in transfer functions.
Transform	Wraps a function that returns a 4x4 coordinate transform matrix.
FrameGraph	Stores named coordinate frames and transform edges.