reservoirpy.mat_gen.Initializer#

class reservoirpy.mat_gen.Initializer(func, autorize_sr=True, autorize_input_scaling=True, autorize_rescaling=True)[source]#

Base class for initializer functions. Allow updating initializer function parameters several times before calling. May perform spectral radius rescaling or input scaling as a post-processing to initializer function results.

Parameters:

  • func (callable) – Initializer function. Should have a shape argument and return a Numpy array or Scipy sparse matrix.

  • autorize_sr (bool, default to True) – Authorize spectral radius rescaling for this initializer.

  • autorize_input_scaling (bool, default to True) – Authorize input_scaling for this initializer.

  • autorize_rescaling (bool, default to True) – Authorize any kind of rescaling (spectral radius or input scaling) for this initializer.

Example

>>> from reservoirpy.mat_gen import random_sparse
>>> init_func = random_sparse(dist="uniform")
>>> init_func = init_func(connectivity=0.1)
>>> matrix = init_func(5, 5)  # actually creates the matrix
>>> matrix = random_sparse(5, 5, dist="uniform", connectivity=0.1)  # also creates the matrix

Methods

__init__(func[, autorize_sr, ...])