# substrate_generator/unit_functions.py
"""Library of functions for single units."""
import numpy as np
from . import utils
[docs]
def logical_or_function(element, weights, state, **kwargs):
return utils.total_weighted_input(element, weights, state) >= 1
[docs]
def logical_and_function(element, weights, state, **kwargs):
# Convention: i,j means i -> j
num_inputs = (weights[:, element] > 0).sum()
return utils.total_weighted_input(element, weights, state) >= num_inputs
[docs]
def logical_parity_function(element, weights, state, **kwargs):
return utils.total_weighted_input(element, weights, state) % 2 >= 1
[docs]
def logical_nor_function(element, weights, state, **kwargs):
return not (logical_or_function(element, weights, state))
[docs]
def logical_nand_function(element, weights, state, **kwargs):
return not (logical_and_function(element, weights, state))
[docs]
def logical_nparity_function(element, weights, state, **kwargs):
return not (logical_parity_function(element, weights, state))
[docs]
def naka_rushton(element, weights, state, exponent=2.0, threshold=1.0, **kwargs):
x = utils.total_weighted_input(element, weights, state) ** exponent
return x / (x + threshold)
[docs]
def boolean_function(element, weights, state, on_inputs=(), **kwargs):
"""An arbitrary boolean function of the element's inputs.
The element is ON exactly when the tuple of its (weighted) input states is
one of ``on_inputs``. All weights must be 0 or 1.
Parameters
----------
element : int
Index of the element whose output is being computed.
weights : numpy.ndarray
The weight matrix (entries restricted to 0 or 1).
state : numpy.ndarray
The state of the substrate.
on_inputs : tuple of tuple, optional
The input patterns for which the element is ON. All patterns must have
the same length, which must equal the number of nonzero input weights.
Returns
-------
bool
The output of the element.
Raises
------
NotImplementedError
If any weight is neither 0 nor 1.
ValueError
If the ``on_inputs`` patterns differ in length, or their length does not
match the number of nonzero input weights.
"""
if np.any((weights != 1) & (weights != 0)):
raise NotImplementedError("weights must be 0 or 1")
if len(set(map(len, on_inputs))) != 1:
raise ValueError("on_inputs must all be the same length")
inputs = tuple(utils.weighted_inputs(element, weights, state))
# Get the length of the first on_input, or use len(inputs) if on_inputs is empty
first_on_input = next(iter(on_inputs), inputs)
if len(inputs) != len(first_on_input):
raise ValueError("nonzero input weights and on_input lengths must match")
return inputs in on_inputs
[docs]
def gauss(x, mu, sigma):
return np.exp(-0.5 * (((x - mu) / sigma) ** 2))
[docs]
def gaussian(
element,
weights,
state,
mu=0.0,
sigma=0.5,
**kwargs,
):
state = utils.binary2spin(state)
x = utils.total_weighted_input(element, weights, state)
return gauss(x, mu, sigma)
