# MIT License
#
# Copyright (c) 2024 DALabNOVA
#
# Permission is hereby granted, free of charge, to any person obtaining a copy
# of this software and associated documentation files (the "Software"), to deal
# in the Software without restriction, including without limitation the rights
# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
# copies of the Software, and to permit persons to whom the Software is
# furnished to do so, subject to the following conditions:
#
# The above copyright notice and this permission notice shall be included in all
# copies or substantial portions of the Software.
#
# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
# SOFTWARE.
import torch
from scipy.stats import entropy
[docs]
def niche_entropy(repr_, n_niches=10):
"""
Calculate the niche entropy of a population.
Parameters
----------
repr_ : list
The list of individuals in the population.
n_niches : int
Number of niches to divide the population into.
Returns
-------
float
The entropy of the distribution of individuals across niches.
Notes
-----
https://www.semanticscholar.org/paper/Entropy-Driven-Adaptive-RoscaComputer/ab5c8a8f415f79c5ec6ff6281ed7113736615682
https://strathprints.strath.ac.uk/76488/1/Marchetti_etal_Springer_2021_Inclusive_genetic_programming.pdf
"""
num_nodes = [len(ind) - 1 for ind in repr_]
min_ = min(num_nodes)
max_ = max(num_nodes)
pop_size = len(repr_)
stride = (max_ - min_) / n_niches
distributions = []
for i in range(1, n_niches + 1):
distribution = (
sum((i - 1) * stride + min_ <= x < i * stride + min_ for x in num_nodes)
/ pop_size
)
if distribution > 0:
distributions.append(distribution)
return entropy(distributions)
[docs]
def gsgp_pop_div_from_vectors(sem_vectors):
"""
Calculate the diversity of a population from semantic vectors.
Parameters
----------
sem_vectors : torch.Tensor
The tensor of semantic vectors.
Returns
-------
float
The average pairwise distance between semantic vectors.
Notes
-----
https://ieeexplore.ieee.org/document/9283096
"""
return torch.sum(torch.cdist(sem_vectors, sem_vectors)) / (
sem_vectors.shape[0] ** 2
)