Welcome to the EvoTorch project! EvoTorch is an open source evolutionary computation library developed at NNAISENSE, built on top of PyTorch.
Get started by installing EvoTorch:
pip install evotorch
With EvoTorch, one can solve various optimization problems, without having to worry about whether or not these problems at hand are differentiable. Among the problem types that are solvable with EvoTorch are:
- Black-box optimization problems (continuous or discrete)
- Reinforcement learning tasks
- Supervised learning tasks
- etc.
Various evolutionary computation algorithms are available in EvoTorch:
- Distribution-based search algorithms:
- PGPE: Policy Gradients with Parameter-based Exploration.
- XNES: Exponential Natural Evolution Strategies.
- SNES: Separable Natural Evolution Strategies.
- CEM: Cross Entropy Method.
- Population-based search algorithms:
- SteadyStateGA: A fully elitist genetic algorithm implementation. Also supports multiple objectives, in which case behaves like NSGA-II.
- CoSyNE: Cooperative Synapse Neuroevolution.
All these algorithms mentioned above are implemented in PyTorch, and therefore, can benefit from vectorization and GPU capabilities of PyTorch. In addition, with the help of the Ray library, EvoTorch can further scale up these algorithms by splitting the workload across:
- multiple CPUs
- multiple GPUs
- multiple computers over a Ray cluster
Below are some code examples which demonstrates the API of EvoTorch.
Any objective function defined to work with PyTorch can be used directly with EvoTorch. A non-vectorized objective function simply receives a solution as a 1-dimensional torch tensor, and returns a fitness as a scalar. A vectorized objective function receives a batch of solutions as a 2-dimensional torch tensor, and returns a 1-dimensional tensor storing the fitnesses. The following example demonstrates how to define and solve the classical Rastrigin problem.
from evotorch import Problem
from evotorch.algorithms import SNES
from evotorch.logging import StdOutLogger, PandasLogger
import math
import matplotlib.pyplot as plt
import torch
# Declare the objective function
def rastrigin(x: torch.Tensor) -> torch.Tensor:
A = 10
(_, n) = x.shape
return A * n + torch.sum((x ** 2) - A * torch.cos(2 * math.pi * x), 1)
# Declare the problem
problem = Problem(
"min",
rastrigin,
initial_bounds=(-5.12, 5.12),
solution_length=100,
vectorized=True,
# device="cuda:0" # enable this line if you wish to use GPU
)
# Initialize the SNES algorithm to solve the problem
searcher = SNES(problem, popsize=1000, stdev_init=10.0)
# Initialize a standard output logger, and a pandas logger
_ = StdOutLogger(searcher, interval=10)
pandas_logger = PandasLogger(searcher)
# Run SNES for the specified amount of generations
searcher.run(2000)
# Get the progress of the evolution into a DataFrame with the
# help of the PandasLogger, and then plot the progress.
pandas_frame = pandas_logger.to_dataframe()
pandas_frame["best_eval"].plot()
plt.show()
The following example demonstrates how to solve reinforcement learning tasks that are available through the gym library.
from evotorch.algorithms import PGPE
from evotorch.logging import StdOutLogger
from evotorch.neuroevolution import GymNE
# Declare the problem to solve
problem = GymNE(
env_name="Humanoid-v4", # Solve the Humanoid-v4 task
network="Linear(obs_length, act_length)", # Linear policy
observation_normalization=True, # Normalize the policy inputs
decrease_rewards_by=5.0, # Decrease each reward by 5.0
num_actors="max", # Use all available CPUs
# num_actors=4, # Explicit setting. Use 4 actors.
)
# Instantiate a PGPE algorithm to solve the problem
searcher = PGPE(
problem,
# Base population size
popsize=200,
# For each generation, sample more solutions until the
# number of simulator interactions reaches this threshold
num_interactions=int(200 * 1000 * 0.75),
# Stop re-sampling solutions if the current population size
# reaches or exceeds this number.
popsize_max=3200,
# Learning rates
center_learning_rate=0.0075,
stdev_learning_rate=0.1,
# Radius of the initial search distribution
radius_init=0.27,
# Use the ClipUp optimizer with the specified maximum speed
optimizer="clipup",
optimizer_config={"max_speed": 0.15},
)
# Instantiate a standard output logger
_ = StdOutLogger(searcher)
# Run the algorithm for the specified amount of generations
searcher.run(500)
# Get the center point of the search distribution,
# obtain a policy out of that point, and visualize the
# agent using that policy.
center_solution = searcher.status["center"]
trained_policy = problem.make_net(center_solution)
problem.visualize(trained_policy)
More examples can be found here.