EMI-Group/evox

Distributed GPU-Accelerated Framework for Evolutionary Computation. Comprehensive Library of Evolutionary Algorithms & Benchmark Problems.

What it solves

EvoX addresses the performance bottlenecks in evolutionary computation by providing a distributed, GPU-accelerated framework. It enables the scalable execution of evolutionary algorithms (EAs) for complex optimization tasks, including single-objective, multi-objective, and neuroevolution, which typically require massive parallel evaluations.

How it works

Built to be compatible with PyTorch, EvoX uses a tailored programming model that allows evolutionary processes to be accelerated on heterogeneous hardware (CPUs and GPUs). It provides a hierarchical architecture that supports distributed workflows across multiple nodes and devices. The framework separates the algorithm logic from the problem definition, allowing users to plug in various EAs and benchmark environments (such as physics engines like Brax) seamlessly.

Who it’s for

It is designed for researchers and developers working in evolutionary computation, neuroevolution, and reinforcement learning who need high-performance optimization tools that can scale across GPUs and distributed systems.

Highlights

  • Massive Algorithm Library: Access to over 140 evolutionary algorithms across the EvoX ecosystem, with 30+ built directly into the package.
  • Extensive Benchmarking: Includes over 240 benchmark problems and environments, ranging from numerical optimization to real-world engineering and RL tasks.
  • High Performance: Achieves over 100x speedups through GPU acceleration and distributed computing.
  • PyTorch Integration: Fully compatible with the PyTorch ecosystem, simplifying the development of neuroevolution models (e.g., evolving MLPs).
  • Visualization Tools: Includes built-in tools and a specialized .exv format for real-time data streaming and analysis of evolutionary processes.

Related

  • Project
    nnaisense/evotorch

    A scalable evolutionary computation library built on PyTorch that solves non-differentiable optimization problems using GPU acceleration and Ray distribution.

  • Project
    RobertTLange/evosax

    evosax is a JAX‑based library that implements 30+ Evolution Strategies (e.g., CMA‑ES, OpenAI‑ES, ARS, SimpleGA) with a vectorised `ask‑eval‑tell` API, enabling fast, hardware‑accelerated black‑box optimisation for neuro‑evolution, RL, and hyper‑parameter search.

  • Project
    pytorch/pytorch

    PyTorch is an open‑source Python library that offers GPU‑accelerated tensor operations and a dynamic, tape‑based autograd system for building and training deep neural networks. It integrates tightly with Python, supports extensions in C/C++ and other languages, and provides fast, low‑overhead execution on CPUs, NVIDIA GPUs, AMD ROCm, and Intel GPUs.

  • Project
    horovod/horovod

    A distributed deep learning training framework that makes it is easy to scale single-GPU training scripts to multiple GPUs and hosts across TensorFlow, PyTorch, Keras, and MXNet.

  • Project
    dynamiqs/dynamiqs

    A GPU-accelerated and differentiable Python library for high-performance simulation of quantum systems, designed for large-scale problems and gradient-based parameter estimation.