github.com/owlbarn/owl ↗

About owlbarn/owl

Owl is a comprehensive scientific computing library written in OCaml that brings numerical and mathematical computing capabilities to the language. The system provides a broad set of tools for researchers and engineers, including n-dimensional arrays (tensors), linear algebra, statistics, optimization, automatic differentiation, neural networks, and signal processing via discrete Fourier transforms. It also supports data frame processing and visualization through integration with external packages.

The library is structured around multiple implementations of n-dimensional arrays to balance different needs: a high-performance version built on C functions for serious computational work, a pure OCaml base version for broader compatibility and safety, and a computation-graph version that enables symbolic computing similar to TensorFlow v1. All three array types can be used with advanced modules like automatic differentiation and neural networks. The project is built on solid mathematical foundations, offering everything from basic mathematical functions to specialized operations like Beta and Gamma functions, ordinary differential equations, interpolation, and various probability distributions.

Owl is maintained by a dedicated team as the de-facto scientific computing framework in OCaml, with a focus on stability and practical usability. The project encourages the community to build specialized tools on top of Owl within the Owlbarn GitHub organization, and it operates collaboratively through pull requests and community discussion on OCaml Discourse and a dedicated Slack channel.