Team

The library was developed by:

Why we built this library

High-dimensional tensors and multivariate functions are often difficult to analyze directly because interaction structure grows rapidly with dimension. HDMRLib was created to support these settings with tools that make lower-order representations easier to compute, inspect, and compare across different computational backends.

The project is intended to support users who need:

  • interpretable lower-order decompositions

  • a consistent workflow for HDMR and EMPR

  • backend flexibility across NumPy, PyTorch, and TensorFlow

  • reusable software for research, experimentation, and teaching

Open-source development

HDMRLib is maintained as an open-source software project. The codebase, issue tracking, and development history are available through the public repository.

We aim to make the library useful both as a research tool and as a software foundation that can be extended in future work.

Contact and repository