Demo repository for the Broadcast Product.
Yusuke Matsui and Tatsuya Yokota "Broadcast Product: Redefining Shape-aligned Element-wise Multiplication and Beyond", Transactions on Machine Learning Research (TMLR), 2026. https://openreview.net/forum?id=zv0OtOPpPO
@article{matsui2026broadcast,
title={Broadcast Product: Redefining Shape-aligned Element-wise Multiplication and Beyond},
author={Matsui, Yusuke and Yokota, Tatsuya},
journal={Transactions on Machine Learning Research},
year={2026},
url={https://openreview.net/forum?id=zv0OtOPpPO}
}demo_python.py and demo_julia.jl contain working code examples for Eq. 4, 5, and 10 from the paper.
broadcast_decomposition/ contains implementations of Broadcast Decomposition in matlab, python, and julia. matlab is the master implementation; python and julia are automatically translated versions. If you find any issues, please open a PR.