# Self-Scaled Broyden Family of Quasi-Newton Methods in JAX

arXiv: https://arxiv.org/abs/2603.10599 · HTML: https://mlsanity.com/paper/2603.10599/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Builds on 2 of its 7 references that are in the map.
- First posted on arXiv 11 Mar 2026.

## Builds on (most cited first, 2 of 2)

- [Optimizing the Optimizer for Physics-Informed Neural Networks and Kolmogorov-Arnold Networks](https://mlsanity.com/paper/2501.16371/) (2025)
- [Optimistix: modular optimisation in JAX and Equinox](https://mlsanity.com/paper/2402.09983/) (2024)

## Built on by (most cited first, 2 of 2)

- [Tackling Failure Modes of PINNs and PIKANs Using Conflict-Free Gradients](https://mlsanity.com/paper/2609.14841/) (2026)
- [Adjoint Method versus Physics-Informed Neural Networks in PDE-Constrained Inverse Problems](https://mlsanity.com/paper/2606.12337/) (2026)

## Abstract (arXiv)

We present a JAX implementation of the Self-Scaled Broyden family of quasi-Newton methods, fully compatible with JAX and building on the Optimistix~citeraderoptimistix2024 optimisation library. The implementation includes BFGS, DFP, Broyden and their Self-Scaled variants (SSBFGS, SSDFP, SSBroyden), as well as L-BFGS, together with a Zoom line search satisfying the strong Wolfe conditions. This is a short technical note, not a research paper, as it does not claim any novel contribution; its purpose is to document the implementation and ease the adoption of these optimisers within the JAX community. The code is available at https://github.com/IvanBioli/ssbroydenoptimistix and on PyPI (pip install ssbroyden-optimistix).