SOAP and Muon optimizers outperform Adam in training machine learning interatomic potentials

Researchers systematically compared matrix-structured optimizers, including SOAP, Muon, and the hybrid SOAP-Muon, against the standard Adam optimizer for training NequIP and Allegro MLIP models. The study found that these new optimizers substantially outperform Adam in both convergence speed and final accuracy.

• SOAP and SOAP-Muon emerged as robust methods with consistent performance gains.
• Muon provided only partial gains relative to Adam.
• Improvements were particularly pronounced under conditions of partial force supervision.

The results indicate that the choice of optimizer is an overlooked yet impactful design axis for machine learning interatomic potentials.