Speaker
Description
Allegro is a machine learning interatomic potential (MLIP) model designed to predict atomic properties in molecules using E(3) equivariant neural networks. When training this model, there tends to be a trade-off between accuracy and inference time. For this reason, we apply multi-objective hyperparameter optimization to both objectives. Additionally, we experiment with modified architectures by constructing variants of Allegro: one extended with additional classical layers and one incorporating quantum-classical hybrid layers based on quantum depth-infused (QDI) circuits. We evaluate all models on QM9, rMD17-aspirin, rMD17-benzene, and a self-generated DFT dataset of copper-lithium structures relevant to battery interfaces. Hyperparameters (6-9 per variant) are optimized with the surrogate-assisted multi-objective algorithm SAMO-COBRA, producing Pareto fronts that reveal accuracy-inference time trade-offs across all four datasets. As results, both variants surpass Allegro in force prediction accuracy across multiple datasets. The classical variant consistently improves over the baseline, while the quantum-classical hybrid variant achieves the best overall force prediction accuracy on the Cu-Li dataset, where it was fully optimized, outperforming the classical variant by approximately 13%. Notably, the hybrid variant also achieves competitive results on the remaining datasets despite using hyperparameters transferred from Cu-Li without dataset-specific optimization, suggesting that quantum-classical hybridization is a promising direction for enhancing MLIP architectures. We complement these results with ZX-calculus redundancy, Fisher-information trainability, and Fourier expressivity analyses of the quantum circuit. Published in Computational Materials Science 270, 114742 (2026).
| Topic | Quantum algorithms for machine learning applications |
|---|---|
| Additional Topic | Hybrid quantum-classical approaches for learning and optimization |