cond-mat.mtrl-sciNov 30, 2025

Worse Forces Can Lead to Better Structures: Trajectory-Level Fine-Tuning of Machine Learning Interatomic Potentials

Authors: Evan Dramko, Yizhi Zhu, Aleksandar Krivokapic, Geoffroy Hautier, Thomas Reps, Christopher Jermaine, Anastasios Kyrillidis

Organizations: Department of Computer Science, Rice University, Houston, USA · Department of Materials Science and Nanoengineering, Rice University, Houston, USA · Rice Advanced Materials Institute, Rice University, Houston, USA · Faculty of Technical Sciences, University of Novi Sad, Novi Sad, Serbia · 

Abstract

Machine learning interatomic potentials (MLIPs) are commonly trained and evaluated by their ability to reproduce first-principles energies and forces. Practical value, however, is often determined by downstream processes such as structural relaxation. Much of MLIP literature is motivated by the expectation that improving force accuracy will produce correspondingly better downstream behavior. Although this expectation is natural, it may not hold in practice due to model inaccuracies. Here, we demonstrate that force accuracy and relaxation accuracy can be driven in opposite directions. Starting from pretrained MLIPs, we differentiate through relaxation trajectories and optimize the the final structures. This intervention produces increases in force-prediction errors while substantially reducing structural errors. Importantly, this represents a trade-off between two aspects of physical fidelity: the learned forces become less faithful to first-principles forces, while the resulting relaxed structures become more faithful to first-principles predictions. Rather than observing a discrepancy between local quality and downstream performance, we demonstrate that this discrepancy can be induced through optimization. Our findings suggest that MLIPs need not be optimized as approximations of first-principles forces, but can instead learn task-specific vector fields that guide relaxation trajectories toward more accurate final structures. By propagating supervision through complete relaxation trajectories, our framework captures dependencies between successive predictions that are absent when structure--force pairs are treated independently. More broadly, these results demonstrate that conventional force metrics can provide an incomplete assessment of downstream performance and highlights the importance of evaluating scientific machine-learning models within computational workflows.

Figures & tables

Appendix figures & tables6 assets

Supplementary material from the paper’s appendix.

Appendix

Explore similar work

Jul 2, 2026

Beyond Adam: SOAP and Muon for Faster, Label-Efficient Training of Machine Learning Interatomic Potentials

Machine learning interatomic potentials (MLIPs) have become a hallmark of AI for scientific simulation. While efforts on new architectures and datasets have led to increasingly accurate and general models, the choice of optimizer for training has largely remained unexplored, defaulting to Adam and its variants in the community. Here, we implement and systematically compare a class of recently proposed matrix-structured optimizers, including Muon, SOAP, and the hybrid SOAP-Muon, for training NequIP and Allegro MLIP models. We find that these optimizers can substantially outperform Adam in both convergence speed and final accuracy. SOAP and SOAP-Muon emerge as robust and consistently strong methods, while Muon only provides partial gains relative to Adam. The improvements are particularly pronounced under partial force supervision. Our results indicate that optimizer choice is an overlooked yet impactful design axis for MLIPs.
Oct 10, 2026

On the Theoretical Limits of Learning Unconstrained Interatomic Potentials

Machine-learned interatomic potentials (MLIPs) accelerate molecular dynamics (MD) simulations by learning the energies and forces that govern atomic motion. Recent work has shown that relaxing physical inductive biases can improve computational efficiency and predictive accuracy, raising the question of whether energy-and-force matching alone is sufficient to ensure accurate sampling. We show that, for unconstrained hypothesis classes, the relationship between prediction error and sampling accuracy is exponentially ill-conditioned in dimension. In particular, prediction errors that decay exponentially with dimension can coexist with nearly maximal total variation distance between the learned and target distributions. We identify two failure mechanisms. The first is kinetic trapping, where relaxation to an accurate learned equilibrium requires exponentially long times, even when simulations are initialized from training configurations at which energies and forces are matched exactly. The second is equilibrium mismatch, where the learned stationary distribution itself remains far from the target. Under explicit assumptions, we show that equilibrium mismatch can persist even in local, symmetry-preserving models, where small per-atom errors coexist with a distributional discrepancy that approaches its maximum as system size increases. Finally, we establish global energy-control conditions that guarantee equilibrium accuracy and identify additional conditions required for rapid relaxation.
May 22, 2026

Multitask learning with semiempirical orbital charges enables sample-efficient MLIPs

Machine learning interatomic potentials (MLIPs) require generating computationally expensive, large-scale training datasets to accurately simulate materials and molecules. Incorporating electronic structure information using multitask learning improves sample efficiency, however, training on full Hamiltonian matrices, which scale quadratically with the number of atoms, is intractable for large datasets. In this work, we show that multitask learning utilizing orbitally resolved semiempirical charges significantly improves sample efficiency and accuracy in MLIPs. To efficiently predict orbital charges, we implement a specialized equivariant model, reducing charge prediction error compared to an invariant baseline. By augmenting training with computationally inexpensive GFN1-xTB orbital charges, which scale linearly with the number of atoms, our model achieves a 58% reduction in energy mean absolute error relative to energy-only baselines, and requires five times less data to match their performance. Against the stronger baseline of joint energy-and-force training, orbital charges still reduce the energy error by 12%. Furthermore, our approach matches or outperforms models trained on expensive density functional theory (DFT) atomic charges, capturing orbitally resolved electronic complexity and forcing the network to learn a physically accurate latent space that spontaneously clusters metals by shared chemical properties. Because orbital charges are only required during training, this approach preserves inference efficiency, providing a scalable recipe for developing accurate, data-efficient foundation models for complex chemical systems.