arXiv subjects

Chuin Wei Tan

Publications and source records attributed to Chuin Wei Tan.

8 recordsLinked to original sources

arXiv2026

Fast and Accurate Foundation Models for Equivariant Machine-Learned Interatomic Potentials

Machine-learned interatomic potentials (MLIPs) have emerged as a transformative tool for computational materials science and chemistry, with universal potentials trained on large and diverse datasets now routinely deployed as 'foundation models' for downstream fine-tuning in targeted chemical spaces. Many scientific applications of the resulting models, such as molecular dynamics (MD), require high inference and training speeds as well as accuracy. In this work, we examine the limits of equivariant MLIPs, which directly encode physical symmetries in model architectures, to achieve these competing targets -- particularly in the regime of extremely large datasets where data efficiency is less critical. We show how this trade-off can be addressed, and present a family of foundation potentials in the NequIP and Allegro equivariant MLIP architectures which achieve leading inference speeds and strong scalability as well as excellent accuracies across a range of community benchmarks -- spanning materials discovery, thermal conductivity prediction, and near-equilibrium mechanical and thermodynamic properties. Accelerations implemented within the NequIP infrastructure now permit training of high-accuracy foundation potentials on ultra-large datasets with dramatically reduced computational cost. Alongside, we show that efforts to improve model accuracy for materials discovery should focus on dataset diversity and improved, consistent descriptions of transition metal compound energy surfaces.

physics.comp-ph↗

arXiv2026

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.

cs.LG↗

arXiv2026

Mixture of Experts Framework in Machine Learning Interatomic Potentials for Atomistic Simulations

First-principles atomistic simulations are essential for understanding complex material phenomena but are fundamentally limited by their computational cost. While Machine Learning Interatomic Potentials (MLIPs) have drastically improved cost for a given accuracy, their inference cost remains a bottleneck for massive systems or long timescales. To address this, we introduce a multifidelity "Mixture-of-Experts" framework based on the E(3)-equivariant Allegro architecture. Our method spatially partitions the simulation domain into a chemically complex region (e.g., reactive interfaces) and a simple region (e.g., bulk lattice), assigning models of varying capacity to each. Among the challenges in such static domain decomposition, the mechanical mismatch between models at the interface is particularly critical, as it can generate artificial stress fields and instability. We address this challenge with a co-training strategy in which the loss function includes agreement constraints -- penalties on per-atom energy and force discrepancies between models evaluated on shared bulk environments -- forcing the independent models to learn a consistent physical description of the bulk material. We validate this approach on a realistic Pt+CO catalytic system, demonstrating that the co-trained models maintain exact energy conservation, align their bulk mechanical response (e.g., equation of state and bulk modulus), and achieve predictive accuracy comparable to a full high-fidelity simulation at more than twice the computational speed.

physics.comp-ph↗

arXiv2025

High-performance training and inference for deep equivariant interatomic potentials

Machine learning interatomic potentials, particularly those based on deep equivariant neural networks, have demonstrated state-of-the-art accuracy and computational efficiency in atomistic modeling tasks like molecular dynamics and high-throughput screening. The size of datasets and demands of downstream workflows are growing rapidly, making robust and scalable software essential. This work presents a major overhaul of the NequIP framework focusing on multi-node parallelism, computational performance, and extensibility. The redesigned framework supports distributed training on large datasets and removes barriers preventing full utilization of the PyTorch 2.0 compiler at train time. We demonstrate this acceleration in a case study by training Allegro models on the SPICE 2 dataset of organic molecular systems. For inference, we introduce the first end-to-end infrastructure that uses the PyTorch Ahead-of-Time Inductor compiler for machine learning interatomic potentials. Additionally, we implement a custom kernel for the Allegro model's most expensive operation, the tensor product. Together, these advancements speed up molecular dynamics calculations on system sizes of practical relevance by up to a factor of 18.

physics.comp-ph↗

arXiv2025

Exploring Charge Density Waves in two-dimensional NbSe2 with Machine Learning

Niobium diselenide (NbSe$_2$) has garnered significant attention due to the coexistence of superconductivity and charge density waves (CDWs) down to the monolayer limit. However, realistic modeling of CDWs-capturing effects such as layer number, twist angle, and strain-remains challenging due to the high computational cost of first-principles methods. Here, we develop a physically informed workflow for training machine-learning interatomic potentials (MLIPs) based on the E(3)-equivariant Allegro architecture, tailored to capture the subtle structural and dynamical signatures of CDWs in mono- and bilayer NbSe$_2$.We find that while CDW lattice distortions are relatively easy to learn, modeling vibrational properties remains more challenging. It requires targeted dataset design and careful hyperparameter tuning, pushing the boundaries and testing the extensibility of current MLIP frameworks. Our MLIPs enable reliable simulations of commensurate and incommensurate CDW phases, including their sensitivity to dimensionality and stacking, as well as CDW dynamics, phonons, and transition temperatures estimated via the stochastic self-consistent harmonic approximation. This work opens new possibilities for studying and tuning CDWs in NbSe$_2$ and other two-dimensional systems, with implications for electron-phonon coupling, superconductivity, and advanced materials design.

cond-mat.mtrl-sci↗

arXiv2024

Unified Differentiable Learning of Electric Response

Predicting response of materials to external stimuli is a primary objective of computational materials science. However, current methods are limited to small-scale simulations due to the unfavorable scaling of computational costs. Here, we implement an equivariant machine-learning framework where response properties stem from exact differential relationships between a generalized potential function and applied external fields. Focusing on responses to electric fields, the method predicts electric enthalpy, forces, polarization, Born charges, and polarizability within a unified model enforcing the full set of exact physical constraints, symmetries and conservation laws. Through application to $α$-SiO$_2$, we demonstrate that our approach can be used for predicting vibrational and dielectric properties of materials, and for conducting large-scale dynamics under arbitrary electric fields at unprecedented accuracy and scale. We apply our method to ferroelectric BaTiO$_3$ and capture the temperature-dependence and time evolution of hysteresis, revealing the underlying microscopic mechanisms of nucleation and growth that govern ferroelectric domain switching.

cond-mat.mtrl-sci↗

arXiv2023

Towards Sustainable Ultrawide Bandgap Van der Waals Materials: An ab initio Screening Effort

The sustainable development of next-generation device technology is paramount in the face of climate change and the looming energy crisis. Tremendous efforts have been made in the discovery and design of nanomaterials that achieve device-level sustainability, where high performance and low operational energy cost are prioritized. However, many of such materials are composed of elements that are under threat of depletion and pose elevated risks to the environment. The role of material-level sustainability in computational screening efforts remains an open question thus far. Here we develop a general van der Waals materials screening framework imbued with sustainability-motivated search criteria. Using ultrawide bandgap (UWBG) materials as a backdrop -- an emerging materials class with great prospects in dielectric, power electronics, and ultraviolet device applications, we demonstrate how this screening framework results in 25 sustainable UWBG layered materials comprising only of low-risks elements. Our findings constitute a critical first-step towards reinventing a more sustainable electronics landscape beyond silicon, with the framework established in this work serving as a harbinger of sustainable 2D materials discovery.

cond-mat.mtrl-sci↗

arXiv2023

Automatic Differentiation for Orbital-Free Density Functional Theory

Differentiable programming has facilitated numerous methodological advances in scientific computing. Physics engines supporting automatic differentiation have simpler code, accelerating the development process and reducing the maintenance burden. Furthermore, fully-differentiable simulation tools enable direct evaluation of challenging derivatives - including those directly related to properties measurable by experiment - that are conventionally computed with finite difference methods. Here, we investigate automatic differentiation in the context of orbital-free density functional theory (OFDFT) simulations of materials, introducing PROFESS-AD. Its automatic evaluation of properties derived from first derivatives, including functional potentials, forces, and stresses, facilitates the development and testing of new density functionals, while its direct evaluation of properties requiring higher-order derivatives, such as bulk moduli, elastic constants, and force constants, offers more concise implementations compared to conventional finite difference methods. For these reasons, PROFESS-AD serves as an excellent prototyping tool and provides new opportunities for OFDFT.

physics.comp-ph↗