ATLAS: A Foundation Neural Sampler for Amorphous Materials

2026-07-21

AI Summary

This work addresses the challenge of efficiently sampling the complex energy landscape of amorphous materials below the glass transition temperature, where conventional methods struggle and data-driven models are hindered by scarce, biased reference ensembles. The authors propose ATLAS, a diffusion-based neural sampler that leverages equivariant graph neural networks to directly generate amorphous structures conforming to the Boltzmann distribution from a target energy function, enabling generalization across scales, temperatures, and compositions. Key innovations include efficient structure generation and manipulation without traditional simulations, composition-amortized pretraining to drastically reduce inverse design costs, and an active search strategy combining observable-guided sampling with large language model collaboration. Experiments demonstrate that ATLAS achieves sub-0.2% free energy error in the 2D Kob–Andersen system using fewer than 1/500th the energy evaluations, reproduces experimental short-range order and optimizes bulk modulus in Cu–Zr and Cr–Co–Ni metallic glasses, and discovers the Pareto front of an eight-component high-entropy metallic glass within 480 evaluations.

Abstract

Amorphous materials exhibit exceptional mechanical and functional properties, yet their rugged energy landscapes are notoriously difficult to sample. Below the glass-transition temperature, conventional molecular dynamics and Monte Carlo become inefficient because equilibration relies on rare barrier-crossing events, while data-driven generative models are constrained by scarce and biased reference ensembles. Here, we introduce ATLAS, an efficient sampler that learns a diffusion process to generate Boltzmann-distributed amorphous structures directly from a target energy function. Parameterized by an equivariant graph neural network, ATLAS generalizes across system size, temperature, and composition. By exploiting the time reversal of the diffusion process, it enables efficient estimation of thermodynamic quantities and steering toward target observables. In two-dimensional Kob-Andersen systems, ATLAS reproduces parallel tempering Markov chain Monte Carlo structural distributions, free energies and entropies, achieving below 0.2% free energy error in the low-temperature glass regime with over 500-fold fewer energy evaluations. In Cu-Zr and Cr-Co-Ni metallic glasses, ATLAS recovers experimentally observed short-range-order trends and steers structures toward prescribed order parameters and optimized bulk moduli. Moreover, composition-amortized pretraining outperforms composition-specific training from scratch, reduces inverse-design costs by several hundred-fold, and enables sampling with expensive universal machine learning interatomic potentials. Coupled to a large language model agent, ATLAS searches an eight-element space for high-entropy metallic glasses balancing stiffness and ductility, identifying a converged Pareto frontier within 480 oracle evaluations. Together, these results establish ATLAS as a foundation model for sampling, steering and designing amorphous materials.

Problem

• amorphous materials
• energy landscape sampling
• glass transition
• Boltzmann distribution
• rare barrier-crossing events

Innovation

• diffusion generative model
• equivariant graph neural network
• amorphous materials sampling
• composition-amortized pretraining
• inverse design