Physical regularized Hierarchical Generative Model for Metallic Glass Structural Generation and Energy Prediction
Qiyuan Chen, Ajay Annamareddy, Ying-Fei Li, Dane Morgan, Bu Wang
TL;DR
GlassVAE introduces a physically regularized, hierarchical graph variational autoencoder tailored for disordered metallic glasses, unifying structure generation and energy prediction in a single latent space. By enforcing invariance to rotations, translations, and permutations and incorporating RDF and energy-regression regularizers, the model learns latent representations that respect both local geometry and global energetics. Empirical results on Cu$_{50}$Zr$_{50}$ show near-perfect energy prediction ($R^2>0.99$) and high-fidelity graph reconstruction (node-type accuracy >95%, distance $R^2\approx0.99$), while enabling realistic generation and energy-conditioned sampling that align with the underlying energy landscape. Theoretical backing links the physics-informed terms to improved latent-space generalization, and the framework demonstrates scalability to multiple trajectories, suggesting practical potential for accelerated discovery and design of disordered materials.
Abstract
Disordered materials such as glasses, unlike crystals, lack long range atomic order and have no periodic unit cells, yielding a high dimensional configuration space with widely varying properties. The complexity not only increases computational costs for atomistic simulations but also makes it difficult for generative AI models to deliver accurate property predictions and realistic structure generation. In this work, we introduce GlassVAE, a hierarchical graph variational autoencoder that uses graph representations to learn compact, rotation, translation, and permutation invariant embeddings of atomic configurations. The resulting structured latent space not only enables efficient generation of novel, physically plausible structures but also supports exploration of the glass energy landscape. To enforce structural realism and physical fidelity, we augment GlassVAE with two physics informed regularizers, a radial distribution function (RDF) loss that captures characteristic short and medium range ordering and an energy regression loss that reflects the broad configurational energetics. Both theoretical analysis and experimental results highlight the critical impact of these regularizers. By encoding high dimensional atomistic data into a compact latent vector and decoding it into structures with accurate energy predictions, GlassVAE provides a fast, physics aware path for modeling and designing disordered materials.