Abstract:Many learning tasks, including learning potential energy surfaces from ab initio calculations, involve global spatial symmetries and permutational symmetry between atoms or general particles. Equivariant graph neural networks are a standard approach to such problems, with one of the most successful methods employing tensor products between various tensors that transform under the spatial group. However, as the number of different tensors and the complexity of relationships between them increase, maintaining parsimony and equivariance becomes increasingly challenging. In this paper, we propose using fusion diagrams, a technique widely employed in simulating SU($2$)-symmetric quantum many-body problems, to design new equivariant components for equivariant neural networks. This results in a diagrammatic approach to constructing novel neural network architectures. When applied to particles within a given local neighborhood, the resulting components, which we term "fusion blocks," serve as universal approximators of any continuous equivariant function defined in the neighborhood. We incorporate a fusion block into pre-existing equivariant architectures (Cormorant and MACE), leading to improved performance with fewer parameters on a range of challenging chemical problems. Furthermore, we apply group-equivariant neural networks to study non-adiabatic molecular dynamics of stilbene cis-trans isomerization. Our approach, which combines tensor networks with equivariant neural networks, suggests a potentially fruitful direction for designing more expressive equivariant neural networks.

Harmonizing SO(3)-Equivariance with Neural Expressiveness: a Hybrid Deep Learning Framework Oriented to the Prediction of Electronic Structure Hamiltonian

Going Beyond Multi-Task Dense Prediction with Synergy Embedding Models

DeepH-2: Enhancing deep-learning electronic structure via an equivariant local-coordinate transformer

General framework for E(3)-equivariant neural network representation of density functional theory Hamiltonian

Efficient and Equivariant Graph Networks for Predicting Quantum Hamiltonian

Infusing Self-Consistency into Density Functional Theory Hamiltonian Prediction via Deep Equilibrium Models

3D Equivariant Pose Regression via Direct Wigner-D Harmonics Prediction

A deep equivariant neural network approach for efficient hybrid density functional calculations

Unifying O(3) Equivariant Neural Networks Design with Tensor-Network Formalism

Transferable E(3) equivariant parameterization for Hamiltonian of molecules and solids

Learning Harmonic Molecular Representations on Riemannian Manifold

SE3Set: Harnessing equivariant hypergraph neural networks for molecular representation learning

SE(3) Equivariant Graph Neural Networks with Complete Local Frames

The prediction of the quality of results in Logic Synthesis using Transformer and Graph Neural Networks

Transferable E(3) Equivariant Parameterization for Hamiltonian of Molecules and Solids

Harmonic (Quantum) Neural Networks

Transformer neural networks and quantum simulators: a hybrid approach for simulating strongly correlated systems

Harmformer: Harmonic Networks Meet Transformers for Continuous Roto-Translation Equivariance

Symmetrical Graph Neural Network for Quantum Chemistry, with Dual R/K Space

Efficient hybrid density functional calculation by deep learning