Cormorant: Covariant Molecular Neural Networks

2019/06/06 by Brandon Anderson, Anderson, Brandon, Truong-Son Hy +3 · 26 citations
Biochemistry, Genetics and Molecular Biology · Computer Science · Materials Science · #Computational Drug Discovery Methods #Computational Physics (physics.comp-ph) #FOS: Computer and information sciences #FOS: Physical sciences #Machine Learning (cs.LG) #Machine Learning (stat.ML) #Machine Learning in Materials Science #Protein Structure and Dynamics

paper · pdf · doi:10.48550/arxiv.1906.04015

openalex publication_date 2019/06/06 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

We propose Cormorant, a rotationally covariant neural network architecture for learning the behavior and properties of complex many-body physical systems. We apply these networks to molecular systems with two goals: learning atomic potential energy surfaces for use in Molecular Dynamics simulations, and learning ground state properties of molecules calculated by Density Functional Theory. Some of the key features of our network are that (a) each neuron explicitly corresponds to a subset of atoms; (b) the activation of each neuron is covariant to rotations, ensuring that overall the network is fully rotationally invariant. Furthermore, the non-linearity in our network is based upon tensor products and the Clebsch-Gordan decomposition, allowing the network to operate entirely in Fourier space. Cormorant significantly outperforms competing algorithms in learning molecular Potential Energy Surfaces from conformational geometries in the MD-17 dataset, and is competitive with other methods at learning geometric, energetic, electronic, and thermodynamic properties of molecules on the GDB-9 dataset.

Citations

• Geometric Deep Learning: Going beyond Euclidean data
• Tensor field networks: Rotation- and translation-equivariant neural networks for 3D point clouds
• Classical Electrodynamics

Cited by

• Hierarchical geometric deep learning enables scalable analysis of molecular dynamics
• CORGI: GNNs with Convolutional Residual Global Interactions for Lagrangian Simulation
• Categorical Equivariant Deep Learning: Category-Equivariant Neural Networks and Universal Approximation Theorems
• Direct Molecular Polarizability Prediction with SO(3) Equivariant Local Frame GNNs
• Learning with Category-Equivariant Architectures for Human Activity Recognition
• Cross Learning between Electronic Structure Theories for Unifying Molecular, Surface, and Inorganic Crystal Foundation Force Fields
• Knowledge Distillation of a Protein Language Model Yields a Foundational Implicit Solvent Model
• Logical GANs: Adversarial Learning through Ehrenfeucht Fraisse Games
• DeepChem Equivariant: SE(3)-Equivariant Support in an Open-Source Molecular Machine Learning Library
• Scaling Laws and Symmetry, Evidence from Neural Force Fields
• Learning from the electronic structure of molecules across the periodic table
• Adaptive Canonicalization with Application to Invariant Anisotropic Geometric Networks
• Learning Inter-Atomic Potentials without Explicit Equivariance
• Shoot from the HIP: Hessian Interatomic Potentials without derivatives
• MolMark: Safeguarding Molecular Structures through Learnable Atom-Level Watermarking
• Learning from Protein Structure with Geometric Vector Perceptrons
• E(n) Equivariant Graph Neural Networks
• DEQuify your force field: More efficient simulations using deep equilibrium models
• Bayesian Prior Construction for Uncertainty Quantification in First-Principles Statistical Mechanics
• Using pretrained graph neural networks with token mixers as geometric featurizers for conformational dynamics
• Hierarchical, rotation‐equivariant neural networks to select structural models of protein complexes
• Guiding Diffusion Models with Reinforcement Learning for Stable Molecule Generation
• Frame Averaging for Invariant and Equivariant Network Design
• Node-Equivariant Message Passing for Efficient and Accurate Machine Learning Interatomic Potentials
• Equi-mRNA: Protein Translation Equivariant Encoding for mRNA Language Models
• Chem3DLLM: 3D Multimodal Large Language Models for Chemistry

Related

• N-body Networks: a Covariant Hierarchical Neural Network Architecture for Learning Atomic Potentials
• Egret-1: Pretrained Neural Network Potentials for Efficient and Accurate Bioorganic Simulation
• Generalizing Neural Wave Functions
• Accurate Many-Body Repulsive Potentials for Density-Functional Tight-Binding from Deep Tensor Neural Networks
• Multi-task learning for electronic structure to predict and explore molecular potential energy surfaces