Molecular Modeling

This section focuses on 3D molecular modeling: representing and reasoning about molecules in three-dimensional space rather than as flat strings or 2D graphs. Notes here cover conformer generation, neural network interatomic potentials, and architectures that model the continuous volumetric space around a molecule. A recurring theme is the tension between atom-centric representations and approaches that capture broader spatial context, including how to learn smooth and physically meaningful potential energy surfaces from data.

Computational Chemistry

MOFFlow assembles metal nodes and organic linkers into Metal-Organic Framework structures

MOFFlow: Flow Matching for MOF Structure Prediction

MOFFlow is the first deep generative model tailored for Metal-Organic Framework (MOF) structure prediction. It utilizes Riemannian flow matching on SE(3) to assemble rigid building blocks (metal nodes and organic linkers), achieving higher accuracy and scalability than atom-based methods on large systems.

Computational Chemistry

Potential energy surface showing molecular conformation space with equilibrium and low energy conformations

DenoiseVAE: Adaptive Noise for Molecular Pre-training

ICLR 2025 paper introducing DenoiseVAE, which learns adaptive, atom-specific noise distributions through a VAE framework to improve denoising-based pre-training for molecular force field prediction, outperforming fixed Gaussian noise approaches on quantum chemistry benchmarks.

Computational Chemistry

Adaptive grid merging visualization for benzene molecule showing multi-resolution spatial discretization

Beyond Atoms: 3D Space Modeling for Molecular Pretraining

ICML 2025 paper introducing SpaceFormer, a Transformer architecture that challenges the atom-centric paradigm by modeling the continuous 3D space surrounding molecules using adaptive multi-resolution grids, achieving state-of-the-art performance on quantum property prediction benchmarks.

Computational Chemistry

A mathematical representation of a potential energy surface (PES)

Dark Side of Forces: Non-Conservative ML Force Models

ICML 2025 analysis rigorously quantifying when non-conservative force models (which predict forces directly) fail in molecular dynamics, demonstrating simulation instabilities and proposing hybrid architectures that capture speed benefits without sacrificing physical correctness.

Computational Chemistry

Efficient DFT Hamiltonian Prediction via Adaptive Sparsity

ICML 2025 methodological paper introducing SPHNet, which uses adaptive network sparsification to overcome the computational bottleneck of tensor products in SE(3)-equivariant networks, achieving up to 7x speedup and 75% memory reduction on DFT Hamiltonian prediction tasks.

Computational Chemistry

Atomic structure of a spherical fullerene

Learning Smooth Interatomic Potentials with eSEN

ICML 2025 paper proposing energy conservation metrics as critical diagnostics for machine learning interatomic potentials and introducing eSEN, a novel architecture designed to bridge the gap between test-set accuracy and real simulation performance on materials property prediction.