Chemistry Research Notes

Research notes on computational chemistry, organized by problem area.

Molecular Design – De novo molecule generation, property prediction, and reaction prediction
Molecular Representations – String notations, learned encoders, cross-modal models, and name translation
Optical Structure Recognition – Extracting molecular structures from images
LLM Applications – General LLMs and VLMs adapted for chemical tasks
Molecular Simulation – Molecular dynamics, force fields, 3D modeling, and neural network potentials
Datasets – Dataset cards for chemistry data

Molecular Simulation

Notes on molecular simulation: molecular dynamics, force fields, 3D conformer generation, and neural network potentials.

Molecular Representations

Notes on molecular representations: string notations (SMILES, InChI, SELFIES), learned encoders, cross-modal models, and name translation.

Optical Chemical Structure Recognition

Notes on recognizing molecular structures from images, covering 35 years of methods: from rule-based vectorization to vision-language models.

LLMs for Chemistry

An archive of notes covering large language models and multimodal models adapted for chemistry tasks. Unlike chemical language models that learn directly from molecular string representations (SMILES, SELFIES), these models build on general-purpose LLM or VLM backbones and are fine-tuned for chemical reasoning, multimodal molecular understanding, or scientific document processing.

Molecular Design

Notes on computational molecular design: de novo generation, benchmarks and evaluation, property prediction, and reaction prediction.

Molecular Databases & Datasets

Dataset cards for molecular enumeration databases (GDB series, ZINC-22) and conformer datasets (GEOM, MARCEL) used in virtual screening and molecular modeling.