Written in parallel with the codebase. API reference, tutorials, and theory-to-code guides will ship here as the platform moves out of private beta.
The primary theoretical reference for the DAC platform. Covers the Markov geometry unifying diffusion maps and transformers, Cartan's structure equation, and the DMAP–NESS framework.
Read on arXiv →The FlashDiffusion source, including the sparse eigensolver, Nyström approximation, and B300-native kernel implementation. Issues and discussions open to beta collaborators.
View on GitHub →Submit your first run. Dataset formats, parameter reference, and how to read your results.
Full endpoint schema, field definitions, response format, and error codes for the FlashDiffusion API.
How Coifman–Lafon diffusion maps translate to the FlashDiffusion pipeline, step by step.
End-to-end walkthrough: submitting an MD trajectory, reading diffusion coordinates, interpreting the embedding.
Particle stack preparation, kernel parameter selection, and using diffusion coordinates for heterogeneity analysis.
Annotated source connecting equations in arXiv:2604.09560 to the FlashDiffusion implementation.
During the beta period the paper and source are the primary references. If you need something specific to run your data — reach out and we will help directly.