mlsanityData to 7 Oct 2026

OMatG-flash: An All-Atom Flow Map with Reinforce Adjoint Matching for Scalable Materials Discovery

arXiv 2609.26402 ↗First posted 22 Sep 2026cs.LGcited by 1

On the map: Diffusion models → Flow matching, time series and science

Abstract (arXiv)

The discovery of novel inorganic materials drives technological breakthroughs in critical fields such as computing and energy storage. Generative AI has promised to accelerate the materials discovery pipeline, but state-of-the-art flow and diffusion models remain bottlenecked by the cost of proposing candidate materials. To address this, we introduce OMatG-flash, an all-atom flow map for inorganic crystal structure prediction (CSP) and de novo generation (DNG). OMatG-flash is a Pareto-optimal inference engine for materials, sampling candidate materials with an order of magnitude fewer inference steps and less wall-clock time than existing flow and diffusion models while demonstrating benchmark performance on par with the state-of-the-art. To enable post-training fine-tuning we apply Reinforce Adjoint Matching to flow maps, further improving match rates and RMSE on the unconditional CSP task. OMatG-flash showcases the potential of flow maps to accelerate generation of high-quality candidate inorganic materials and demonstrates a step forward in sample throughput necessary for data-hungry materials discovery workflows.

Key facts. Cited by 1 paper in the map (data through 7 Oct 2026). Placed in Flow matching, time series and science, one of 14 directions in Diffusion models. Builds on 56 of its 85 references that are in the map. First posted on arXiv 22 Sep 2026.

Builds on

56 of 85 references.

Closest ancestors · 10

Same direction, newest first

  1. 2025Generalised Flow Maps for Few-Step Generative Modelling on Riemannian Manifolds24 Oct
  2. 2025Space Group Conditional Flow Matching28 Sep
  3. 2025Kinetic Langevin Diffusion for Crystalline Materials Generation4 Jul
  4. 2025How to build a consistency model: Learning flow maps via self-distillation24 May
  5. 2025All-atom Diffusion Transformers: Unified generative modelling of molecules and materials5 Mar
  6. 2025Open Materials Generation with Stochastic Interpolants4 Feb
  7. 2023Crystal Structure Prediction by Joint Equivariant Diffusion30 Jul
  8. 2023Practical and Asymptotically Exact Conditional Sampling in Diffusion Models30 Jun
  9. 2022Flow Matching for Generative Modeling6 Oct
  10. 2022Flow Straight and Fast: Learning to Generate and Transfer Data with Rectified Flow7 Sep

Other references · 46

  1. 2017Proximal Policy Optimization Algorithmscited by 14,441
  2. 2020Score-Based Generative Modeling through Stochastic Differential Equationscited by 8,964
  3. 2024DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Modelscited by 8,851
  4. 2015Deep Unsupervised Learning using Nonequilibrium Thermodynamicscited by 7,005
  5. 2022Scalable Diffusion Models with Transformerscited by 5,931
  6. 2023Consistency Modelscited by 1,778
  7. 2023Training Diffusion Models with Reinforcement Learningcited by 939
  8. 2023Stochastic Interpolants: A Unifying Framework for Flows and Diffusionscited by 897
  9. 2021E(3)-Equivariant Graph Neural Networks for Data-Efficient and Accurate Interatomic Potentialscited by 701
  10. 2025Flow-GRPO: Training Flow Matching Models via Online RLcited by 605
  11. 2018Tune: A Research Platform for Distributed Model Selection and Trainingcited by 395
  12. 2021Crystal Diffusion Variational Autoencoder for Periodic Material Generationcited by 268
34 more
  1. 2024Adjoint Matching: Fine-tuning Flow and Diffusion Generative Models with Memoryless Stochastic Optimal Controlcited by 222
  2. 2024FlowMM: Generating Materials with Riemannian Flow Matchingcited by 105
  3. 2025UMA: A Family of Universal Models for Atomscited by 99
  4. 2023MatterGen: a generative model for inorganic materials designcited by 98
  5. 2025Feynman-Kac Correctors in Diffusion: Annealing, Guidance, and Product of Expertscited by 85
  6. 2024Orb: A Fast, Scalable Neural Network Potentialcited by 63
  7. 2026Flow Map Language Models: One-step Language Modeling via Continuous Denoisingcited by 49
  8. 2025SymmCD: Symmetry-Preserving Crystal Generation with Diffusion Modelscited by 48
  9. 2026Meta Flow Maps enable scalable reward alignmentcited by 32
  10. 2026Diamond Maps: Efficient Reward Alignment via Stochastic Flow Mapscited by 26
  11. 2025Wyckoff Transformer: Generation of Symmetric Crystalscited by 24
  12. 2025LeMat-GenBench: A Unified Evaluation Framework for Crystal Generative Modelscited by 20
  13. 2025Guiding Generative Models to Uncover Diverse and Novel Crystals via Reinforcement Learningcited by 19
  14. 2026Reinforce Adjoint Matching: Scaling RL Post-Training of Diffusion and Flow-Matching Modelscited by 17
  15. 2025Space Group Equivariant Crystal Diffusioncited by 15
  16. 2025FALCON: Few-step Accurate Likelihoods for Continuous Flowscited by 12
  17. 2025PLaID++: A Preference Aligned Language Model for Targeted Inorganic Materials Designcited by 11
  18. 2025All that structure matches does not glittercited by 10
  19. 2026Crystalite: A Lightweight Transformer for Efficient Crystal Modelingcited by 9
  20. 2026Self-conditioned Flow Map Language Models via Fixed-point Flowscited by 9
  21. 2026MACE-POLAR-1: A Polarisable Electrostatic Foundation Model for Molecular Chemistrycited by 7
  22. 2026Open Materials Generation with Inference-Time Reinforcement Learningcited by 6
  23. 2025Boltzmann generators for amorphous particle systemscited by 5
  24. 2026Learning Hamiltonian Flow Maps: Mean Flow Consistency for Large-Timestep Molecular Dynamicscited by 5
  25. 2026Riemannian MeanFlowcited by 5
  26. 2026MolCrystalFlow: Molecular Crystal Structure Prediction via Flow Matchingcited by 5
  27. 2025Guided Diffusion for the Discovery of New Superconductorscited by 4
  28. 2026Strong Stochastic Flow Mapscited by 4
  29. 2025AI-Driven Expansion and Application of the Alexandria Databasecited by 3
  30. 2026Fast Organic Crystal Structure Prediction with Unit Cell Flow Matchingcited by 3
  31. 2026SLayerGen: a Crystal Generative Model for all Space and Layer Groupscited by 2
  32. 2026Few-Step Boltzmann Generators via Scalable Likelihood Flow Mapscited by 2
  33. 2026Few-step Cofolding with All-Atom Flow Mapscited by 1
  34. 2026Discovering Crystal Structure Prediction Algorithms with an AI Co-Scientistcited by 1

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 1.

1 paper.

  1. 2026Riemannian Flow Models with Reinforcement Learning for Molecular Crystal Structure Predictioncited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026EP-Flow: Disordered Crystal Structure Prediction without Site-Level Annotations
  2. 2026GLASS: Global Latent Aggregation with Slot-based Set Decoding for Scalable All-Atom Crystal Generation
  3. 2026uFlowCSP: Crystal Structure Prediction using Mean flow generative models
  4. 2026OxiGen: Oxidation-State-Aware Crystal Generation
  5. 2026Finetuning-Free Diffusion Model with Adaptive Constraint Guidance for Inorganic Crystal Structure Generation
  6. 2025Establishing baselines for generative discovery of inorganic crystals
  7. 2026Chemical filters for ultra-high-throughput materials screening and generation
  8. 2024FlowLLM: Flow Matching for Material Generation with Large Language Models as Base Distributions