mlsanityData to 14 Oct 2026

CrystalGRPO: Target-Aligned and Coverage-Preserving Reinforcement Learning for Flow-Based Crystal Structure Prediction

arXiv 2608.06582 ↗First posted 6 Aug 2026cs.LGcited by 1

On the map: Deep reinforcement learning methods → RL for energy, networks and operations

Abstract (arXiv)

Flow-based generative models can efficiently produce candidate structures for crystal structure prediction (CSP), but their pretrained objectives do not directly optimize downstream target recovery. Reinforcement-learning post-training offers a flexible solution, yet existing approaches rely primarily on energy rewards and coordinate-only stochastic policies. Predicted energy does not identify the reference polymorph, while reward-driven concentration can reduce the candidate coverage required for Top-N recovery. We introduce CrystalGRPO, a CSP-aligned post-training framework that extends existing ODE-to-SDE policy constructions to the joint coordinate--lattice state. CrystalGRPO combines MACE-predicted energy with a StructureMatcher-based recovery score and provides two operating modes: CrystalGRPO-Q, which prioritizes single-draw recovery, and CrystalGRPO-C, which combines full-trajectory reference regularization with a coverage-aware group advantage to preserve finite-budget target recovery. Across MP-20 and MPTS-52 with PXRDGen and OMatG backbones, both variants reduce one- and twenty-sample RMSE relative to coordinate-only reinforcement in all four backbone--dataset settings. CrystalGRPO-Q consistently improves Top-1, whereas CrystalGRPO-C achieves a higher Top-20 across all settings.

Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Placed in RL for energy, networks and operations, one of 15 directions in Deep reinforcement learning methods. Builds on 23 of its 33 references that are in the map. First posted on arXiv 6 Aug 2026.

Builds on

23 of 33 references.

Closest ancestors · 2

Same direction, newest first

  1. 2026Open Materials Generation with Inference-Time Reinforcement Learning31 Jan
  2. 2025Guiding Generative Models to Uncover Diverse and Novel Crystals via Reinforcement Learning10 Nov

Other references · 21

  1. 2017Proximal Policy Optimization Algorithmscited by 15,215
  2. 2024DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Modelscited by 9,379
  3. 2022Flow Matching for Generative Modelingcited by 5,702
  4. 2023Training Diffusion Models with Reinforcement Learningcited by 982
  5. 2025Does Reinforcement Learning Really Incentivize Reasoning Capacity in LLMs Beyond the Base Model?cited by 982
  6. 2025Flow-GRPO: Training Flow Matching Models via Online RLcited by 661
  7. 2021Crystal Diffusion Variational Autoencoder for Periodic Material Generationcited by 270
  8. 2023Crystal Structure Prediction by Joint Equivariant Diffusioncited by 148
  9. 2024Fine-Tuning of Continuous-Time Diffusion Models as Entropy-Regularized Controlcited by 143
  10. 2024FlowMM: Generating Materials with Riemannian Flow Matchingcited by 109
  11. 2024Fine-Tuned Language Models Generate Stable Inorganic Materials as Textcited by 93
  12. 2025Pass@K Policy Optimization: Solving Harder Reinforcement Learning Problemscited by 87
9 more
  1. 2024Space Group Constrained Crystal Generationcited by 73
  2. 2025SymmCD: Symmetry-Preserving Crystal Generation with Diffusion Modelscited by 51
  3. 2025Open Materials Generation with Stochastic Interpolantscited by 28
  4. 2025WyckoffDiff -- A Generative Diffusion Model for Crystal Symmetrycited by 26
  5. 2025A Periodic Bayesian Flow for Material Generationcited by 16
  6. 2025LLM Meets Diffusion: A Hybrid Framework for Crystal Material Generationcited by 13
  7. 2025All that structure matches does not glittercited by 12
  8. 2025Generative Models for Crystalline Materialscited by 4
  9. 2025AI-Assisted Rapid Crystal Structure Generation Towards a Target Local Environmentcited by 2

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 1.

1 paper.

  1. 2026Reinforcement Learning on the Discrete Composition Channel of a Crystal Generator: Validated Gains and Reward Hackingcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026Let CSP Be Your ANCHOR: Adaptive Crystal Search over Frozen Structure Priors
  2. 2026Riemannian Flow Models with Reinforcement Learning for Molecular Crystal Structure Prediction
  3. 2026CrystalReasoner: Reasoning and RL for Property-Conditioned Crystal Structure Generation
  4. 2026EP-Flow: Disordered Crystal Structure Prediction without Site-Level Annotations
  5. 2026uFlowCSP: Crystal Structure Prediction using Mean flow generative models
  6. 2026APO: Unsupervised Atomic Policy Optimization for 3D Structure Prediction of Atomic Systems
  7. 2026Deep Generative Crystal Structure Prediction: A Benchmark Study and a Controlled Test of Prototype Dependence
  8. 2026Generative structure search for efficient and diverse discovery of molecular and crystal structures