This work introduces CrystalGRPO, a CSP-aligned post-training framework that extends existing ODE-to-SDE policy constructions to the joint coordinate--lattice state 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.
Abstract
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.
Crystal structure prediction (CSP) is fundamental to computational materials discovery. Generative models including CDVAE, DiffCSP, FlowMM, and CrystalFlow learn stable-crystal distributions directly, but diffusion and flow-matching inference requires tens to thousands of sequential network evaluations per candidate. W...
Sourin Dey, Dipannoy Das Gupta, Wei Lai et al.· 1 citation
UniPath is proposed, a novel framework based on Learnable-Time Flow Matching that unifies crystal structure and energy prediction within a single probability path, and theoretically proves that under optimal velocity fields, these paths converge to a consistent trajectory.
Songyou Li, Mingze Li, Qianpu Liu et al.· Proceedings of the 32nd ACM...· 1 citation
Present generative CSP models function largely as implicit, softer-edged prototype libraries rather than genuinely de novo predictors, and enlarging this residual capacity, rather than aggregate match rate alone, is the central open problem.
Lai Wei, Rong-Zhi Dong, Ying Feng et al.· 1 citation
Crystal Structure Prediction (CSP), the task of determining stable atomic arrangements from chemical composition alone, remains a central challenge in computational materials science with direct implications for accelerating materials discovery. While recent diffusion-based generative models achieve impressive results...
Lu Yang, Tian-Tian Xu, Xiu-Feng Liu et al.· Proceedings of the 32nd ACM...· 0 citations
This work introduces ANCHOR, a GRPO composition policy trained with multi-objective rewards around a frozen CSP model, and continuous adaptive novelty (CAN), a graded novelty score against known structures and a growing discovery history.
Emma Lei Hovmand, J. Elsborg, Melih Kandemir et al.· arXiv.org· 0 citations
Protein optimization remains a longstanding goal in life sciences. Existing machine learning-assisted directed evolution (MLDE) methods primarily rely on sequence-only features, overlooking the critical spatial constraints and co-evolutionary interactions encoded in protein structures. However, directly integrating str...
Zi-Kun Nie, Su-Yuan Zhao, Yi-Zhen Luo et al.· 0 citations
Related blog posts
MIT News · Artificial Intelligence· news.mit.eduNov 17, 2026
By studying the computational foundations of language learning, Associate Professor Jacob Andreas builds intelligent systems that can communicate effectively with humans.
Microsoft Research Blog· microsoft.comNov 16, 2026
AI coding agents are increasingly help developers fix bugs and build features largely on their own, but every interaction with their models consumes tokens. In each session, an agent sends instructions, tool descriptions, and session history to its model. Most efforts to reduce this cost focus on individual tasks. Yet developers return to the same […] The post TokenEconomizer: Learning from Experience to Make AI Coding Agents More Efficient appeared first on Microsoft Research.
MIT News · Artificial Intelligence· news.mit.eduNov 13, 2026