Loading
Life sciences · Preprint
arXiv · September 9, 2026
Posted before peer review. The findings may change or fail to hold.
uFlowCSP is a generative model for crystal structure prediction that achieves comparable or superior accuracy to existing methods (CrystalFlow, DiffCSP) while requiring substantially fewer network evaluations and wall-clock time. The method learns mean probability-flow velocity rather than instantaneous flow, enabling single-step inference matching or exceeding methods requiring hundreds to thousands of evaluations. This represents a significant computational efficiency gain, but the work is unpublished and requires peer review.
Algorithmic development and computational benchmarking. Computational materials science benchmark datasets (MP-20 from Materials Project, CSPBench); no human or biological subjects.. Intervention: uFlowCSP: MeanFlow-based crystal structure prediction model with chemistry- and symmetry-aware Transformer architecture, trained with coarse crystal-system token.. Compared with: CrystalFlow (flow-matching method at 100 and 2,000 evaluations) and DiffCSP (diffusion-based method at ~20,000 evaluations).
One-step uFlowCSP achieves 78.38% match rate on MP-20, matching CrystalFlow (78.34%) with 100x fewer evaluations and ~10x lower wall-clock time Five-step uFlowCSP reaches 83.64% on MP-20, exceeding CrystalFlow (78.34%) and DiffCSP (77.93%) with 20x fewer evaluations Under CSPBench energy-ranked top-five criterion, five-step uFlowCSP achieves 72% structure match, 72% space-group match, and 65% consensus match rates
Safety was not reported in the material analysed. Check the source before drawing any conclusion about harm.
The source did not state who this applies to in practice.
A novel computational method for crystal structure prediction showing improved efficiency and performance, but published on arXiv without peer review.
As stated by the source record.
Quoted from the source exactly as published.
Graded across the dimensions that decide whether you should act, each from what the source actually supports. There is no single score, and where a dimension was not assessed it says so.
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. We introduce uFlowCSP, a MeanFlow-based CSP model that learns the average, rather than instantaneous, probability-flow velocity. It generates a complete structure in one to five evaluations, delivering 5x-58x faster inference with equal or better performance. A chemistry- and symmetry-aware Transformer uses canonical atom ordering, global composition, and per-token chemistry embeddings. A coarse crystal-system token is used only during training; it provides additive gains, particularly improving space-group agreement despite being absent at inference, which remains formula-only. On MP-20 with 20 candidates per target, one step matches CrystalFlow (78.38% vs. 78.34%) with 100x fewer evaluations and about 10x lower wall-clock time. Five steps reach 83.64%, exceeding CrystalFlow (78.34% at 2,000 evaluations) and DiffCSP (77.93% at about 20,000), while using 20x fewer evaluations. uFlowCSP generates 10,000 structures in 0.39-1.31 minutes, versus 6.5 for CrystalFlow and 76.1 for DiffCSP. Under CSPBench's energy-ranked top-five structure-and-space-group criterion, five-step uFlowCSP reaches 72%/72%/65% structure, space-group, and consensus match rates. CrystalFlow reaches 78%/73%/68% at 100 steps but falls to 49%/32%/31% at five. Thus, uFlowCSP improves accuracy per network evaluation, not merely peak accuracy.
Taken from the source record, never inferred. Follow any of these and new work involving them reaches your briefing.