CS Paper Radar

Symmetry-Breaking De Novo Crystal Generation via Markovian Jump Diffusion

Van Khoa Nguyen, Alexandros Kalousis

cs.LG ·submitted 2026-08-13 16:41 UTC · v1

Abstract

Generating crystals has recently attracted significant interest due to their broad applications in materials science. However, existing generative models struggle to produce complete crystallographic specifications, limiting their ability to capture global symmetry and structural dependencies. In particular, current state-of-the-art approaches generate crystals only up to site symmetries and rely on sampling space groups from empirical distributions during generation. Inspired by \emph{spontaneous symmetry breaking} in physics, where crystals break symmetries under external conditions, we propose a novel diffusion-based framework that generates full structure specifications by reversing from the lowest-symmetry priors. Our method leverages a Markovian jump-diffusion process to model these symmetry-breaking dynamics, enabling it to traverse different space groups in a physically motivated manner. Our model, dubbed \emph{Symmetry-breaking Crystal Diffusion} (SbCD), introduces a principled approach to explicitly incorporate inter-space-group transitions into the generative process. In de novo generation experiments on MP20 and MPTS-52, SbCD outperforms its symmetry-preserving counterpart by a substantial margin, offering a promising perspective for generative modeling of crystalline materials.

Comment
Under review

Closest papers

Nearest neighbours by meaning across the whole index.

CrystalAR: Autoregressive Modeling for Symmetric Crystal Generation

cs.AI cond-mat.mtrl-sci ·2026-10-20 ·Ziqiao Meng, Rongsheng Chen, Zhi Zheng, Nikita Kazeev, Zhenkun Wang +2 more
similarity 0.70
Abstract

Deep generative modeling has achieved remarkable progress in crystal generation. Given the rich symmetries inherent in crystalline structures, recent work has sought to explicitly incorporate space-group symmetry into generative models, enabling more effective exploration of novel symmetry templates. Yet introducing symmetry templates increases the combinatorial complexity of stoichiometric relationships between atomic species and crystallographic sites, while creating a fundamental tension between flexible exploration of discrete crystallographic templates and faithful realization of these templates as continuous three-dimensional structures. To address these challenges, we propose CrystalAR, a unified autoregressive framework that models Wyckoff symmetry templates and complete crystal structures within a single generative sequence. Specifically, CrystalAR reformulates continuous coordinate generation as a symmetry-conditioned hierarchy of discrete spatial localization decisions, autoregressively refining the positions of all symmetry-inequivalent sites across resolution levels while jointly modeling their spatial configuration within each level. This discrete hierarchical formulation further enables constraint-aware search over both crystallographic templates and geometric realizations. As a result, CrystalAR retains strong symmetry-template generation capability while achieving competitive de novo generation performance and template-free crystal structure prediction (CSP) accuracy. These results suggest that symmetry-template generation can be retained alongside broadly competitive de novo generation performance and competitive CSP without reliance on symmetry templates retrieved from prior data, providing a promising alternative paradigm for symmetry-preserving crystal generation.

0.26

Orbitol: Crystal Generation on Wyckoff Orbits

cs.LG ·2026-10-21 ·Ivan Žugec, Joshua Rosenthal, Ivor Lončarić, Jan-Willem van de Meent, Tin Hadži Veljković
similarity 0.68
Abstract · 39 pages, 9 figures, 13 tables

Generative models propose crystalline materials by learning composition and geometry, but unconditional sampling gives no direct control over the space group, which constrains whether responses such as piezoelectricity are symmetry-allowed. We present Orbitol, a crystal generator built on the Crystalite diffusion Transformer that replaces atom tokens with Wyckoff-orbit tokens. Given a space group and a collection of occupied Wyckoff positions, Orbitol generates the atom type and free coordinates of each orbit together with a compatible lattice, so every valid structure is invariant under the operations of the requested group. It generates structures two to three orders of magnitude faster than the symmetry-aware baselines at the evaluated settings. On MP-20, 13.8% of Orbitol's attempts yielded stable or metastable, unique and novel structures that retained the requested space group after relaxation, compared with 10.8% for SGFM, the next best model. We also introduce SPARC, a score that combines retention of the raw detected space group with agreement between the retained-group and training distributions. Finally, we used Orbitol to generate candidates in piezoelectric crystal classes. Of the structures selected by internal screening, 89% passed our density functional theory checks, and 79% also matched no Materials Project or Alexandria entry with the same space group. Their median figure of merit was nine times that of the piezoelectric materials computed in the Materials Project, and 24% of the new candidates ranked in that set's top 10%.

0.20

Deep Generative Crystal Structure Prediction: A Benchmark Study and a Controlled Test of Prototype Dependence

cs.LG cond-mat.mtrl-sci ·2026-09-22 ·Lai Wei, Rongzhi Dong, Ying Feng, Madeline Miklos, Jianjun Hu
benchmark similarity 0.65
Abstract · 18 pages

Deep generative models are widely reported to enable de novo crystal structure prediction (CSP), but their capability has not been measured consistently against template-based methods. We evaluate 12 representative generative CSP models, spanning latent-variable, diffusion, flow-matching, autoregressive, and manifold random-walk architectures, against TCSP 2.0 on 180 test structures and a leakage-controlled subset of 46. All methods use identical structure-matching, symmetry, and consensus criteria. Template retrieval is the strongest single method, reaching 68.3% top-1 success; symmetry-aware EquiCSP (66.4%) and Uni-3DAR (62.9%) form the next tier. However, comparison with TCSP 2.0 shows that most structures correctly predicted by generative models are also correctly predicted by template substitution. Thus, the set of structures uniquely reachable by generation is small, limiting its practical advantage for discovering structures outside existing prototype libraries. To test the source of this performance, we removed entire stoichiometric prototype families from the training set and retrained the strongest generative model. Accuracy declined by 50-78% across four families, establishing that performance is substantially prototype-dependent. A small minority of structures survived removal of their prototype family, demonstrating a real but limited retrieval-independent predictive capacity. Present generative CSP models therefore function largely as implicit, softer-edged prototype libraries rather than genuinely de novo predictors. Enlarging this residual capacity, rather than aggregate match rate alone, is the central open problem.

0.31

Diffusion Tree Sampling for Inference Time Adaptation of Crystal Generative Models

cs.LG cond-mat.mtrl-scics.AI ·2026-11-12 ·Daniel T. Levy, Vineet Jain, Tara Akhound-Sadegh, Sékou-Oumar Kaba, Siamak Ravanbakhsh
top venue similarity 0.65
Abstract · 24 pages, 7 figures. Accepted to NeurIPS 2026

Diffusion-based generative models have become strong priors over plausible crystal structures, but turning that prior into promising candidates requires steering toward several conflicting properties at once. Fine-tuning is expensive and produces a new model for every new objective, while differentiable guidance is not applicable to many real-world tasks. We instead adapt these generative models at inference time using Diffusion Tree Sampling (DTS), a Monte Carlo tree search procedure over the denoising trajectory. We make two technical contributions on top of DTS that are particularly relevant for AI-for-science applications. First, we show how to use existing property-conditional generative models (e.g., models fine-tuned for classifier-free guidance) as guided proposals inside DTS, without distorting the target distribution. Second, we extend DTS to multi-objective sampling, so that a single search produces samples spanning an entire Pareto front. We instantiate both adaptations on de novo crystal generation: improving the stability of crystals generated by different models, using pretrained conditional models to generate crystals belonging to rare space groups or with high bulk modulus, and generating crystals along a Pareto front of high dielectric constant and high bandgap materials. Across these tasks, DTS-based inference-time adaptation successfully improves model outputs without modifying the models themselves.

0.29

uFlowCSP: Crystal Structure Prediction using Mean flow generative models

cs.AI cond-mat.mtrl-scics.LG ·2026-09-09 ·Sourin Dey, Dipannoy Das Gupta, Lai Wei, Sadman Sadeed Omee, Jianjun Hu
similarity 0.63
Abstract

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.

0.29

Synthetic Crystal Data Generation for Diffusion Pretraining in Material Property Prediction

cs.LG cond-mat.mtrl-sci ·2026-11-17 ·Aamod Varma, Shuyi Jia, Victor Fung
published similarity 0.63
Abstract · 14 pages, 2 figures, 6 tables. Accepted at the AI4Mat Workshop at NeurIPS 2026

Diffusion-based pretraining has emerged as a powerful strategy for learning transferable representations in graph neural networks for materials property prediction without requiring quantum-mechanical labels such as energies, forces, and stresses for every pretraining structure. However, the method retains a critical dependency: it requires structures residing near local minima of a potential energy surface, which have traditionally been obtained through costly DFT relaxations of known or physically synthesized compounds. This reliance constrains both the scale and the structural diversity of available pretraining data, leaving large regions of crystal structure space unexplored. In this work, we connect diffusion-based pretraining with synthetic structure generation to reduce this bottleneck. We introduce a synthetic data generation pipeline combining three complementary strategies: perturb-and-relax, generative sampling via MatterGen, and protostructure enumeration. Where structural relaxation is required, we use fast machine-learned interatomic potentials instead of performing a new DFT relaxation for each generated structure. We further compare a coordinate-denoising control with continuous-time variance-preserving (VP) and sub-VP diffusion objectives using a randomly initialized Orb-v2 backbone. The source comparison covers seven MatBench property-prediction tasks, and an eighth task, \texttt{mp\_gap}, is included in the protostructure scaling study. Every synthetic-pretraining condition numerically outperforms random initialization on all seven source-comparison tasks. Scaling the protostructure corpus from 20k to 165k under a fixed 400-epoch schedule numerically matches or exceeds MP-20k pretraining on most tasks; because this protocol increases both data diversity and optimization updates, the result reflects joint data-and-compute scaling.

0.27

Reinforcement Learning on the Discrete Composition Channel of a Crystal Generator: Validated Gains and Reward Hacking

cs.LG cond-mat.mtrl-sci ·2026-10-02 ·Pawan Prakash, Philipp Höllmer, Addis Fuhr, Peter Hirschfeld, P. Ganesh +2 more
benchmarkcode/data similarity 0.62
Abstract · 26 pages, 4 figures, 12 tables. Code: https://github.com/paprakash/OMatGRPO. Models and structures: https://huggingface.co/paprakash/OMatGRP…

Inverse materials design is a long-standing goal of computational materials discovery. Generative models for crystalline materials are typically trained to match the distribution of a structure database, while nothing in their training objective points them at specific design goals such as targeted properties. We use group-relative policy optimization (GRPO) to align a generative model based on stochastic interpolants and discrete flow matching with general black-box reward functions through reinforcement learning. Atom types are generated by a discrete flow and the policy gradient of our generalization of GRPO directly acts on the likelihoods of the atom-type transitions, which differentiates our work from previous reinforcement-learning approaches for diffusion and flow-based generative models of crystalline materials. We introduce a reward function that raises the yield of metastable, unique and novel structures (mSUN) from 13.4% for the pretrained model to 45.5% for the reinforced model, as evaluated by a community benchmark. Our reward also improves the performance of a reinforcement learning framework for crystalline materials based on latent denoising diffusion models. At the same time, we find that directly reinforcing atom-type transition likelihoods enables reward exploitation that has to be prevented with explicit guards. The same analysis also exposes a gap in the community metric. Single-element structures in distinct packings are counted as metastable, unique and novel materials and inflate mSUN without yielding any new compounds. A stability claim is only as good as its reference hull. We report every result split by the number of reference phases behind it and argue that benchmarks should do the same.

0.25

Data-driven candidate structure ensemble generation beyond the convex hull with crystALL approach

cs.CE cond-mat.mtrl-sci ·2026-11-02 ·Adam M. Krajewski, Ricardo Amaral, Jonathan W. Siegel, Luke A. Myers, Rushi Gong +2 more
code/data similarity 0.61
Abstract · 19 pages, 7 figures, software repository https://github.com/PhasesResearchLab/crystALL

Real materials consist of multiple stable or metastable states under real-world conditions, existing as statistical ergodic ensembles of configurations rather than ideal crystals. Yet, data-driven computational materials discovery increasingly biases itself toward finding "new stable structures" from ground-state DFT calculations alone. This limits our long-term ability to discover materials whose properties emerge from metastable phases, to explain experimental observations, or to model thermodynamic behavior comprehensively. Predicting ergodic sets of atomic configurations of known and hypothetical solids is a fundamental challenge in theoretical crystallography, approached via many complementary methods. We discuss this conceptually and introduce crystALL, a pragmatic and powerful data-driven approach to discovering ensembles beyond the convex hull, embracing structural diversity over stability alone, enabling robust experiment matching in cases evading traditional methods. crystALL leverages 5 million relaxed structures from OPTIMADE and MPDD to rapidly generate a comprehensive candidate set by: (1) species substitutions; (2) ML feature-space analysis with pySIPFENN; (3) representative structure selection via clustering and energy-minimization or property matching; and (4) construction of the final candidate ensemble for experimental or ab initio validation. We also uniquely address the common challenge of experimental composition uncertainties spanning many integer stoichiometries by new efficient searches across anonymized compositional ranges rather than fixed formulas. Overall, crystALL's ensemble-focused philosophy provides critical advantages over the prevailing convex-hull-searching paradigm, which often fails to account for key experimental observations and increases AI/ML discovery bias over time - akin to model collapse in LLMs retrained on iteratively generated content.

0.23