Pith. sign in

REVIEW 3 major objections 5 minor 82 references

Crystal structure prediction with host-guided inpainting generation and foundation potentials

T0 review · 3 major / 5 minor · reviewed 2026-08-16 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Conditional inpainting on a symmetrized host rescues symmetric crystal generation by diffusion models at large cell sizes, and the pipeline finds a new stable Li-Si polymorph.

desk verdict Useful engineering for host-guided inpainting, but the headline symmetry gain is partly baked in by the symmetrized host and loose spglib tolerances. read the letter →

arxiv 2504.16893 v2 pith:JW2OR6QE submitted 2025-04-23 cond-mat.mtrl-sci physics.comp-ph

Peichen Zhong , Xinzhe Dai , Bowen Deng , Gerbrand Ceder , Kristin A. Persson This is my paper · ORCID
classification cond-mat.mtrl-sciphysics.comp-ph
keywords crystalstructurepredictiondiffusionmodelsinpaintinggenerationsymmetryintercalationchemistrymachinelearninginteratomicpotentialsLi-Sialloysfoundation
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Diffusion-based generative models for crystals learn local bonding environments well but fail to assemble them into coherent, symmetric crystals once the unit cell grows, a failure the paper traces to the locality bias of graph neural networks. To fix this, the paper proposes conditional generation: a host framework is built by removing the atom type with the most flexible coordination (Zn or Li) and symmetrizing the remainder with loosened symmetry-matching tolerances, and the removed atoms are then placed back by diffusion inpainting. On the ZnS–P$_2$S$_5$ and Li–Si systems, the inpainting route yields a higher fraction of structures with nontrivial space groups than unconditional generation. After relaxation with a universal machine-learning interatomic potential, the workflow finds a Li$_5$Si$_2$ polymorph that sits below the known convex hull, along with low-energy metastable polymorphs. The practical payoff is that, if the scaffold is chemically meaningful, the same machinery offers a route to intercalation compounds, structural modification, and symmetry-preserving generation at larger cell sizes.

What carries the argument

Score-based diffusion on fractional coordinates with a variance-exploding SDE, where a custom SE(3)-equivariant graph neural network predicts the denoising score. The conditional component is the inpainting loop: a binary mask separates framework atoms from guest atoms, the framework is re-noised at each step to the correct noise level and blended back in, and resampling repeatedly revisits earlier diffusion steps so the guest positions harmonize with the host. The host itself is constructed by removing the flexible-coordination species and symmetrizing the remainder with a symmetry finder at tolerances up to site tolerance 2.0 and angle tolerance 30 degrees. A pretrained universal interatomic potential relaxes every generated structure and supplies decomposition energies for screening, with r2SCAN DFT used for final verification.

What would settle it

Take a host framework obtained at the loosest tolerances (site tolerance 2.0, angle tolerance 30 degrees), relax it with DFT or a tight-tolerance force field, and re-determine its space group at standard tolerances; if the refined symmetry collapses to P1 or P-1, the scaffold that guides inpainting is an artifact. A complementary test is to run inpainting into a deliberately randomized, P1 version of the same host and compare symmetry success rates and decomposition energies with the symmetrized version.

Watch

Extended reading notes

Core claim

The paper's central claim is that the failure of unconditional diffusion models to produce symmetric crystals at large cell sizes is not fundamental to diffusion itself but follows from the locality of the graph neural network that provides the score: the model reproduces correct short-range motifs yet outputs disordered mosaics because it never couples atomic positions to long-range crystallographic order. The remedy is to supply that order externally. CHGGen first generates and relaxes candidate structures, removes the species with the broadest coordination distribution, symmetrizes the remaining framework by repeatedly loosening the tolerance of a symmetry finder, and then runs masked inpainting so that only the guest atoms are diffused inside that fixed, high-symmetry host. In the two demonstrated systems, inpainting raises the symmetry success rate far above the unconditional baseline of under five percent, and the downstream relaxation-plus-DFT pipeline identifies Li$_5$Si$_2$ in space group $R\bar{3}m$ as a thermodynamically stable phase below the known convex hull.

Load-bearing premise

The load-bearing premise is that deleting the atom type with the broadest coordination distribution and symmetrizing what remains at strongly loosened matching tolerances produces a real crystallographic scaffold rather than relabelling a disordered configuration as a higher-symmetry space group.

Editorial extensions

If this is right

  • Inpainting on a symmetrized host should allow diffusion models to generate ordered crystals well beyond the roughly 20-atom scale where unconditional generation degrades.
  • Host-guest generation gives a direct route to intercalation compounds and partially occupied structures: fix the known framework and let diffusion place the mobile or interstitial species.
  • Because conditional and unconditional generation differ only in the mask applied during reverse diffusion, any future generative model can inherit the symmetry-improving strategy without retraining.
  • Foundation-potential relaxation tightens generated structures into local minima, but its systematic softening means screening thresholds for DFT follow-up should be set well below the usual 0.1 eV per atom; the paper suggests roughly 30 meV per atom.
  • The Li-Si case shows the pipeline can propose phases that fall below the known convex hull and are later confirmed by DFT, so generative inpainting can serve as a hypothesis generator for phase-diagram completion.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • If the scaffold-symmetrization step is chemically sound rather than a tolerance artifact, the same host-guided conditional scheme should transfer to substitutional disorder, defective frameworks, and interfaces where the bulk stays fixed, territories where unconditional generation is currently unreliable.
  • A direct test of the method's ability to place guests correctly would be to inpaint lithium into a known intercalation host, such as a lithiated chloride or sulfide framework, and compare the generated lithium sites and ordering with experimentally determined ones; the paper demonstrates the workflow but does not quantify site-level accuracy.
  • The symmetry success metric counts any space group above P1 or P-1; future work could tighten the metric to require specific target space groups or Wyckoff positions, which would separate genuine prototype discovery from mere tolerance-assisted ordering.
  • Combining CHGGen with explicit symmetry-constrained or Wyckoff-position-based generation could push the recovered symmetries from the moderate monoclinic cases reported here toward more complex framework prototypes such as NASICON-type structures.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The paper proposes CHGGen, a conditional crystal generation framework that combines score-based diffusion inpainting with a symmetrized host structure and the CHGNet foundation potential for relaxation and thermodynamic screening. The workflow removes atoms with broad coordination distributions (Zn or Li), symmetrizes the remaining framework with spglib using progressively looser tolerances, inpaints guest atoms into that framework, and relaxes the resulting structures with CHGNet before selected r2SCAN DFT verification. The method is demonstrated on the Zn-P-S and Li-Si chemical systems. The central claim is that inpainting into a symmetrized host yields a higher fraction of non-P1/non-P-1 crystal structures than unconditional generation, with a DFT-confirmed stable Li5Si2 polymorph in the Li-Si system.

Significance. If the reported symmetry gain is robust, CHGGen would be a practical modular tool for host-guided structure generation, particularly for intercalation chemistry and partial-occupancy systems, and the availability of code is a genuine strength. The paper also gives a clean demonstration of using a foundation potential for relaxation and screening in a generative pipeline. However, the central comparison is currently weakened by the construction of the host, the loose spglib tolerances used both to create and to score symmetric structures, and the lack of statistical grounding. The authors themselves state in Section IV that the symmetry refinement approach 'remains preliminary' and that it 'relies on spglib by simply increasing the tolerance threshold,' which directly bears on the headline claim.

major comments (3)
  1. [III.B and Methods; Eqs. (7) and (8)] The headline symmetry-success comparison in Figures 4f and 6 is partly built into the algorithm. The host framework is symmetrized by spglib with site tolerance up to 2.0 Å and angle tolerance up to 30°, and during every reverse step Eq. (7) re-noises the host positions as x_host_{t-1} = x_host_0 + σ_{t-1} z, followed by the mask combination in Eq. (8). The final host therefore retains the imposed symmetry by construction, and the guest atoms are the only free part. To support the claim that inpainting 'generates a higher fraction of symmetric structures than unconditional generation,' the authors need at least an unsymmetrized-host control, a report of the root-mean-squared displacement between the initial relaxed framework and its symmetrized projection, and a success-rate evaluation at strict spglib tolerances (for example, site tolerance no larger than 0.01 Å). Without these, the apparent improvement may be an artifact of tolerance inflation rather than a genuine crystallographic ordering effect.
  2. [Figure 4f, Figure 6, and Methods (success rate)] The success-rate metric counts every non-P1/non-P-1 space group as success, including C2 and Cm, and the authors acknowledge in Section IV that the method 'predominantly yields structures with moderate symmetry (e.g., C2, Cm).' The comparison also lacks sample counts, error bars, and any statistical test, so the statement that inpainting achieves 'significantly higher' success rates is not quantitatively supported. Please report the number of generated structures per condition, the distribution of assigned space groups, and a stricter definition of success (for example, only the crystal system or Laue class expected from the intended host framework). A bootstrap confidence interval on the success-rate difference would also be appropriate.
  3. [Section IV (Discussion)] The manuscript contains an explicit limitation statement: 'the current symmetry refinement approach remains preliminary as it relies on spglib by simply increasing the tolerance threshold.' This concession is in direct tension with the abstract's claim that the inpainting method 'generates a higher fraction of symmetric structures than unconditional generation.' As written, the central claim is defensible only under the assumption that loosened spglib tolerances produce chemically meaningful scaffolds, which is exactly what the skeptical reader will question. The authors should either provide evidence that the symmetrized frameworks correspond to genuine crystallographic scaffolds (for example, by comparing them to known structure prototypes or by reporting displacement distributions), or reframe the headline claim as a demonstration of conditional generation conditioned on a user-supplied host rather than as a general symmetry-success improvement.
minor comments (5)
  1. [Figure 6 caption] The caption contains a typo: 'GHGGen' should be 'CHGGen'.
  2. [Methods, Model architecture] The library name 'pytorch-lighting' should be 'pytorch-lightning'.
  3. [Section IV, first paragraph of final discussion] The phrase 'often relys' should be 'often relies'.
  4. [Algorithm, Inpainting Generation] The resampling 'jump back' step applies z~N(0,1) and then sets x_t ← x_{t-1} + sqrt(σ_{t-1}^2 - σ_{t-2}^2) z, but the text and reference do not clarify whether this is intended as the forward transition under the VE-SDE; a sentence explaining the consistency of this step with the sampler would improve reproducibility.
  5. [Section III.D] The Li5Si2 (R-3m) structure is acknowledged as having been identified previously by Tipton et al. and Morris et al., so the paper should present it as a validation of the pipeline rather than as a de novo discovery; the current phrasing 'successfully predicted' is acceptable but could be sharpened to avoid overclaiming novelty.

Circularity Check

1 steps flagged · score 6.0 of 10

Symmetry gain is largely inherited from the symmetrized host by construction: Eqs. (7)-(8) reset the host to x_host_0 at every reverse step, so the reported improvement in symmetry success rate is substantially a built-in consequence of conditioning on a symmetrized framework.

  1. self definitional [Section II (Inpainting), Eqs. (7)-(8); Section III.B and Methods (host symmetrization and symmetry success rate)]
    "The remaining structure (framework) undergoes symmetry refinement using spglib through incremental structural matching tolerance to obtain a space group with higher symmetry (i.e., until the space group is not P1). ... xhost_{t−1} = x host_0 + σ_{t−1}z, z∼N(0,I). ... xt−1 = (1−m)⊙x host_{t−1} + m⊙x_{t−1} ... resulting in a final crystal structure that is closely aligned with the original host structure with minimal deviation (e.g., σ_min = 0.001 ˚A)."

    The host is explicitly symmetrized before inpainting, and Eq. (7) resets the host atoms to x_host_0 + σ_{t−1}z at every reverse step. At the final step σ_min = 0.001 Å, so the host sublattice is, to within 0.001 Å, exactly the symmetrized input host. Eq. (8) then overwrites the host portion of the denoised configuration with this fixed host. Thus the host's space group is supplied by the input and preserved by construction; the measured 'success rate of identifying symmetric crystal structures' largely measures this imposed host symmetry rather than symmetry discovered by the generative model. The only free degrees of freedom are the guest atoms, which may or may not break the host symmetry.

full rationale

The paper's thermodynamic predictions (e.g., Li5Si2 R-3m and C2/m metastable polymorphs) are checked by independent r2SCAN-DFT calculations and are not circular. CHGNet [12] is an external pretrained potential with its own benchmarks, and the same-group workshop report [44] is not load-bearing for the symmetry claim. The circularity is confined to the headline symmetry-success comparison: the host is deliberately symmetrized before generation, and Eqs. (7)-(8) reset the host sublattice to that symmetrized x_host_0 at every reverse step with final σ_min = 0.001 Å. The full structure's space group therefore inherits the host's imposed symmetry unless the newly placed guest atoms break it; the measured 'higher fraction of symmetric structures' is substantially a built-in consequence of the conditional input, not a symmetry discovered by the diffusion model. The authors' own Section IV caveat that the symmetry refinement 'remains preliminary as it relies on spglib by simply increasing the tolerance threshold', together with the tolerance escalation used for the host (stol up to 2.0 Å, angle up to 30°) and the final scoring at 0.1 Å/10°, reinforces that input and output share the same permissive symmetry criterion. This is partial circularity in the central claim, leading to a score of 6.

Assumptions & free parameters 6 free parameters · 6 assumptions · 0 invented entities

All non-trivial inputs are listed. The free parameters control lattice volume, sampling noise, and the symmetry metric itself; the most consequential is the spglib tolerance range because it directly defines whether a structure counts as symmetric. The domain assumptions concern the transferability of a Materials Project-trained diffusion model and CHGNet to target chemistries. No new physical entities are introduced.

free parameters (6)
  • Atomic volume prior V0 = 24 Å3 (Zn-P-S), 16 Å3 (Li-Si)
    Set by hand in Methods; determines all sampled lattice volumes before relaxation, so it controls the volume distribution of generated structures.
  • Signal-to-noise ratio delta = 0.4
    Set in Hyperparameters; controls the ancestral sampling and Langevin corrector, affecting generated structure quality and symmetry likelihood.
  • spglib matching tolerances = stol 0.1 to 2.0; angle 10 to 30 degrees
    Tolerance inflation is the mechanism that lifts structures out of P1; the symmetry success rate is directly sensitive to these chosen thresholds.
  • Energy threshold for DFT screening = Ed < 0.1 eV/atom (30 meV/atom discussed)
    Structures below this hand-set threshold are submitted to DFT; it affects which candidates are reported as stable or metastable.
  • Space-group exclusion sets for success metric = {P1,P-1} for Zn-P-S; {P1,P-1,Pm} for Li-Si
    The metric definition in Methods differs between systems, making cross-system success rates not directly comparable.
  • Diffusion sampling steps (T, M, r) = not specified in main text
    Number of predictor, corrector, and resampling steps are fixed in the implementation but not reported; they affect sample quality and symmetry yield.
assumptions (6)
  • standard math The denoising score matching objective Eq. (6) yields an accurate estimate of the score function for periodic fractional-coordinate crystals.
    Standard score-matching theory from Refs [38-40] is invoked without proof.
  • domain assumption The Materials Project training set with Ehull < 0.1 eV is a sufficient distribution for transfer to Zn-P-S and Li-Si chemistries.
    The model is trained on 109,805 MP structures; transferability to the target chemistries is assumed.
  • domain assumption CHGNet energy predictions and relaxations are accurate enough (within about 30 meV/atom) to screen thermodynamic stability before DFT.
    CHGNet is the workhorse for relaxation and Ed screening; the paper documents its average error and systematic softening behavior.
  • ad hoc to paper Removing the atoms with broad coordination distributions (Zn or Li) leaves a chemically meaningful framework that can be symmetrized by spglib.
    This is the core structural premise of CHGGen; no independent validation is given that the remaining P-S or Si framework is a stable crystallographic scaffold.
  • standard math The masked resampling scheme (RePaint) converges to the conditional distribution of guest atoms given the host framework.
    Relies on Refs [43,44]; the theory is not re-derived in this paper.
  • domain assumption spglib tolerance inflation yields physically meaningful space groups.
    The paper itself concedes this is preliminary and mostly yields monoclinic C2/Cm frameworks.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Crystal structure prediction with host-guided inpainting generation and foundation potentials." pith.science (2026). https://pith.science/paper/JW2OR6QE

@misc{pith2026250416893,
  author       = {Pith},
  title        = {Pith review of: Crystal structure prediction with host-guided inpainting generation and foundation potentials},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/JW2OR6QE}},
  note         = {Machine review of arXiv:2504.16893}
}
abstract

Unconditional crystal structure generation with diffusion models faces challenges in identifying symmetric crystals as the unit cell size increases. We present the Crystal Host-Guided Generation (CHGGen) framework to address this challenge through conditional generation using an inpainting method, which optimizes a fraction of atomic positions within a predefined and symmetrized host structure to improve the success rate for symmetric structure generation. By integrating inpainting structure generation with a foundation potential for structure optimization, we demonstrate the method on the ZnS-P$_2$S$_5$ and Li-Si chemical systems, where the inpainting method generates a higher fraction of symmetric structures than unconditional generation. The practical significance of CHGGen extends to enabling the structural modification of crystal structures, particularly for systems with partial occupancy or intercalation chemistry. The inpainting method also allows for seamless integration with other generative models, providing a versatile framework for accelerating materials discovery.

Figures

Figures reproduced from arXiv: 2504.16893 by the authors.

Figure 1
Figure 1. FIG. 1 [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. e, P atoms predominantly occupy tetrahedral sites (4-coordinated), consistent with known Li–P–S crystal structures [46]. In contrast, Li exhibits a broad dis￾tribution of coordination numbers, with peaks at 5- fold (∼40%), 4-fold (∼25%), and 6-fold (∼30%) geome￾tries. Notably, these coordination statistics qualitatively align with patterns observed in the MP training dataset (Fig. 2f), where P atoms maintain rigid t… view at source ↗
Figure 3
Figure 3. illustrates the CHGGen computational work￾flow. The process begins with sampling various Bravais lattices at a fixed volume through a random search over lattice constant ratios and angles (see Methods in SI). The unit cell volume is determined as N ∗ V0, where N represents the number of atoms and V0 denotes the atomic volume. The V0 can be initialized either from related crystalline phases or predicted by compositio… view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: c presents the CHGNet-predicted decomposi￾tion energies (Ed) to quantify the thermodynamic stabil￾ity. The distribution shows a median Ed of 0.07 eV/atom, indicating that the majority of generated structures are metastable in this chemical space, while 6.5% of the gen-…
Figure 5
Figure 5. Figure 5: FIG. 5 [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: FIG. 6. Comparison of success rates for identifying symmet [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

82 extracted references · 56 canonical work pages

  1. [1]

    Z. Lu, B. Zhu, B. W. B. Shires, D. O. Scanlon, and C. J. Pickard, Ab initio random structure searching for battery cathode materials, The Journal of Chemical Physics154 (2021)

    2021

  2. [2]

    D. Zhou, I. Bier, B. Santra, L. D. Jacobson, C. Wu, A. Garaizar Suarez, B. R. Almaguer, H. Yu, R. Abel, R. A. Friesner, and L. Wang, A robust crystal structure prediction method to support small molecule drug de- velopment with large scale validation and blind study, Nature Communications16, 2210 (2025)

    work page 2025

  3. [3]

    Zhang, T

    Z. Zhang, T. Cui, M. J. Hutcheon, A. M. Shipley, H. Song, M. Du, V. Z. Kresin, D. Duan, C. J. Pickard, and Y. Yao, Design Principles for High-Temperature Su- perconductors with a Hydrogen-Based Alloy Backbone at Moderate Pressure, Physical Review Letters128, 047001 (2022)

    work page 2022

  4. [4]

    V. V. Gusev, D. Adamson, A. Deligkas, D. Antypov, C. M. Collins, P. Krysta, I. Potapov, G. R. Darling, M. S. Dyer, P. Spirakis, and M. J. Rosseinsky, Optimality guar- antees for crystal structure prediction, Nature619, 68 (2023)

    work page 2023

  5. [5]

    C. J. Pickard and R. Needs, Ab initio random struc- ture searching, Journal of Physics: Condensed Matter 23, 053201 (2011). 11

    work page 2011

  6. [6]

    A. R. Oganov and C. W. Glass, Crystal structure predic- tion using ab initio evolutionary techniques: Principles and applications, The Journal of chemical physics124 (2006)

    work page 2006

  7. [7]

    Y. Wang, J. Lv, L. Zhu, and Y. Ma, Crystal structure prediction via particle-swarm optimization, Phys. Rev. B82, 094116 (2010)

    work page 2010

  8. [8]

    Hautier, C

    G. Hautier, C. Fischer, V. Ehrlacher, A. Jain, and G. Ceder, Data Mined Ionic Substitutions for the Dis- covery of New Compounds, Inorganic Chemistry50, 656 (2011)

    work page 2011

Show all 82 references
  1. [9]

    Huang, D

    W. Huang, D. A. Kitchaev, S. T. Dacek, Z. Rong, A. Ur- ban, S. Cao, C. Luo, and G. Ceder, Finding and proving the exact ground state of a generalized Ising model by convex optimization and MAX-SAT, Physical Review B 94, 134424 (2016)

    2016

  2. [10]

    Ferrari, F

    A. Ferrari, F. K¨ ormann, M. Asta, and J. Neugebauer, Simulating short-range order in compositionally complex materials, Nature Computational Science3, 221 (2023)

    2023

  3. [11]

    Chen and S

    C. Chen and S. P. Ong, A universal graph deep learn- ing interatomic potential for the periodic table, Nature Computational Science2, 718 (2022)

    2022

  4. [12]

    B. Deng, P. Zhong, K. Jun, J. Riebesell, K. Han, C. J. Bartel, and G. Ceder, CHGNet as a pretrained universal neural network potential for charge-informed atomistic modelling, Nature Machine Intelligence5, 1031 (2023)

    2023

  5. [13]

    Batatia, P

    I. Batatia, P. Benner, Y. Chiang, A. M. Elena, D. P. Kov´ acs, J. Riebesell, X. R. Advincula, M. Asta, W. J. Baldwin, N. Bernstein,et al., A foundation model for atomistic materials chemistry, arXiv preprint arXiv:2401.00096 (2023)

    2023 arXiv

  6. [14]

    J. Kim, J. Kim, J. Kim, J. Lee, Y. Park, Y. Kang, and S. Han, Data-Efficient Multifidelity Training for High- Fidelity Machine Learning Interatomic Potentials, Jour- nal of the American Chemical Society147, 1042 (2025)

    2025

  7. [15]

    Merchant, S

    A. Merchant, S. Batzner, S. S. Schoenholz, M. Aykol, G. Cheon, and E. D. Cubuk, Scaling deep learning for materials discovery, Nature624, 80 (2023)

    2023

  8. [16]

    Zhang, X

    D. Zhang, X. Liu, X. Zhang, C. Zhang, C. Cai, H. Bi, Y. Du, X. Qin, A. Peng, J. Huang,et al., DPA-2: a large atomic model as a multi-task learner, npj Comput. Mater.10, 293 (2024)

    2024

  9. [17]

    H. Yang, C. Hu, Y. Zhou, X. Liu, Y. Shi, J. Li, G. Li, Z. Chen, S. Chen, C. Zeni,et al., Mattersim: A deep learning atomistic model across elements, temperatures and pressures, arXiv preprint arXiv:2405.04967 (2024)

    2024 arXiv

  10. [18]

    X. Fu, B. M. Wood, L. Barroso-Luque, D. S. Levine, M. Gao, M. Dzamba, and C. L. Zitnick, Learning smooth and expressive interatomic potentials for physical prop- erty prediction, arXiv preprint arXiv:2502.12147 (2025)

    2025 arXiv

  11. [19]

    Z. Ren, S. I. P. Tian, J. Noh, F. Oviedo, G. Xing, J. Li, Q. Liang, R. Zhu, A. G. Aberle, S. Sun, X. Wang, Y. Liu, Q. Li, S. Jayavelu, K. Hippalgaonkar, Y. Jung, and T. Buonassisi, An invertible crystallographic repre- sentation for general inverse design of inorganic crystals...

    2022

  12. [20]

    H. Park, A. Onwuli, and A. Walsh, Exploration of crys- tal chemical space using text-guided generative artificial intelligence, chemRxiv (2024)

    2024

  13. [21]

    Cheng, Response Matching for Generating Materials and Molecules, Journal of Chemical Theory and Compu- tation20, 9259 (2024)

    B. Cheng, Response Matching for Generating Materials and Molecules, Journal of Chemical Theory and Compu- tation20, 9259 (2024)

    2024

  14. [22]

    Al- though promising, CDVAE-generated structures are pre- dominantly thermodynamically unstable or lack symme- try [23, 24]

    introduced CDVAE that uses a variational autoen- coder to sample lattice parameters and compositions and a diffusion model to optimize atomic coordinates. Al- though promising, CDVAE-generated structures are pre- dominantly thermodynamically unstable or lack symme- try [23, 24...

    2025 arXiv

  15. [23]

    N. J. Szymanski and C. J. Bartel, Establishing base- lines for generative discovery of inorganic crystals, arXiv preprint arXiv:2501.02144 (2025)

    2025 arXiv

  16. [24]

    T. Xie, X. Fu, O.-E. Ganea, R. Barzilay, and T. Jaakkola, Crystal diffusion variational autoencoder for periodic ma- terial generation, International Conference on Learning Representations (ICLR) (2021)

    2021

  17. [25]

    G. Kurz, I. Gilitschenski, and U. D. Hanebeck, Effi- cient evaluation of the probability density function of a wrapped normal distribution, in2014 Sensor Data Fusion: Trends, Solutions, Applications (SDF)(IEEE,

  18. [26]

    D. Levy, S. S. Panigrahi, S.-O. Kaba, Q. Zhu, K. L. K. Lee, M. Galkin, S. Miret, and S. Ravanbakhsh, SymmCD: Symmetry-preserving crystal generation with diffusion models, International Conference on Learning Represen- tations (ICLR) (2025)

    2025

  19. [27]

    and solid-solid interface in Supplementary Informa- tion. A. Locality bias of GNN-based generative models To investigate the limitations of unconditional gen- eration at large scales, we generated 10 supercell con- figurations of (Li 2S-P2S5)10 and (Li 4S2-P2S5)8 with the para...

    2023

  20. [28]

    V. D. Bortoli, E. Mathieu, M. Hutchinson, J. Thornton, Y. W. Teh, and A. Doucet, Riemannian score-based gen- erative modelling, Advances in Neural Information Pro- cessing Systems (NeurIPS) (2022)

    2022

  21. [29]

    R. Jiao, W. Huang, P. Lin, J. Han, P. Chen, Y. Lu, and Y. Liu, Crystal structure prediction by joint equivari- ant diffusion, Advances in Neural Information Processing Systems (NeurIPS) (2023)

    2023

  22. [30]

    C. Zeni, R. Pinsler, D. Z¨ ugner, A. Fowler, M. Horton, X. Fu, Z. Wang, A. Shysheya, J. Crabb´ e, S. Ueda, R. Sor- dillo, L. Sun, J. Smith, B. Nguyen, H. Schulz, S. Lewis, C.-W. Huang, Z. Lu, Y. Zhou, H. Yang, H. Hao, J. Li, C. Yang, W. Li, R. Tomioka, and T. Xie, A generative...

    2025

  23. [31]

    A. Jain, S. P. Ong, G. Hautier, W. Chen, W. D. Richards, S. Dacek, S. Cholia, D. Gunter, D. Skinner, G. Ceder, and K. A. Persson, Commentary: The Materials Project: A materials genome approach to accelerating materials innovation, APL Materials1, 011002 (2013)

    2013

  24. [32]

    Schmidt, N

    J. Schmidt, N. Hoffmann, H. Wang, P. Borlido, P. J. M. A. Carri¸ co, T. F. T. Cerqueira, S. Botti, and M. A. L. Marques, Machine-Learning-Assisted Determination of the Global Zero-Temperature Phase Diagram of Mate- rials, Advanced Materials35(2023)

    2023

  25. [33]

    S. Yang, K. Cho, A. Merchant, P. Abbeel, D. Schu- urmans, I. Mordatch, and E. D. Cubuk, Scal- able diffusion for materials generation, arXiv preprint arXiv:2311.09235 (2023)

    2023 arXiv

  26. [34]

    Aykol, A

    M. Aykol, A. Merchant, S. Batzner, J. N. Wei, and E. D. Cubuk, Predicting emergence of crystals from amorphous precursors with deep learning potentials, Nature Compu- tational Science5, 105 (2024)

    2024

  27. [35]

    Gruver, A

    N. Gruver, A. Sriram, A. Madotto, A. G. Wilson, C. L. Zitnick, and Z. Ulissi, Fine-tuned language models gener- ate stable inorganic materials as text, International Con- ference on Learning Representations (ICLR) (2024)

    2024

  28. [36]

    L. M. Antunes, K. T. Butler, and R. Grau-Crespo, Crystal structure generation with autoregressive large language modeling, Nature Communications15, 10570 (2024)

    2024

  29. [37]

    H. H. Li, J.-X. Shen, and K. A. Persson, A rapid lithium- ion cathode discovery pipeline and its exemplary appli- cation, Energy Advances3, 255 (2024)

    2024

  30. [38]

    Y.-L. Liao, B. Wood, A. Das, and T. Smidt, Equiformerv2: Improved equivariant transformer for scaling to higher-degree representations, International Conference on Learning Representations (ICLR) (2024)

    2024

  31. [39]

    Rønne, A

    N. Rønne, A. Aspuru-Guzik, and B. Hammer, Generative diffusion model for surface structure discovery, Physical 12 Review B110, 235427 (2024)

    2024

  32. [40]

    Y. Song, J. Sohl-Dickstein, D. P. Kingma, A. Kumar, S. Ermon, and B. Poole, Score-based generative mod- eling through stochastic differential equations, Interna- tional Conference on Learning Representations (ICLR) (2020)

    2020

  33. [41]

    Vincent, A Connection Between Score Matching and Denoising Autoencoders, Neural Computation23, 1661 (2011)

    P. Vincent, A Connection Between Score Matching and Denoising Autoencoders, Neural Computation23, 1661 (2011)

    2011

  34. [42]

    J. Ho, A. Jain, and P. Abbeel, Denoising diffusion proba- bilistic models, Advances in Neural Information Process- ing Systems (NeurIPs)33, 6840 (2020)

    2020

  35. [43]

    C. Duan, Y. Du, H. Jia, and H. J. Kulik, Accurate tran- sition state generation with an object-aware equivari- ant elementary reaction diffusion model, arXiv preprint arXiv:2304.06174 (2023)

    2023 arXiv

  36. [44]

    M. Li, R. Okabe, M. Cheng, A. Chottratanapituk, N. T. Hung, X. Fu, B. Han, Y. Wang, W. Xie, R. Cava,et al., Structural constraint integration in generative model for discovery of quantum material candidates, (2024)

    2024

  37. [45]

    Lugmayr, M

    A. Lugmayr, M. Danelljan, A. Romero, F. Yu, R. Tim- ofte, and L. Van Gool, RePaint: Inpainting using De- noising Diffusion Probabilistic Models, IEEE/CVF Con- ference on Computer Vision and Pattern Recognition (CVPR) (2022)

    2022

  38. [46]

    X. Dai, P. Zhong, B. Deng, Y. Chen, and G. Ceder, In- painting crystal structure generations with score-based denoising, inICML 2024 AI for Science Workshop (2024)

    2024

  39. [47]

    Waroquiers, X

    D. Waroquiers, X. Gonze, G. M. Rignanese, C. Welker- Nieuwoudt, F. Rosowski, M. G¨ obel, S. Schenk, P. Degel- mann, R. Andr´ e, R. Glaum, and G. Hautier, Statistical analysis of coordination environments in Oxides, Chem- istry of Materials29, 8346 (2017)

    2017

  40. [48]

    B. Lee, K. Jun, B. Ouyang, and G. Ceder, Weak Cor- relation between the Polyanion Environment and Ionic Conductivity in Amorphous Li–P–S Superionic Conduc- tors, Chemistry of Materials35, 891 (2023)

    2023

  41. [49]

    S. Gong, K. Yan, T. Xie, Y. Shao-Horn, R. Gomez- Bombarelli, S. Ji, and J. C. Grossman, Examining graph neural networks for crystal structures: Limitations and opportunities for capturing periodicity, Science Advances 9, eadi3245 (2023)

    2023

  42. [50]

    R. E. A. Goodall and A. A. Lee, Predicting materials properties without crystal structure: deep representation learning from stoichiometry, Nature Communications11, 6280 (2020)

    2020

  43. [51]

    Szczuka, B

    C. Szczuka, B. Karasulu, M. F. Groh, F. N. Sayed, T. J. Sherman, J. D. Bocarsly, S. Vema, S. Menkin, S. P. Emge, A. J. Morris, and C. P. Grey, Forced Disorder in the Solid Solution Li 3 P–Li 2 S: A New Class of Fully Reduced Solid Electrolytes for Lithium Metal Anodes, Journal...

    2022

  44. [52]

    B. Deng, Y. Choi, P. Zhong, J. Riebesell, S. Anand, Z. Li, K. Jun, K. A. Persson, and G. Ceder, Systematic soften- ing in universal machine learning interatomic potentials, npj Computational Materials11, 9 (2025)

    2025

  45. [53]

    D. H. S. Tan, Y.-T. Chen, H. Yang, W. Bao, B. Sreenarayanan, J.-M. Doux, W. Li, B. Lu, S.-Y. Ham, B. Sayahpour, J. Scharf, E. A. Wu, G. Deysher, H. E. Han, H. J. Hah, H. Jeong, J. B. Lee, Z. Chen, and Y. S. Meng, Carbon-free high-loading silicon anodes enabled by sulfide solid...

    2021

  46. [54]

    Artrith, A

    N. Artrith, A. Urban, and G. Ceder, Constructing first- principles phase diagrams of amorphous Li x Si using machine-learning-assisted sampling with an evolutionary algorithm, The Journal of Chemical Physics148(2018)

    2018

  47. [55]

    W. W. Tipton, C. R. Bealing, K. Mathew, and R. G. Hennig, Structures, phase stabilities, and electrical po- tentials of li-si battery anode materials, Phys. Rev. B 87, 184114 (2013)

    2013

  48. [56]

    A. J. Morris, C. P. Grey, and C. J. Pickard, Thermo- dynamically stable lithium silicides and germanides from density functional theory calculations, Phys. Rev. B90, 054111 (2014)

    2014

  49. [57]

    Sivonxay, M

    E. Sivonxay, M. Aykol, and K. A. Persson, The lithiation process and Li diffusion in amorphous SiO 2 and Si from first-principles, Electrochimica Acta331, 135344 (2020)

    2020

  50. [58]

    H. Park, Z. Li, and A. Walsh, Has generative artificial intelligence solved inverse materials design?, Matter7, 2355 (2024)

    2024

  51. [59]

    Gasteiger, F

    J. Gasteiger, F. Becker, and S. G¨ unnemann, Gem- Net: Universal Directional Graph Neural Networks for Molecules, Advances in Neural Information Processing Systems (NeurIPs) (2021)

    2021

  52. [60]

    Batzner, A

    S. Batzner, A. Musaelian, L. Sun, M. Geiger, J. P. Mailoa, M. Kornbluth, N. Molinari, T. E. Smidt, and B. Kozinsky, E(3)-equivariant graph neural networks for data-efficient and accurate interatomic potentials, Nature Communications13, 2453 (2022)

    2022

  53. [61]

    Zaidi, M

    S. Zaidi, M. Schaarschmidt, J. Martens, H. Kim, Y. W. Teh, A. Sanchez-Gonzalez, P. Battaglia, R. Pascanu, and J. Godwin, Pre-training via denoising for molecular property prediction, International Conference on Learn- ing Representations (ICLR) (2023)

    2023

  54. [62]

    Riebesell, R

    J. Riebesell, R. E. A. Goodall, P. Benner, Y. Chiang, B. Deng, G. Ceder, M. Asta, A. A. Lee, A. Jain, and K. A. Persson, Matbench discovery – a framework to evaluate machine learning crystal stability predictions, arXiv preprint arXiv:2308.14920 (2023)

    2023 arXiv

  55. [63]

    Y.-L. Liao, T. Smidt, M. Shuaibi, and A. Das, Gen- eralizing Denoising to Non-Equilibrium Structures Im- proves Equivariant Force Fields, Transactions on Ma- chine Learning Research (TMLR) (2024)

    2024

  56. [64]

    J. Gan, P. Zhong, Y. Du, Y. Zhu, C. Duan, H. Wang, C. P. Gomes, K. A. Persson, D. Schwalbe-Koda, and W. Wang, Large language models are innate crystal structure generators, arXiv preprint arXiv:2502.20933 (2025)

    2025

  57. [65]

    Yinet al., A LaCl 3-based lithium superionic con- ductor compatible with lithium metal, Nature616, 77 (2023)

    Y.-C. Yinet al., A LaCl 3-based lithium superionic con- ductor compatible with lithium metal, Nature616, 77 (2023)

    2023

  58. [66]

    X. He, Q. Bai, Y. Liu, A. M. Nolan, C. Ling, and Y. Mo, Crystal structural framework of lithium super-ionic con- ductors, Advanced Energy Materials9, 1902078 (2019)

    2019

  59. [67]

    J.-X. Shen, M. Horton, and K. A. Persson, A charge- density-based general cation insertion algorithm for gen- erating new li-ion cathode materials, npj Computational Materials6, 161 (2020)

    2020

  60. [68]

    R. Zhu, W. Nong, S. Yamazaki, and K. Hippalgaonkar, Wycryst: Wyckoff inorganic crystal generator frame- work, arXiv preprint arXiv:2311.17916 (2023)

    2023 arXiv

  61. [69]

    Z. Cao, X. Luo, J. Lv, and L. Wang, Space group in- formed transformer for crystalline materials generation, arXiv preprint arXiv:2403.15734 (2024)

    2024

  62. [70]

    Kazeev, W

    N. Kazeev, W. Nong, I. Romanov, R. Zhu, A. Ustyuzhanin, S. Yamazaki, and K. Hippalgaonkar, 13 Wyckoff transformer: Generation of symmetric crystals, arXiv preprint arXiv:2503.02407 (2025)

    2025

  63. [71]

    T. Hsu, B. Sadigh, N. Bertin, C. W. Park, J. Chap- man, V. Bulatov, and F. Zhou, Score-based denoising for atomic structure identification, npj Computational Ma- terials10, 155 (2024)

    2024

  64. [72]

    Geiger and T

    M. Geiger and T. Smidt, e3nn: Euclidean neural net- works, arXiv preprint arXiv:2207.09453 (2022)

    2022 arXiv

  65. [73]

    C. J. Bartel, A. Trewartha, Q. Wang, A. Dunn, A. Jain, and G. Ceder, A critical examination of compound stabil- ity predictions from machine-learned formation energies, npj Computational Materials6, 97 (2020)

    2020

  66. [74]

    H.-C. Wang, S. Botti, and M. A. L. Marques, Predicting stable crystalline compounds using chemical similarity, npj Computational Materials7, 12 (2021)

    2021

  67. [75]

    S. P. Ong, W. D. Richards, A. Jain, G. Hautier, M. Kocher, S. Cholia, D. Gunter, V. L. Chevrier, K. A. Persson, and G. Ceder, Python Materials Genomics (py- matgen): A robust, open-source python library for mate- rials analysis, Computational Materials Science68, 314 (2013)

    2013

  68. [76]

    Bitzek, P

    E. Bitzek, P. Koskinen, F. G¨ ahler, M. Moseler, and P. Gumbsch, Structural relaxation made simple, Phys- ical Review Letters97, 1 (2006)

    2006

  69. [77]

    A. Wang, R. Kingsbury, M. McDermott, M. Horton, A. Jain, S. P. Ong, S. Dwaraknath, and K. A. Persson, A framework for quantifying uncertainty in DFT energy corrections, Scientific Reports11, 15496 (2021)

    2021

  70. [78]

    Kresse and J

    G. Kresse and J. Furthm¨ uller, Efficiency of ab-initio total energy calculations for metals and semiconductors using a plane-wave basis set, Computational Materials Science 6, 15 (1996)

    1996

  71. [79]

    Kresse and D

    G. Kresse and D. Joubert, From ultrasoft pseudopoten- tials to the projector augmented-wave method, Physical Review B59, 1758 (1999)

    1999

  72. [80]

    J. Sun, A. Ruzsinszky, and J. P. Perdew, Strongly Con- strained and Appropriately Normed Semilocal Density Functional, Physical Review Letters115, 036402 (2015)

    2015

  73. [81]

    J. W. Furness, A. D. Kaplan, J. Ning, J. P. Perdew, and J. Sun, Accurate and Numerically Efficient r 2SCAN Meta-Generalized Gradient Approximation, The Jour- nal of Physical Chemistry Letters11, 8208 (2020), arXiv:2008.03374

    2020 arXiv

  74. [82]

    Kingsbury, A

    R. Kingsbury, A. S. Gupta, C. J. Bartel, J. M. Munro, S. Dwaraknath, M. Horton, and K. A. Persson, Perfor- mance comparison of r2SCAN and SCAN metaGGA den- sity functionals for solid materials via an automated, high-throughput computational workflow, Physical Re- view Material...

    2022

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.