Pith. sign in

REVIEW 2 major objections 5 minor 1 cited by

DiffCrysGen: A Score-Based Diffusion Model for Design of Diverse Inorganic Crystalline Materials

T0 review · 2 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read A single diffusion model yields rare-earth-free magnet candidates.

desk verdict A real generative-materials pipeline whose own FM/AFM recheck undercuts its headline success rate; worth a serious referee, but the property-design claim needs to be recomputed. read the letter →

arxiv 2505.07442 v1 pith:BXXL4S6L submitted 2025-05-12 cond-mat.mtrl-sci physics.comp-ph

Sourav Mal , Subhankar Mishra , Prasenjit Sen This is my paper · ORCID
classification cond-mat.mtrl-sciphysics.comp-ph PACS 81.05.Zx75.50.Ww
keywords score-baseddiffusionmodelcrystalstructuregenerationinvertiblereal-spacecrystallographicrepresentationrare-earth-freepermanentmagnetsmagnetocrystallineanisotropymaterialsdiscoverydensityfunctionaltheoryvalidationinversedesign
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

DiffCrysGen claims that one score-based diffusion model, trained on a unified 2D matrix representation of crystal structures, can learn the joint distribution of atom types, fractional coordinates, and lattice parameters directly from data, without hand-crafted priors or decoupled modules for each structural component. The authors demonstrate this by generating over a million candidate inorganic materials, filtering for rare-earth-free compositions, and using property predictors plus density-functional theory to identify 97 materials meeting both formation-energy and magnetization targets. Among the dynamically stable survivors, five have ferromagnetic ground states, and two combine high saturation magnetization with strong magnetocrystalline anisotropy. If the claim holds, generative materials design need not encode crystal symmetry or chemistry by hand; the model learns those rules from the data distribution itself.

What carries the argument

The central object is IRCR, a set of five 2D matrices—element one-hot encoding, lattice constants and angles, fractional coordinates, site occupancy, and elemental properties—that encode the unit cell invertibly, so generated matrices can be decoded back into structures. The score-based diffusion model uses a variance-exploding SDE: noise is added linearly in time without scaling the data, and a noise-conditional denoiser $D_\theta$ trained with an $L_2$ loss estimates the score $\nabla_x \log p_t(x)$ through $(D_\theta(x_t, \sigma) - x_t)/\sigma^2$. This single network replaces the separate atom, lattice, and coordinate channels of prior models, and the same IRCR input feeds convolutional predictors for formation energy and saturation magnetization that screen the generated pool before costly DFT calculations.

What would settle it

Recompute the FM-versus-AFM energy difference for all 107 DFT-relaxed materials that passed the $M_s \geq 1$ T screen, not just the final 13; if the majority of those turn out to be antiferromagnetic, the claimed 80.16% success rate for magnetic design collapses to a much smaller ferromagnetic subset.

Watch

Extended reading notes

Core claim

The core claim is that an expressive denoising network can implicitly capture crystallographic priors—symmetry, chemical validity, and the interdependence of composition, positions, and lattice—when trained on a sufficiently large dataset encoded as Invertible Real-Space Crystallographic Representation (IRCR) 2D matrices. In a variance-exploding score-based framework, the model diffuses the full matrix representation to noise and learns to reverse the process by predicting clean data at each noise level. The authors report that 28.19% of generated rare-earth-free materials fall in high-symmetry space groups, a sharp contrast to the comparative VA E models, and that 97 of 121 DFT-relaxable candidates meet the design targets ($h_{\mathrm{form}} \leq -0.2$ eV/atom and $M_s \geq 1$ T). After phonon and magnetic-order checks, 13 materials are dynamically stable; among the high-anisotropy ones, LiFeO and ScFe$_4$O$_5$ exhibit ferromagnetic ground states with $K_1$ of 4.20 and 1.91 MJ/m$^3$. This, the paper argues, shows that symmetry and chemistry can be learned rather than imposed.

Load-bearing premise

The pipeline treats the saturation-magnetization labels and the trained $M_s$ predictor as describing ferromagnetic ground states; if most screened candidates are actually antiferromagnetic or nonmagnetic, the reported permanent-magnet success rate is considerably overestimated.

Editorial extensions

If this is right

  • If the joint-distribution claim holds, generation quality should improve further with dataset size; the current model still under-covers high-symmetry structures relative to the training distribution (71.8% monoclinic/triclinic outputs versus 17.97% in training).
  • The unconditional base model opens a direct route to property-conditioned generation through adapter modules or classifier-free guidance, steering outputs toward target compositions, space groups, or properties.
  • The pipeline also surfaces strongly stabilized antiferromagnetic compounds such as Mn2AlRh and Mn4BePd5, which are irrelevant for permanent magnets but useful for spintronics.
  • Because the IRCR representation is invertible and flexible, extending it beyond ternary compositions would let the same framework generate quaternary or higher-order materials without a new architectural design.

Reading between the lines

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

  • A natural test of the data-driven claim is to train the same architecture on an unfiltered dataset without the $M_s \geq 10^{-5}$ T magnetization cut; if the model's magnetic bias disappears, the filter itself may be driving candidate chemistry rather than the score model learning magnetism.
  • Re-labelling the training set with AFM-aware ground-state magnetization could turn the same pipeline into a generator for ferrimagnets or altermagnets, directly expanding the target property space beyond ferromagnets.
  • The near-degenerate FM/AFM state of LiFe2O2 (0.25 meV/atom) suggests the generative model can place compounds at magnetic phase boundaries; guided diffusion might systematically discover other tunable magnetic phases.
  • One could benchmark DiffCrysGen against symmetry-aware generative models on the same 2D IRCR input, isolating whether the observed symmetry gains come from the diffusion objective or from the representation itself.
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

2 major / 5 minor

Summary. The paper introduces DiffCrysGen, a score-based diffusion model (VE SDE) trained on IRCR 2D matrices of crystalline unit cells from the Alexandria database. The model jointly generates element, coordinate, occupancy, and lattice matrices through a single denoiser, without explicit symmetry-aware priors. The authors train property predictors for formation energy and saturation magnetization using the same representation, generate about 1.26 million candidates, filter for rare-earth-free compositions and predicted targets (hform <= -0.2 eV/atom, Ms >= 1 T, dmin >= 1 Å, ternary-only, space group > 16), and submit 140 candidates to DFT relaxation, convex-hull distance, SOC-based K1, phonons, and FM/AFM energy comparisons. They report a validity rate of 86.42%, a success rate of 80.16% among structurally valid materials and 65% overall, and identify 13 dynamically stable materials, five of which have K1 >= 1 MJ/m3. A subsequent FM/AFM analysis shows that 8 of the 13 are antiferromagnetic, leaving only LiFeO and ScFe4O5 as FM high-anisotropy candidates.

Significance. The central methodological claim is attractive: a single score network on a unified 2D representation can learn the joint distribution of composition, coordinates, and lattice without decoupled modules or explicit symmetry priors. The paper includes a direct VAE comparison on the same dataset, showing a much improved space-group distribution. The DFT validation workflow is unusually thorough for a generative-model paper, including relaxation, hull distance, phonon dispersions, SOC-based K1, and a genuinely external FM/AFM check. However, the headline success rates and the permanent-magnet candidate list are computed under an FM-only assumption that the paper itself partially retracts in Section III.E, and the method section omits the discrete decoding step. With those points corrected, the work would be a credible contribution to generative materials discovery.

major comments (2)
  1. [III.D, III.E; Tables II and III] The headline success rates (80.16% among structurally valid materials, 65% overall) and the list of materials with K1 >= 1 MJ/m3 are computed under an FM-only magnetization assumption. Section III.E explicitly concedes this assumption, and Table III shows that 8 of the 13 dynamically stable candidates have AFM ground states, including LiFe2O2 and KFe2O2 with K1 > 4 MJ/m3 in Table II. The FM/AFM comparison is performed for only 13 of the 54 materials on which K1 was computed, so the extent of AFM contamination in the 97-material and 54-material sets is unknown. Consequently, the claimed design success rate and the permanent-magnet candidate list are not reliable as stated; the authors should recompute success under a consistent FM+Ms+K1 target or expand the FM/AFM check to all materials that feed the success statistics.
  2. [II.B (Eqs. 5, 8, 9)] The generative model is defined as a continuous VE SDE on IRCR matrices, but the element matrix E and occupancy matrix O are one-hot categorical. The paper does not describe how the continuous denoiser output is converted back to a valid discrete crystal (e.g., rounding, masking, or a separate validity check). Without this step, the procedure is not reproducible and the claim that atom types are jointly generated without task-specific priors is incomplete, since any fixed decoding rule is itself a prior. Please specify the decoding procedure in detail in the Methods.
minor comments (5)
  1. [Table II] The table title states 'final 14 materials' but the table contains 13 rows; the text also says 13 dynamically stable materials. Please correct the numbering.
  2. [III.D] The units for K1 are inconsistent: '17 materials demonstrated significant anisotropy (K1≥ 0.5 MJ/m)' should read 'MJ/m3'; please check all instances and figure captions.
  3. [III.C] The definition of 'novel compositions absent from the training set' (959,122 of 1,264,466) is not given; specify whether novelty is determined by composition only, and how the comparison is performed.
  4. [III.B] The comparison with graph neural networks trained on the entire Alexandria database does not name the specific model; add a reference or a brief description for context.
  5. [Fig. 2] The caption contains a typo: 'Learing curve' should be 'Learning curve'; please fix.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: the central generation claim is validated by external DFT, phonon, and FM/AFM calculations, and the FM-only screening assumption is a stated correctness limitation, not a definitional reduction.

full rationale

The paper's core derivation chain is not circular. DiffCrysGen is a standard score-based diffusion model trained on an invertible representation (IRCR) drawn from the Alexandria database; the representation is an input encoding, not the target result. The generated candidates are screened by IRCR-based property predictors, but the headline success rates, stability checks, and property claims are then re-evaluated by external DFT: structure relaxation, formation energy, saturation magnetization, convex-hull energy, K1 with spin-orbit coupling, and phonon dispersion. These external calculations break any fitted-input loop, so the 'prediction' of stable, high-magnetization candidates is not equivalent by construction to the training labels. The most important caveat is stated in Section III.E: 'So far it has been implicitly assumed that all the magnetic materials have ferromagnetic ground states,' and the later FM/AFM comparison shows 8 of 13 checked survivors are antiferromagnetic. This is an explicit limitation that weakens the quantitative success claim and the permanent-magnet candidate list, but it is a physical-assumption error or overstatement, not a circular reduction: the FM/AFM energies are computed from first principles and are external to the generative model. The self-citations to the authors' earlier IRCR work (ref 33) and earlier statistical anisotropy analysis (ref 50) are used as representation tools and screening heuristics, respectively, not as unverified uniqueness theorems or ansatze that predetermine the final DFT-validated outcomes. No fitted parameter is renamed as a prediction, and no equation in the paper reduces the derived result to the model's own inputs. Therefore no significant circularity is present, and the honest score is 0.

Assumptions & free parameters 9 free parameters · 4 assumptions · 0 invented entities

The central results rest on domain assumptions about the matrix representation, the ferromagnetic meaning of the magnetization labels, and the out-of-distribution generalization of fitted predictors. The dataset and screening thresholds are hand-chosen and would require sensitivity analysis to establish that the generative model, not the filters, drives the reported success.

free parameters (9)
  • Dataset stability cutoff E_hull = 0.1 eV/atom
    Hand-chosen convex-hull threshold in Section III.A that restricts training data to stable and metastable materials; this bakes stability into the generated distribution.
  • Dataset formation-energy cutoff h_form = 0 eV/atom
    Hand-chosen threshold in Section III.A that removes unstable materials from Alex-1, shaping the learned chemistry.
  • Dataset magnetization cutoff M_s = 1e-5 T
    Hand-chosen threshold in Section III.A that keeps only magnetic materials in Alex-1; the model never sees non-magnetic compositions.
  • Candidate screening M_s target = 1 T
    Hand-chosen target in Section III.C for selecting property-aligned candidates before DFT.
  • Candidate screening h_form target = -0.2 eV/atom
    Hand-chosen target in Section III.C for the formation-energy filter.
  • Minimum interatomic distance d_min = 1 A
    Heuristic filter in Section III.C to remove unphysical close contacts; no sensitivity analysis is provided.
  • Space-group filter = >16
    Hand-chosen cutoff in Section III.C to keep only high-symmetry structures; this directly defines the 140-candidate DFT set.
  • Composition filter = ternary-only
    Restricts the final DFT set to 635 ternary compounds in Section III.C; justified by diversity but arbitrary.
  • Diffusion and denoiser hyperparameters
    Noise schedule, sigma_max, network depth and width, batch size, and training steps are not reported in Section II.B; these are hand-chosen and needed to reproduce the model.
assumptions (4)
  • domain assumption IRCR is an invertible, lossless encoding of unit-cell lattice, composition, and fractional coordinates for the supported materials.
    Invoked in Section II.A and inherited from ref 33; if decoding noisy matrices fails or loses symmetry information, generated matrices may not correspond to valid crystals.
  • domain assumption Alexandria saturation magnetization values correspond to ferromagnetic spin arrangements.
    Sections III.A and III.E: the Ms predictor was trained on FM-only magnetization; this is stated as an implicit assumption and is later shown to fail for 8 of 13 finalists.
  • domain assumption The trained IRCR-based CNN property predictors generalize to generated out-of-distribution crystals.
    Section III.B-C uses the predictors to prune millions of generations to 862 candidates; no domain-shift or uncertainty validation is reported.
  • domain assumption DFT relaxation, E_hull <= 0.4 eV/atom, phonon stability, and the E = K1 sin^2(theta) extraction define viable permanent-magnet candidates.
    Section III.D and Appendix A; these computational criteria are standard but do not guarantee synthesizability or address finite-temperature magnetism.

how reviews work

0 comments
Cite this review

Pith. "Pith review of DiffCrysGen: A Score-Based Diffusion Model for Design of Diverse Inorganic Crystalline Materials." pith.science (2026). https://pith.science/paper/BXXL4S6L

@misc{pith2026250507442,
  author       = {Pith},
  title        = {Pith review of: DiffCrysGen: A Score-Based Diffusion Model for Design of Diverse Inorganic Crystalline Materials},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/BXXL4S6L}},
  note         = {Machine review of arXiv:2505.07442}
}
read the original abstract

Crystal structure generation is a foundational challenge in materials discovery, particularly in designing functional inorganic crystalline materials with desired properties. Most existing diffusion-based generative models for crystals rely on complex, hand-crafted priors and modular architectures to separately model atom types, atomic positions, and lattice parameters. These methods often require customized diffusion processes and conditional denoising, which can introduce additional model complexities and inconsistencies. Here we introduce DiffCrysGen, a fully data-driven, score-based diffusion model that jointly learns the distribution of all structural components in crystalline materials. With crystal structure representation as unified 2D matrices, DiffCrysGen bypasses the need for task-specific priors or decoupled modules, enabling end-to-end generation of atom types, fractional coordinates, and lattice parameters within a single framework. Our model learns crystallographic symmetry and chemical validity directly from large-scale datasets, allowing it to scale to complex materials discovery tasks. As a demonstration, we applied DiffCrysGen to the design of rare-earth-free magnetic materials with high saturation magnetization, showing its effectiveness in generating stable, diverse, and property-aligned candidates for sustainable magnet applications.

Figures

Figures reproduced from arXiv: 2505.07442 by the authors.

Figure 1
Figure 1. FIG. 1. Distribution of (a) number of atoms in unit cell, (b) space groups, (c) convex hull, (d) formation energy and (e) saturation magnetization [PITH_FULL_IMAGE:figures/full_fig_p005_1.png] view at source ↗
Figure 2
Figure 2. FIG. 2. (a) Learing curve of the diffusion model, (b)-(c) Parity plots of [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. FIG. 3. Visualization of the crystal structures of the top eight materials that are closest to the convex hull ( [PITH_FULL_IMAGE:figures/full_fig_p010_3.png] view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: FIG. 4. Visualization of the crystal structures of the bottom five materials, which exhibit high saturation magnetization ( [PITH_FULL_IMAGE:figures/full_fig_p010_4.png]
Figure 5
Figure 5. Figure 5: FIG. 5. Phonon dispersions of the top eight materials that are closest to the convex hull ( [PITH_FULL_IMAGE:figures/full_fig_p011_5.png]
Figure 6
Figure 6. Figure 6: FIG. 6. Phonon dispersions of the bottom five materials, which exhibit high saturation magnetization ( [PITH_FULL_IMAGE:figures/full_fig_p012_6.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. VASP Plugins: Linking the Vienna ab-initio Simulation Package with Python

    cond-mat.mtrl-sci 2026-07 accept novelty 5.5 of 10

    A C++/pybind11 shared-memory plugin layer exposes VASP SCF and ionic data as NumPy arrays so Python can modify structure, forces, local potential, and occupancies in place.

Reference graph

Works this paper leans on

62 extracted references · 9 canonical work pages · cited by 1 Pith paper

  1. [1]

    Jain , author S

    author author A. Jain , author S. P. \ Ong , author G. Hautier , author W. Chen , author W. D. \ Richards , author S. Dacek , author S. Cholia , author D. Gunter , author D. Skinner , author G. Ceder , \ and\ author K. A. \ Persson ,\ 10.1063/1.4812323 journal journal APL Mater. \ volume 1 ,\ pages 011002 ( year 2013 ) ,\ http://arxiv.org/abs/https://pubs...

    doi:10.1063/1.4812323

  2. [2]

    author author J. E. \ Saal , author S. Kirklin , author M. Aykol , author B. Meredig , \ and\ author C. Wolverton ,\ 10.1007/s11837-013-0755-4 journal journal JOM \ volume 65 ,\ pages 1501 ( year 2013 ) NoStop

    doi:10.1007/s11837-013-0755-4 2013

  3. [3]

    Choudhary , author K

    author author K. Choudhary , author K. F. \ Garrity , author A. C. E. \ Reid , author B. DeCost , author A. J. \ Biacchi , author A. R. \ Hight Walker , author Z. Trautt , author J. Hattrick-Simpers , author A. G. \ Kusne , author A. Centrone , author A. Davydov , author J. Jiang , author R. Pachter , author G. Cheon , author E. Reed , author A. Agrawal ,...

  4. [4]

    Curtarolo , author W

    author author S. Curtarolo , author W. Setyawan , author G. L. \ Hart , author M. Jahnatek , author R. V. \ Chepulskii , author R. H. \ Taylor , author S. Wang , author J. Xue , author K. Yang , author O. Levy , author M. J. \ Mehl , author H. T. \ Stokes , author D. O. \ Demchenko , \ and\ author D. Morgan ,\ https://doi.org/10.1016/j.commatsci.2012.02.0...

    doi:10.1016/j.commatsci.2012.02.005

  5. [5]

    Schmidt , author H.-C

    author author J. Schmidt , author H.-C. \ Wang , author T. F. T. \ Cerqueira , author S. Botti , \ and\ author M. A. L. \ Marques ,\ 10.1038/s41597-022-01177-w journal journal Scientific Data \ volume 9 ,\ pages 64 ( year 2022 ) NoStop

    doi:10.1038/s41597-022-01177-w

  6. [6]

    Schmidt , author L

    author author J. Schmidt , author L. Pettersson , author C. Verdozzi , author S. Botti , \ and\ author M. A. L. \ Marques ,\ 10.1126/sciadv.abi7948 journal journal Science Advances \ volume 7 ,\ pages eabi7948 ( year 2021 ) ,\ http://arxiv.org/abs/https://www.science.org/doi/pdf/10.1126/sciadv.abi7948 https://www.science.org/doi/pdf/10.1126/sciadv.abi7948 NoStop

    doi:10.1126/sciadv.abi7948 2021

  7. [7]

    Schmidt , author N

    author author J. Schmidt , author N. Hoffmann , author H.-C. \ Wang , author P. Borlido , author P. J. M. A. \ Carriço , author T. F. T. \ Cerqueira , author S. Botti , \ and\ author M. A. L. \ Marques ,\ https://doi.org/10.1002/adma.202210788 journal journal Advanced Materials \ volume 35 ,\ pages 2210788 ( year 2023 ) ,\ http://arxiv.org/abs/https://adv...

    doi:10.1002/adma.202210788

  8. [8]

    Xie \ and\ author J

    author author T. Xie \ and\ author J. C. \ Grossman ,\ 10.1103/PhysRevLett.120.145301 journal journal Phys. Rev. Lett. \ volume 120 ,\ pages 145301 ( year 2018 ) NoStop

    doi:10.1103/physrevlett.120.145301 2018

Show all 62 references
  1. [9]

    Chen , author W

    author author C. Chen , author W. Ye , author Y. Zuo , author C. Zheng , \ and\ author S. P. \ Ong ,\ 10.1021/acs.chemmater.9b01294 journal journal Chem. Mater. \ volume 31 ,\ pages 3564 ( year 2019 ) NoStop

    2019 doi

  2. [10]

    Choudhary \ and\ author B

    author author K. Choudhary \ and\ author B. DeCost ,\ 10.1038/s41524-021-00650-1 journal journal npj Computational Materials \ volume 7 ,\ pages 185 ( year 2021 ) NoStop

    2021 doi

  3. [11]

    Ramesh , author M

    author author A. Ramesh , author M. Pavlov , author G. Goh , author S. Gray , author C. Voss , author A. Radford , author M. Chen , \ and\ author I. Sutskever ,\ https://arxiv.org/abs/2102.12092 title Zero-shot text-to-image generation , \ ( year 2021 ),\ http://arxiv.org/abs/...

    2021 arXiv

  4. [12]

    Yu , author Y

    author author J. Yu , author Y. Xu , author J. Y. \ Koh , author T. Luong , author G. Baid , author Z. Wang , author V. Vasudevan , author A. Ku , author Y. Yang , author B. K. \ Ayan , author B. Hutchinson , author W. Han , author Z. Parekh , author X. Li , author H. Zhang , ...

    2022 arXiv

  5. [13]

    Anil \ and\ author et al

    author author R. Anil \ and\ author et al. ,\ https://arxiv.org/abs/2305.10403 title Palm 2 technical report , \ ( year 2023 ),\ http://arxiv.org/abs/2305.10403 arXiv:2305.10403 [cs.CL] NoStop

    2023 arXiv

  6. [14]

    author author OpenAI ,\ https://arxiv.org/abs/2303.08774 title Gpt-4 technical report , \ ( year 2024 ),\ http://arxiv.org/abs/2303.08774 arXiv:2303.08774 [cs.CL] NoStop

    2024 arXiv

  7. [15]

    Ho , author W

    author author J. Ho , author W. Chan , author C. Saharia , author J. Whang , author R. Gao , author A. Gritsenko , author D. P. \ Kingma , author B. Poole , author M. Norouzi , author D. J. \ Fleet , \ and\ author T. Salimans ,\ https://arxiv.org/abs/2210.02303 title Imagen vi...

    arXiv

  8. [16]

    Singer , author A

    author author U. Singer , author A. Polyak , author T. Hayes , author X. Yin , author J. An , author S. Zhang , author Q. Hu , author H. Yang , author O. Ashual , author O. Gafni , author D. Parikh , author S. Gupta , \ and\ author Y. Taigman ,\ https://arxiv.org/abs/2209.1479...

    2022 arXiv

  9. [17]

    author author D. P. \ Kingma \ and\ author M. Welling ,\ @noop title Auto-encoding variational bayes , \ ( year 2022 ),\ http://arxiv.org/abs/1312.6114 arXiv:1312.6114 [stat.ML] NoStop

    2022 arXiv

  10. [18]

    author author D. P. \ Kingma \ and\ author M. Welling ,\ 10.1561/2200000056 journal journal Foundations and Trends in Machine Learning \ volume 12 ,\ pages 307 ( year 2019 ) NoStop

    2019 doi

  11. [19]

    Noh , author J

    author author J. Noh , author J. Kim , author H. S. \ Stein , author B. Sanchez-Lengeling , author J. M. \ Gregoire , author A. Aspuru-Guzik , \ and\ author Y. Jung ,\ https://doi.org/10.1016/j.matt.2019.08.017 journal journal Matter \ volume 1 ,\ pages 1370 ( year 2019 ) NoStop

    2019 doi

  12. [20]

    Ren , author S

    author author Z. Ren , author S. I. P. \ Tian , author J. Noh , author F. Oviedo , author G. Xing , author J. Li , author Q. Liang , author R. Zhu , author A. G. \ Aberle , author S. Sun , author X. Wang , author Y. Liu , author Q. Li , author S. Jayavelu , author K. Hippalgao...

    doi

  13. [21]

    Xie , author X

    author author T. Xie , author X. Fu , author O. Ganea , author R. Barzilay , \ and\ author T. S. \ Jaakkola ,\ https://arxiv.org/abs/2110.06197 journal journal CoRR \ volume abs/2110.06197 ( year 2021 ) ,\ http://arxiv.org/abs/2110.06197 2110.06197 NoStop

    2021 arXiv

  14. [22]

    author author C. J. \ Court , author B. Yildirim , author A. Jain , \ and\ author J. M. \ Cole ,\ 10.1021/acs.jcim.0c00464 journal journal J. Chem. Inf. Model. \ volume 60 ,\ pages 4518 ( year 2020 ) ,\ note pMID: 32866381 ,\ http://arxiv.org/abs/https://doi.org/10.1021/acs.jc...

    2020 doi

  15. [23]

    author author C. J. \ Court , author A. Jain , \ and\ author J. M. \ Cole ,\ 10.1021/acs.chemmater.1c01368 journal journal Chem. Mater. \ volume 33 ,\ pages 7217 ( year 2021 ) ,\ http://arxiv.org/abs/https://doi.org/10.1021/acs.chemmater.1c01368 https://doi.org/10.1021/acs.che...

    2021 doi

  16. [24]

    author author I. J. \ Goodfellow , author J. Pouget-Abadie , author M. Mirza , author B. Xu , author D. Warde-Farley , author S. Ozair , author A. Courville , \ and\ author Y. Bengio ,\ @noop title Generative adversarial networks , \ ( year 2014 ),\ http://arxiv.org/abs/1406.2...

    2014 arXiv

  17. [25]

    Long , author N

    author author T. Long , author N. M. \ Fortunato , author I. Opahle , author Y. Zhang , author I. Samathrakis , author C. Shen , author O. Gutfleisch , \ and\ author H. Zhang ,\ 10.1038/s41524-021-00526-4 journal journal Npj Comput. Mater. \ volume 7 ,\ pages 66 ( year 2021 ) NoStop

    doi

  18. [26]

    Zhao , author M

    author author Y. Zhao , author M. Al-Fahdi , author M. Hu , author E. M. D. \ Siriwardane , author Y. Song , author A. Nasiri , \ and\ author J. Hu ,\ https://doi.org/10.1002/advs.202100566 journal journal Adv. Sci. \ volume 8 ,\ pages 2100566 ( year 2021 ) ,\ http://arxiv.org...

    doi

  19. [27]

    Kim , author S

    author author B. Kim , author S. Lee , \ and\ author J. Kim ,\ 10.1126/sciadv.aax9324 journal journal Sci. Adv. \ volume 6 ,\ pages eaax9324 ( year 2020 a ) ,\ http://arxiv.org/abs/https://www.science.org/doi/pdf/10.1126/sciadv.aax9324 https://www.science.org/doi/pdf/10.1126/s...

    2020 doi

  20. [28]

    Nouira , author N

    author author A. Nouira , author N. Sokolovska , \ and\ author J.-C. \ Crivello ,\ @noop title Crystalgan: learning to discover crystallographic structures with generative adversarial networks , \ ( year 2019 ),\ http://arxiv.org/abs/1810.11203 arXiv:1810.11203 [cs.LG] NoStop

    2019 arXiv

  21. [29]

    Kim , author J

    author author S. Kim , author J. Noh , author G. H. \ Gu , author A. Aspuru-Guzik , \ and\ author Y. Jung ,\ 10.1021/acscentsci.0c00426 journal journal ACS Cent. Sci. \ volume 6 ,\ pages 1412 ( year 2020 b ) ,\ note pMID: 32875082 ,\ http://arxiv.org/abs/https://doi.org/10.102...

    2020 doi

  22. [30]

    Choubisa , author M

    author author H. Choubisa , author M. Askerka , author K. Ryczko , author O. Voznyy , author K. Mills , author I. Tamblyn , \ and\ author E. H. \ Sargent ,\ https://doi.org/10.1016/j.matt.2020.04.016 journal journal Matter \ volume 3 ,\ pages 433 ( year 2020 ) NoStop

    2020 doi

  23. [31]

    Wines , author T

    author author D. Wines , author T. Xie , \ and\ author K. Choudhary ,\ 10.1021/acs.jpclett.3c01260 journal journal J. Phys. Chem. Lett. \ volume 14 ,\ pages 6630 ( year 2023 ) ,\ note pMID: 37462366 ,\ http://arxiv.org/abs/https://doi.org/10.1021/acs.jpclett.3c01260 https://do...

    2023 doi

  24. [32]

    Lyngby \ and\ author K

    author author P. Lyngby \ and\ author K. S. \ Thygesen ,\ 10.1038/s41524-022-00923-3 journal journal Npj Comput. Mater. \ volume 8 ,\ pages 232 ( year 2022 ) NoStop

    2022 doi

  25. [33]

    Mal , author G

    author author S. Mal , author G. Seal , \ and\ author P. Sen ,\ 10.1021/acs.jpclett.4c00068 journal journal The Journal of Physical Chemistry Letters \ volume 15 ,\ pages 3221 ( year 2024 ) NoStop

    2024 doi

  26. [34]

    Lucas , author G

    author author J. Lucas , author G. Tucker , author R. Grosse , \ and\ author M. Norouzi ,\ https://openreview.net/forum?id=r1xaVLUYuE title Understanding posterior collapse in generative latent variable models , \ ( year 2019 ) NoStop

    2019

  27. [35]

    Srivastava , author L

    author author A. Srivastava , author L. Valkov , author C. Russell , author M. U. \ Gutmann , \ and\ author C. Sutton ,\ https://arxiv.org/abs/1705.07761 title Veegan: Reducing mode collapse in gans using implicit variational learning , \ ( year 2017 ),\ http://arxiv.org/abs/1...

    2017 arXiv

  28. [36]

    Dhariwal \ and\ author A

    author author P. Dhariwal \ and\ author A. Nichol ,\ https://arxiv.org/abs/2105.05233 title Diffusion models beat gans on image synthesis , \ ( year 2021 ),\ http://arxiv.org/abs/2105.05233 arXiv:2105.05233 [cs.LG] NoStop

    2021 arXiv

  29. [37]

    Sohl-Dickstein , author E

    author author J. Sohl-Dickstein , author E. A. \ Weiss , author N. Maheswaranathan , \ and\ author S. Ganguli ,\ https://arxiv.org/abs/1503.03585 title Deep unsupervised learning using nonequilibrium thermodynamics , \ ( year 2015 ),\ http://arxiv.org/abs/1503.03585 arXiv:1503...

    2015 arXiv

  30. [38]

    Ho , author A

    author author J. Ho , author A. Jain , \ and\ author P. Abbeel ,\ https://arxiv.org/abs/2006.11239 title Denoising diffusion probabilistic models , \ ( year 2020 ),\ http://arxiv.org/abs/2006.11239 arXiv:2006.11239 [cs.LG] NoStop

    2006 arXiv

  31. [39]

    Song , author J

    author author Y. Song , author J. Sohl-Dickstein , author D. P. \ Kingma , author A. Kumar , author S. Ermon , \ and\ author B. Poole ,\ https://arxiv.org/abs/2011.13456 title Score-based generative modeling through stochastic differential equations , \ ( year 2021 ),\ http://...

    2011 arXiv

  32. [40]

    Karras , author M

    author author T. Karras , author M. Aittala , author T. Aila , \ and\ author S. Laine ,\ https://arxiv.org/abs/2206.00364 title Elucidating the design space of diffusion-based generative models , \ ( year 2022 ),\ http://arxiv.org/abs/2206.00364 arXiv:2206.00364 [cs.CV] NoStop

    2022 arXiv

  33. [41]

    Zeni , author R

    author author C. Zeni , author R. Pinsler , author D. Z \"u gner , author A. Fowler , author M. Horton , author X. Fu , author Z. Wang , author A. Shysheya , author J. Crabb \'e , author S. Ueda , author R. Sordillo , author L. Sun , author J. Smith , author B. Nguyen , author...

    doi

  34. [42]

    Jiao , author W

    author author R. Jiao , author W. Huang , author P. Lin , author J. Han , author P. Chen , author Y. Lu , \ and\ author Y. Liu ,\ https://arxiv.org/abs/2309.04475 title Crystal structure prediction by joint equivariant diffusion , \ ( year 2024 a ),\ http://arxiv.org/abs/2309....

    2024 arXiv

  35. [43]

    Jiao , author W

    author author R. Jiao , author W. Huang , author Y. Liu , author D. Zhao , \ and\ author Y. Liu ,\ https://arxiv.org/abs/2402.03992 title Space group constrained crystal generation , \ ( year 2024 b ),\ http://arxiv.org/abs/2402.03992 arXiv:2402.03992 [cs.LG] NoStop

    2024 arXiv

  36. [44]

    author author F. E. \ Kelvinius , author O. B. \ Andersson , author A. S. \ Parackal , author D. Qian , author R. Armiento , \ and\ author F. Lindsten ,\ https://arxiv.org/abs/2502.06485 title Wyckoffdiff -- a generative diffusion model for crystal symmetry , \ ( year 2025 ),\...

    2025

  37. [45]

    Park , author A

    author author H. Park , author A. Onwuli , \ and\ author A. Walsh ,\ 10.26434/chemrxiv-2024-rw8p5 journal journal ChemRxiv \ ( year 2024 ),\ 10.26434/chemrxiv-2024-rw8p5 NoStop

    2024 doi

  38. [46]

    Luo , author C

    author author Y. Luo , author C. Liu , \ and\ author S. Ji ,\ https://arxiv.org/abs/2307.02707 title Towards symmetry-aware generation of periodic materials , \ ( year 2023 ),\ http://arxiv.org/abs/2307.02707 arXiv:2307.02707 [cs.LG] NoStop

    2023 arXiv

  39. [47]

    O'Shea \ and\ author R

    author author K. O'Shea \ and\ author R. Nash ,\ @noop title An introduction to convolutional neural networks , \ ( year 2015 ),\ http://arxiv.org/abs/1511.08458 arXiv:1511.08458 [cs.NE] NoStop

    2015 arXiv

  40. [48]

    Schmidt , author T

    author author J. Schmidt , author T. F. \ Cerqueira , author A. H. \ Romero , author A. Loew , author F. Jäger , author H.-C. \ Wang , author S. Botti , \ and\ author M. A. \ Marques ,\ https://doi.org/10.1016/j.mtphys.2024.101560 journal journal Materials Today Physics \ volu...

    2024

  41. [49]

    Zhao , author E

    author author Y. Zhao , author E. M. D. \ Siriwardane , author Z. Wu , author N. Fu , author M. Al-Fahdi , author M. Hu , \ and\ author J. Hu ,\ 10.1038/s41524-023-00987-9 journal journal Npj Comput. Mater. \ volume 9 ,\ pages 38 ( year 2023 ) NoStop

    doi

  42. [50]

    Mal \ and\ author P

    author author S. Mal \ and\ author P. Sen ,\ https://doi.org/10.1016/j.jmmm.2023.171590 journal journal J. Magn. Magn. Mater. \ volume 589 ,\ pages 171590 ( year 2024 ) NoStop

    2023

  43. [51]

    Sakurai , author R

    author author M. Sakurai , author R. Wang , author T. Liao , author C. Zhang , author H. Sun , author Y. Sun , author H. Wang , author X. Zhao , author S. Wang , author B. Balasubramanian , author X. Xu , author D. J. \ Sellmyer , author V. Antropov , author J. Zhang , author ...

    doi

  44. [52]

    Nieves , author S

    author author P. Nieves , author S. Arapan , author J. Maudes-Raedo , author R. Marticorena-Sánchez , author N. Del Brío , author A. Kovacs , author C. Echevarria-Bonet , author D. Salazar , author J. Weischenberg , author H. Zhang , author O. Vekilova , author R. Serrano-Lópe...

    doi

  45. [53]

    Zhang , author A

    author author L. Zhang , author A. Rao , \ and\ author M. Agrawala ,\ https://arxiv.org/abs/2302.05543 title Adding conditional control to text-to-image diffusion models , \ ( year 2023 ),\ http://arxiv.org/abs/2302.05543 arXiv:2302.05543 [cs.CV] NoStop

    2023 arXiv

  46. [54]

    Ho \ and\ author T

    author author J. Ho \ and\ author T. Salimans ,\ https://arxiv.org/abs/2207.12598 title Classifier-free diffusion guidance , \ ( year 2022 ),\ http://arxiv.org/abs/2207.12598 arXiv:2207.12598 [cs.LG] NoStop

    2022 arXiv

  47. [55]

    Kresse \ and\ author J

    author author G. Kresse \ and\ author J. Furthmüller ,\ https://doi.org/10.1016/0927-0256(96)00008-0 journal journal Comput. Mater. Sci. \ volume 6 ,\ pages 15 ( year 1996 ) NoStop

    1996 doi

  48. [56]

    Kresse \ and\ author J

    author author G. Kresse \ and\ author J. Furthm\"uller ,\ 10.1103/PhysRevB.54.11169 journal journal Phys. Rev. B \ volume 54 ,\ pages 11169 ( year 1996 ) NoStop

    1996 doi

  49. [57]

    author author P. E. \ Bl\"ochl ,\ 10.1103/PhysRevB.50.17953 journal journal Phys. Rev. B \ volume 50 ,\ pages 17953 ( year 1994 ) NoStop

    1994 doi

  50. [58]

    author author J. P. \ Perdew , author K. Burke , \ and\ author M. Ernzerhof ,\ 10.1103/PhysRevLett.77.3865 journal journal Phys. Rev. Lett. \ volume 77 ,\ pages 3865 ( year 1996 ) NoStop

    1996 doi

  51. [59]

    author author S. L. \ Dudarev , author G. A. \ Botton , author S. Y. \ Savrasov , author C. J. \ Humphreys , \ and\ author A. P. \ Sutton ,\ 10.1103/PhysRevB.57.1505 journal journal Phys. Rev. B \ volume 57 ,\ pages 1505 ( year 1998 ) NoStop

    doi

  52. [60]

    Baroni , author S

    author author S. Baroni , author S. de Gironcoli , author A. Dal Corso , \ and\ author P. Giannozzi ,\ 10.1103/RevModPhys.73.515 journal journal Rev. Mod. Phys. \ volume 73 ,\ pages 515 ( year 2001 ) NoStop

    2001 doi

  53. [61]

    Togo , author L

    author author A. Togo , author L. Chaput , author T. Tadano , \ and\ author I. Tanaka ,\ 10.1088/1361-648X/acd831 journal journal J. Phys. Condens. Matter \ volume 35 ,\ pages 353001 ( year 2023 ) NoStop

    2023 doi

  54. [62]

    Togo ,\ 10.7566/JPSJ.92.012001 journal journal J

    author author A. Togo ,\ 10.7566/JPSJ.92.012001 journal journal J. Phys. Soc. Jpn. \ volume 92 ,\ pages 012001 ( year 2023 ) NoStop

    2023 doi

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.