Pith. sign in

REVIEW 4 major objections 5 minor 74 references

One-Shot Generative Design for Disordered Metamaterials via Self-Organizing Neural Cellular Automata

T0 review · 4 major / 5 minor · reviewed 2026-08-02 · deepseek-v4-flash

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

Pith's one-line read A single local growth rule, learned from one template, can generate and steer disordered microstructures, transferring across domains and discretizations.

desk verdict Convincing grid-based NCA for one-shot microstructure generation and steering, with a strong cloaking demo; the cross-discretization transfer claim is the soft underbelly and needs quantitative evidence before the paper's full generality can be accepted. read the letter →

arxiv 2607.14475 v1 pith:KUHWD5TR submitted 2026-07-16 cs.CE cs.LG

Yujie Xiang , Liwei Wang This is my paper
classification cs.CEcs.LG
keywords disorderedmetamaterialsneuralcellularautomatagenerativedesignone-shotlearningmicrostructuregenerationmultiscaleoptimizationmechanicalcloakingself-organization
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

Disordered metamaterials offer mechanical properties that ordered lattices cannot, but their design has been trapped between hand-crafted rules and data-hungry generative models. This paper claims that one local growth rule — a neural cellular automaton trained from a single template — can grow statistically equivalent disordered microstructures from random seeds. It further claims that the same frozen rule can be steered after training by editing only its fixed perception operators (rotating gradients, reshaping the metric, adjusting boundary speed), giving continuous control over orientation, anisotropy, scale, thickness, and stiffness without retraining or new data. Because the rule is local and interpreted as a generalized PDE, it transfers to irregular domains and arbitrary mesh discretizations, and it can be coupled to multiscale optimization to grow a smooth mechanical cloak that restores the homogeneous displacement field around a void.

What carries the argument

Key machinery: the neural cellular automaton (NCA) — each cell carries a multi-channel state; fixed Sobel/Laplacian kernels gather neighborhood information; a shared network maps the perception to a residual update; a Bernoulli mask breaks symmetry. The pivotal interpretation: the update is a forward-Euler step of a generalized PDE, so the network approximates Fθ(u, ∇u, ∇²u). Controls modify only the perception operators — rotation R(θ), Riemannian metric g(s1,s2)=exp([[s1,s2],[s2,-s1]]), anisotropic Laplacian scaled by μ, boundary speed s_x|∇y u|+s_y|∇x u| — while the network stays frozen. For meshes, grid kernels become angular weights from local-frame neighbor offsets, degenerating to gri

What would settle it

Train an NCA on a regular grid template; run the same frozen network on a moderately irregular triangular mesh using the paper's intrinsic stencil; then compare two-point correlation functions, lineal-path functions, and homogenized stiffness tensors between the grid-grown and mesh-grown samples. If these deviate by more than the sample-to-sample variation observed on the grid, the discretization-transfer claim is refuted.

Watch

Extended reading notes

Core claim

Central claim: a neural cellular automaton trained from a single template via a style loss is a generalized PDE — its local perception (state, gradient, Laplacian) mapped by a shared network to an update is a forward-Euler step of ∂u/∂t = Fθ(u, ∇u, ∇²u). Then the fixed perception operators are control handles: rotating gradients rotates the microstructure and stiffness; a Riemannian metric g(s1,s2) with scale μ controls anisotropy and feature size; a signed boundary speed dilates/erodes the solid phase directionally. The same PDE view justifies mesh transfer: replace grid kernels with intrinsic angular stencils in local frames, keep the same frozen weights. End-to-end, the authors optimize a

Load-bearing premise

The claim that the same trained NCA weights work on any mesh rests on the assertion in Section 3.2 that the NCA update approximates a continuous PDE whose solution is independent of domain and discretization; the paper states this without a proof, a convergence study, or quantitative comparison of mesh-grown versus grid-grown morphology and properties.

Editorial extensions

If this is right

  • A single trained NCA, one per template, reproduces six morphologically distinct disordered microstructures (leaf vein, vascular tissue, trabecular bone, biofilm, spider silk, metal fracture) without any dataset.
  • Post-training steering expands the accessible stiffness space continuously: the anisotropy ratio C11/C22 spans from 0.027 to 19.88 (nearly three orders of magnitude) with only four control parameters and no retraining.
  • The learned rule transfers to irregular domains and arbitrary mesh discretizations by adapting only the perception stencil, and its asynchronous growth supports multiple growth sources, parallelizable generation, and in-situ repair of a damaged bone microstructure.
  • Coupling the NCA with VAE-latent multiscale optimization yields a mechanical cloak: the same optimized control fields grow the structure at different resolutions, with mean relative displacement error falling from 26.12%±1.71% at 500×500 to 16.86%±0.63% at 1000×1000 across independent samples.

Reading between the lines

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

  • If the PDE-transfer claim holds, the NCA framework effectively decouples the learned growth rule from the computational mesh, so a single training run on a canonical grid could be reused across arbitrary geometries, manufacturing resolutions, and even point clouds — turning microstructure generation into a mesh-agnostic service.
  • The steering parameters (s1, s2, sx, sy) form an interpretable, low-dimensional parameterization of the learned dynamics; a natural extension the paper does not pursue is to invert the control-to-stiffness map (e.g., with a surrogate or normalizing flow) for on-demand generation of a target stiffness tensor, eliminating the need for the VAE-optimization round-trip.
  • The in-situ repair result suggests a broader principle: because the fixed surrounding tissue acts as a boundary condition, the same asynchronous-growth mechanism could be used for self-healing digital materials or for generating tissue-matching scaffolds around existing bone — a property that follows from the local, PDE-like dynamics rather than from any visual loss engineering.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. The paper proposes a Neural Cellular Automata (NCA) framework for generating disordered microstructures. A single NCA update rule is trained from one template using a frozen VGG-16 feature extractor and a sliced optimal transport style loss, then iteratively grows statistically equivalent microstructures. The authors introduce four post-training steering mechanisms — gradient rotation, Riemannian-metric anisotropy/scale, and directional boundary thickness — that modify only the fixed local perception operators, enabling control of orientation, anisotropy, scale, and thickness without retraining. They further claim that the learned rule transfers to irregular domains and arbitrary discretizations by redefining local perception operators on meshes, and demonstrate spatially varying growth, asynchronous growth, and in-situ repair. Finally, they couple the generator with a VAE-latent topology optimization pipeline and apply it to a mechanical cloaking problem, reporting a homogenized relative error of 8.10% and full-scale NCA-generated errors of 26.12%±1.71% at 500×500 and 16.86%±0.63% at 1000×1000 resolution.

Significance. If the claims hold, this is a strong contribution to generative design of disordered metamaterials. The grid-based generation and steering results are well supported: the one-shot training objective is sensible, the morphology is validated with two-point correlation and lineal-path functions, and the property-space exploration (10,000 samples; anisotropy ratio from 0.027 to 19.88) is extensive. The mechanical cloaking demonstration is quantified with full-scale FEA at two resolutions, which is more than most papers in this area provide. The central weakness is that the paper's headline generalizability claims — transfer across arbitrary discretizations and self-repair — rest on visual demonstrations and an unvalidated PDE analogy. The grid-based core of the paper is credible; the transfer claim needs substantially more evidence before the 'arbitrary discretization' contribution can be accepted.

major comments (4)
  1. [§3.2] The central claim that the learned NCA rule transfers across domains and discretizations rests on the assertion: 'the same PDE can be solved and converge to the same result regardless of the domain or discretization, as long as the local operators are adapted' (Sec. 3.2). No proof, convergence study, or quantitative comparison is provided. The only evidence is a single qualitative image (Fig. 15(b)); mesh-grown morphology is never compared with grid-grown morphology via S2, lineal-path, volume fraction, or homogenized stiffness, and no test at varying mesh irregularity is reported. Since 'arbitrary discretizations' is a headline contribution, this is a load-bearing gap. Please add quantitative mesh-vs-grid validation, or substantially weaken the claim.
  2. [Eq. (12), §3.2] The adapted stencils in Eq. (12) are not discretization-invariant as claimed. w_x and w_y scale as inverse L1 distance from the center vertex, with no area or edge-length factors, so on a nonuniform mesh they change the effective gradient magnitude and Laplacian normalization relative to the training grid. The statement that these weights 'degenerate exactly to the 2D grid kernels' is not correct even on a uniform grid: for an axis-aligned neighbor at (1,0), Eq. (12) gives w_x=2 and w_lap=2, whereas the standard Sobel x kernel has weight 1 for that neighbor and the standard 4-neighbor Laplacian has weight 1 (and 0 for diagonal neighbors). Feeding such rescaled or differently weighted perception vectors into a network trained on grid-specific filter responses is not guaranteed to reproduce the same dynamics. The stencils must be normalized to preserve the scale and sparsity of the grid ke
  3. [§2.1.1, Eq. (3)] The interpretation of the NCA update as a PDE is formal: Eq. (3) is a rewrite of Eq. (1) with a stochastic Bernoulli mask, and no well-posedness, consistency, or convergence analysis is given. This may be acceptable as an analogy for steering homogeneous growth, but the paper uses the PDE analogy as the foundation for the discretization-transfer claim. Without at least a numerical consistency check (e.g., grid refinement showing convergence of generated statistics to a limit), the PDE analogy cannot bear the weight placed on it in Sec. 3.2.
  4. [§3.3, Fig. 17] The asynchronous growth and in-situ repair demonstrations are presented without quantitative evaluation. The repair result is judged visually; there is no measurement of how well the regrown patch matches the surrounding trabecular statistics (e.g., S2, lineal path, volume fraction) or whether the repaired region restores effective stiffness or connectivity. Since 'self-repair' and 'asynchronous growth' are listed as contributions, these claims need quantitative support, or they should be clearly framed as qualitative demonstrations.
minor comments (5)
  1. [§2.2.2, Eq. (7)-(8)] The Riemannian-metric perception notation is dense. In Eq. (7), the term μ^{-1/2} g^{-1}∇u and the definition of the anisotropic Laplacian in Eq. (8), including the factor √|μI|, are difficult to parse. Please expand the derivation and define each symbol explicitly (e.g., whether μ is a scalar and how √|μI| reduces when μ is scalar).
  2. [Table 1] The 'Uncloaked reference' and 'Optimized result' rows report no standard deviation or number of samples, while the NCA-generated rows report mean±std over 20 samples. Clarify whether the first two are deterministic single FEA runs, and add the resolution at which the 'Optimized result' was computed.
  3. [§3.2, Fig. 15] Figure 15(b) is the sole evidence for surface-mesh transfer. The figure lacks information about mesh resolution, number of vertices, and the degree of irregularity. At minimum, add these details and, ideally, side-by-side grid-grown and mesh-grown results at comparable resolution.
  4. [§5, Conclusion] The conclusion states that 'the learned growth rule can be transferred across different domain geometries and representations' as if this were established. Given the current evidence, this overstates the support; please align the conclusion with the actual validation level.
  5. [Throughout] There are many minor typos and spacing errors (e.g., 'improves theaverage' and missing spaces around equations in the text). A careful proofreading pass is recommended.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity; the core training, control, and multiscale demonstrations are independently validated rather than reducing to their inputs.

full rationale

The paper's central derivation chain is not circular. The NCA is trained from a single template by matching frozen VGG-16 style statistics (Eq. 4), and the resulting microstructures are validated with different statistical descriptors—two-point correlation and lineal-path functions (Fig. 3)—so the validation is not identical to the training objective by construction. The post-training controls are explicit modifications of the perception operators: gradient rotation (Eq. 5), Riemannian-metric reshaping (Eqs. 7–8), and boundary-speed thickness evolution (Eq. 9). These are direct actuation mechanisms, not fitted predictions, and their morphological and homogenized-stiffness consequences are checked by independent generation and numerical homogenization (Figs. 6, 8, 9, 11). The multiscale optimization is constrained to the property space sampled from the same NCA (Sec. 2.3) via a VAE latent tube (Eqs. 16–18), so the optimized stiffness distribution is admittedly limited to the model's reachable property space; however, the cloaking objective is external, and the final performance is evaluated by full-scale finite-element simulation of separately grown NCA microstructures (Table 1), which yields independent errors (26.12% and 16.86%) rather than recovering the fitted homogenized value (8.10%). Thus no claimed prediction is forced by an input by construction. The paper contains self-citations by the corresponding author (e.g., refs. [13,15,41,44]), but they are contextual references and are not load-bearing evidence for a uniqueness or prediction claim. The main weakness—the discretization-transfer claim resting on an unproved PDE analogy and mesh stencils (Eq. 12) that are not shown to be discretization-invariant—is a matter of evidence and correctness, not circularity.

Assumptions & free parameters 5 free parameters · 5 assumptions · 0 invented entities

No new physical entities are introduced; hidden state channels and stochastic masks are model-internal mechanisms, not claimed as physical. The framework depends on broad domain assumptions rather than fitted constants; hand-chosen hyperparameters (Tables A1–A2) are listed but are not fitted to the target cloaking result. The main unvalidated assumption is the PDE-equivalence of the learned NCA rule, which underlies the cross-discretization transfer claim.

free parameters (5)
  • NCA hidden width D = 96
    MLP hidden width in Table A1; chosen by hand, not fitted to the target result.
  • stochastic update mask probability p = 0.5
    Bernoulli mask probability in Sec. 2.1.1; hand-chosen to break symmetry and act as spatial dropout.
  • VAE latent dimension d_z = 4
    Latent dimensionality in Table A2; chosen by hand for the optimization constraint formulation.
  • latent tube radius fraction r_frac = 0.05
    Tube radius scaling in Table A2 and Sec. 4.1; hand-chosen to balance feasibility and design freedom.
  • constraint limit gamma = 0.2
    Allowable mean fraction of elements outside the tube (Eq. 18, Table A2); chosen by hand.
assumptions (5)
  • domain assumption The learned NCA update (Eq. 3) is a discretization of a continuous PDE, so the same network weights transfer across domains and discretizations once local perception operators are adapted.
    Invoked in Sec. 2.1.1 (Eq. 3) and Sec. 3.2; this is the load-bearing premise for the cross-discretization transfer claim, but no convergence proof is given.
  • domain assumption A frozen ImageNet VGG-16 with sliced optimal transport style loss is a sufficient statistical descriptor to learn local growth rules that preserve the mechanical properties of the template.
    Used in Sec. 2.1.2; validation uses two-point/lineal path and homogenized surfaces for six cases, but there is no guarantee for arbitrary templates.
  • standard math The angularly-weighted mesh stencil in Eq. (12) yields mesh analogs of the Sobel/Laplacian grid kernels that are compatible with the trained grid perception.
    Sec. 3.2 Eq. (12); a plausible discretization assumption, but no numerical convergence test against the grid operators is reported.
  • domain assumption The 10,000-sample generated property space is representative and smooth enough for VAE training and latent-tube optimization.
    Sec. 2.3 and 4.1; scale steering is excluded from the property space, and its effect on anisotropy is inconsistently stated in Fig. 9.
  • domain assumption Homogenized effective stiffness C describes the macroscale FEA response at 50×50 elements and at 500×500/1000×1000 resolutions.
    Sec. 4; standard multiscale practice, but disorder fluctuations evidently create the gap between the 8.1% homogenized optimum and the 16.9–26.1% realized cloak errors.

how reviews work

0 comments
Cite this review

Pith. "Pith review of One-Shot Generative Design for Disordered Metamaterials via Self-Organizing Neural Cellular Automata." pith.science (2026). https://pith.science/paper/KUHWD5TR

@misc{pith2026260714475,
  author       = {Pith},
  title        = {Pith review of: One-Shot Generative Design for Disordered Metamaterials via Self-Organizing Neural Cellular Automata},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/KUHWD5TR}},
  note         = {Machine review of arXiv:2607.14475}
}
read the original abstract

Disordered metamaterials feature microstructures with inherent randomness and irregularity, enabling them to achieve broader property coverage and superior performance unavailable in their regular counterparts. Despite their promise, designing disordered microstructures is substantially harder than designing regular ones. Their design remains trapped between manual parameterizations with limited expressiveness, and generative AI that is data-hungry and struggles to generalize. To address these limitations, we propose a generative design framework based on Neural Cellular Automata that dynamically grows complex microstructures through learned local interaction rules, inspired by the self-organizing processes in natural materials. This framework requires only a single training template, yet accommodates diverse disordered microstructures and adapts to irregular domains and arbitrary discretizations. By manipulating the learned local rules, we can steer the growth process to generate microstructures unseen during training, providing control over orientation, anisotropy, and directional thickness without retraining. As a dynamic, local growth process, it naturally produces spatially varying microstructures that transition smoothly to enable location-specific mechanical properties. We demonstrate this in a multiscale mechanical cloaking design, where microstructures vary across the space to meet an optimized heterogeneous property distribution. Our design enables excellent cloaking performance without complicated post-processing and incompatible assembly common in existing methods. This data-efficient, generalizable approach opens access to previously intractable disordered materials for biomedical implants and soft robotics.

Figures

Figures reproduced from arXiv: 2607.14475 by the authors.

Figure 1
Figure 1. Overview of the proposed generative framework for disordered microstructures. (a) Learning and controls on homogeneous microstructure. An NCA model is trained to iteratively grow a microstructure through local perception and mapping stages, where each cell perceives its neighborhood and maps to a new state. Four steering mechanisms provide continuous control over the orientation, thickness, anisotropy, and scale of … view at source ↗
Figure 2
Figure 2. Growth process and model architecture of the NCA framework illustrated on the leaf vein microstructure. Each cell carries a multi-channel state vector 𝐮 𝑡 , whose first channel encodes the solid-void phase, corresponding to the local regions in the microstructure. The model iteratively applies a shared local update rule through fixed perception and mapping stages to grow the microstructure. Local neighborhood inform… view at source ↗
Figure 3
Figure 3. Comparison between the leaf template and generated samples on the two-point correlation function 𝑆2 , and the lineal-path function 𝐿. The close overlap between the generated average curves and the template curves, together with the narrow standard-deviation bands, shows that the generated leaf structures reproduce the key morphological features of the template, including spatial correlations and solid phase connecti… view at source ↗
Figures from the paper (18 more)
Figure 4
Figure 4. Figure 4: Validation of the NCA generative framework on six biologically and physically distinct disordered microstructures. (a) Six template microstructures (outer ring) and their corresponding NCA-generated realizations (inner ring), each produced by an independently trained m…
Figure 5
Figure 5. Figure 5: Orientation steering applied to the local perception operation. The native gradients extracted by the Sobel filters are rotated by a prescribed steering angle 𝜃 via a rotation matrix 𝐑(𝜃) before being passed to the trained update network, controlling the growth to foll…
Figure 6
Figure 6. Figure 6: Generated microstructures and corresponding elastic surfaces under orientation steering. An arrow with a gradually changing color shows the increasing orientation angle 𝜃 at the top. The elastic surfaces of 100 independently generated samples at each 𝜃 represent the gr…
Figure 7
Figure 7. Figure 7: Anisotropy and scale controls applied to the NCA perception stage. The original perception is modified by a Riemannian metric tensor 𝐠(𝑠1 , 𝑠2 ) parameterized by the stretching parameter 𝑠1 and shear parameter 𝑠2 , which reshapes the local perceptual space to stretch o…
Figure 8
Figure 8. Figure 8: Generated microstructures and corresponding elastic surfaces under anisotropy steering, following the same layout as [PITH_FULL_IMAGE:figures/full_fig_p010_8.png]
Figure 9
Figure 9. Figure 9: Generated microstructures and corresponding elastic surfaces under scale steering, following the same layout as [PITH_FULL_IMAGE:figures/full_fig_p011_9.png]
Figure 10
Figure 10. Figure 10: Directional thickness control applied as a post-processing boundary evolution stage. After the NCA generation, the solid-void boundary is detected through local gradients, using the same Sobel filters in the perception stage. A signed speed operator parameterized by t…
Figure 11
Figure 11. Figure 11: Generated microstructures and corresponding elastic surfaces under directional thickness steering, following the same layout as [PITH_FULL_IMAGE:figures/full_fig_p012_11.png]
Figure 12
Figure 12. Figure 12: Mechanical property distribution of the generated dataset. The pairwise distribution matrix shows the joint and marginal distributions of the six independent stiffness components 𝐶11, 𝐶12, 𝐶16, 𝐶22, 𝐶26, and 𝐶66 across 10,000 generated microstructures, where the diago…
Figure 13
Figure 13. Figure 13: Individual controlling effect of each control parameter on the stiffness distribution. Each panel shows the distribution of a pair of stiffness components, colored by the value of the corresponding control parameter. (a) Increasing the stretching parameter 𝑠1 shifts s…
Figure 14
Figure 14. Figure 14: Spatially graded control and microstructure transition in the NCA framework. (a) Smooth transitions produced by spatially varying individual control parameters. Each parameter varies continuously from top to bottom across the domain, leading to gradual changes in orie…
Figure 15
Figure 15. Figure 15: Transfer of the trained NCA framework from a uniform grid discretization to a surface-mesh discretization by redefining local perception operators. (a) A prescribed global flow direction 𝐝 is projected onto the local tangent plane at each mesh vertex to define a consi…
Figure 16
Figure 16. Figure 16: Asynchronous growth in an irregular domain guided by prescribed growth sources and sequence. (a) Two source points (triangle marks) define the growth sequence that activates cells progressively from the sources outward. Once the cells are activated, they keep growing …
Figure 17
Figure 17. Figure 17: In-situ repair of a damaged trabecular bone microstructure utilizing the NCA generation framework. A circular region is removed from a real cross-sectional bone slice to simulate a localized structural defect. An NCA model is trained on the neighboring region surround…
Figure 18
Figure 18. Figure 18: Optimization pipeline and problem setup for the mechanical cloaking case. (a) The macroscopic domain is divided into a cloaking region Ω𝑐 , a reference region Ω𝑟 , and a void region Ω𝑣 , loaded by compressive pressure on the left and right edges while rigid-body motio…
Figure 19
Figure 19. Figure 19: Optimized stiffness distribution and projected control parameter fields in the cloaking region. (a) Spatially varying distributions of the six stiffness components, where the angular distribution pattern reflects the directional stiffness redistribution required to re…
Figure 20
Figure 20. Figure 20: NCA-generated full cloaking structure at 500 × 500 resolution and its displacement field comparison. The left panel shows the generated heterogeneous microstructure in the given resolution. The right panels compare 𝑈𝑥 and 𝑈𝑦 across three configurations, including the …
Figure 21
Figure 21. Figure 21: NCA-generated full cloaking structure at 1000× 1000 resolution and its displacement field comparison, following the same layout as [PITH_FULL_IMAGE:figures/full_fig_p024_21.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

74 extracted references · 3 linked inside Pith

  1. [1]

    P. Jiao, J. Mueller, J. R. Raney, X. Zheng, A. H. Alavi, Mechanical metamaterials and beyond, Nat. Commun. 14 (2023) 6004

    2023

  2. [2]

    Bertoldi, V

    K. Bertoldi, V. Vitelli, J. Christensen, M. van Hecke, Flexible mechanical metamaterials, Nat. Rev. Mater. 2 (2017) 17066

    2017

  3. [3]

    D.Lee,W.Chen,L.Wang,Y.-C.Chan,W.Chen, Data-drivendesignformetamaterialsandmultiscalesystems:areview, AdvancedMaterials 36 (2024) 2305254

    2024

  4. [4]

    Bückmann, M

    T. Bückmann, M. Thiel, M. Kadic, R. Schittny, M. Wegener, An elasto-mechanical unfeelability cloak made of pentamode metamaterials, Nat. Commun. 5 (2014) 4130

    2014

  5. [5]

    Boddapati, C

    J. Boddapati, C. Daraio, Planar structured materials with extreme elastic anisotropy, Mater. Des. 246 (2024) 113348

    2024

  6. [6]

    Z. Lin, L. S. Novelino, H. Wei, N. A. Alderete, G. H. Paulino, H. D. Espinosa, S. Krishnaswamy, Digital mechanical metamaterial with programmable functionality, Adv. Mater. 36 (2024) 2406263

    2024

  7. [7]

    Zhang, M

    J. Zhang, M. Xiao, L. Gao, A. Alù, F. Wang, Self-bridging metamaterials surpassing the theoretical limit of Poisson’s ratios, Nat. Commun. 14 (2023) 4041

    2023

  8. [8]

    A. J. D. Shaikeea, H. Cui, M. O’Masta, X. R. Zheng, V. S. Deshpande, The toughness of mechanical metamaterials, Nat. Mater. 21 (2022) 297–304

    2022

Show all 74 references
  1. [9]

    J. U. Surjadi, B. F. G. Aymon, A. Kumar, L. Wu, J. Lem, K. N. Kamrin, C. M. Portela, Hierarchical granular metamaterials, arXiv (2026) 2606.22795

    2026 arXiv

  2. [10]

    K. Liu, R. Sun, C. Daraio, Growth rules for irregular architected materials with programmable properties, Science 377 (2022) 975–981

    2022

  3. [11]

    C. E. Gregg, D. Catanoso, B. O. Irene, I. Kostitsyna, M. E. Ochalek, T. J. Olatunde, I. W. Park, F. M. Sebastianelli, E. M. Taylor, G. T. Trinh, K. C. Cheung, Ultralight, strong, and self-reprogrammable mechanical metamaterials, Sci. Robot. 9 (2024) eadi2746

    2024

  4. [12]

    S. A. Cummer, J. Christensen, A. Alù, Controlling sound with acoustic metamaterials, Nat. Rev. Mater. 1 (2016) 16001

    2016

  5. [13]

    L. Wang, Z. Liu, D. Da, Y.-C. Chan, W. Chen, P. Zhu, Generalized de-homogenization via sawtooth-function-based mapping and its demonstrationondata-drivenfrequencyresponseoptimization,ComputerMethodsinAppliedMechanicsandEngineering395(2022)114967

    2022

  6. [14]

    Z. Gao, X. Zhang, Y. Wu, Y. Ni, Y. Lu, C. Xia, H. Wang, H. Wang, A metamaterial scaffold beyond modulus limits: enhanced osteogenesis and angiogenesis of critical bone defects, Nat. Commun. 16 (2025) 2094

    2025

  7. [15]

    L. Wang, J. Boddapati, K. Liu, P. Zhu, C. Daraio, W. Chen, Mechanical cloak via data-driven aperiodic metamaterial design, Proc. Natl. Acad. Sci. U.S.A. 119 (2022) e2122185119

    2022

  8. [16]

    Z.Feng,H.Xu, Designingperiodicmetamaterialsinspiredbystochasticmicrostructuresviatexture-synthesis-drivenexplainableAI, J.Mech. Des. 148 (2026) 051703

    2026

  9. [17]

    Pahlavani, K

    H. Pahlavani, K. Tsifoutis-Kazolis, M. C. Saldivar, P. Mody, J. Zhou, M. J. Mirzaali, A. A. Zadpoor, Deep learning for size-agnostic inverse design of random-network 3D printed mechanical metamaterials, Adv. Mater. 36 (2024) 2303481

    2024

  10. [18]

    X. Fang, J. Wen, L. Cheng, D. Yu, H. Zhang, P. Gumbsch, Programmable gear-based mechanical metamaterials, Nat. Mater. 21 (2022) 869–876

    2022

  11. [19]

    Zaiser, S

    M. Zaiser, S. Zapperi, Disordered mechanical metamaterials, Nat. Rev. Phys. 5 (2023) 679–688

    2023

  12. [20]

    Oftadeh, M

    R. Oftadeh, M. Perez-Viloria, J. C. Villa-Camacho, A. Vaziri, A. Nazarian, Biomechanics and mechanobiology of trabecular bone: a review, J. Biomech. Eng. 137 (2015) 0108021

    2015

  13. [21]

    X.Wang,L.Zhu,L.Sun,N.Li,Breakingthetradeoffsbetweendifferentmechanicalpropertiesinbioinspiredhierarchicallatticemetamaterials, Adv. Funct. Mater. 33 (2023) 2305978

    2023

  14. [22]

    M.Shen,K.Liu,S.Mao,C.Daraio, Self-supervisedAIfordecodinganddesigningdisorderedmetamaterials, Sci.Adv.12(2026)eadx7389

    2026

  15. [23]

    J. Gim, N. Schnitzer, L. M. Otter, Y. Cui, S. Motreuil, F. Marin, S. E. Wolf, D. E. Jacob, A. Misra, R. Hovden, Nanoscale deformation mechanics reveal resilience in nacre ofPinna nobilisshell, Nat. Commun. 10 (2019) 4822

    2019

  16. [24]

    S. Luan, E. Chen, J. John, S. Gaitanaros, A data-driven framework for structure-property correlation in ordered and disordered cellular metamaterials, Sci. Adv. 9 (2023) eadi1453

    2023

  17. [25]

    Y.Liu,B.Xia,K.Liu,Y.Zhou,K.Wei, Robustnessanddiversityofdisorderedstructuresonsoundabsorptionanddeformationresistance, J. Mech. Phys. Solids 190 (2024) 105751

    2024

  18. [26]

    S.Fulco,M.K.Budzik,H.Xiao,D.J.Durian,K.T.Turner, Disorderenhancesthefracturetoughnessof2Dmechanicalmetamaterials, PNAS Nexus 4 (2025) pgaf023

    2025

  19. [27]

    Fulco, P

    S. Fulco, P. K. Purohit, M. K. Budzik, K. T. Turner, Fracture of disordered and stochastic lattice materials, arXiv (2025) 2508.21187

    2025 arXiv

  20. [28]

    Y.Jia,K.Liu,X.S.Zhang, Modulatestressdistributionwithbio-inspiredirregulararchitectedmaterialstowardsoptimaltissuesupport, Nat. Commun. 15 (2024) 4072

    2024

  21. [29]

    Y. Liu, B. Xia, Y. Zhou, K. Wei, Disordered mechanical metamaterials with programmable properties, Acta Mater. 285 (2025) 120700

    2025

  22. [30]

    D. R. Reid, N. Pashine, J. M. Wozniak, H. M. Jaeger, A. J. Liu, S. R. Nagel, J. J. de Pablo, Auxetic metamaterials from disordered networks, Proc. Natl. Acad. Sci. U.S.A. 115 (2018) E1384–E1390

    2018

  23. [31]

    Hanifpour, C

    M. Hanifpour, C. F. Petersen, M. J. Alava, S. Zapperi, Mechanics of disordered auxetic metamaterials, Eur. Phys. J. B 91 (2018) 271

    2018

  24. [32]

    Dhulipala, C

    S. Dhulipala, C. M. Portela, Curvature-guided mechanics and design of spinodal and shell-based architected materials, J. Mech. Phys. Solids 204 (2025) 106273

    2025

  25. [33]

    H. Xu, Y. Li, C. Brinson, W. Chen, A descriptor-based design methodology for developing heterogeneous microstructural materials system, J. Mech. Des. 136 (2014) 051007

    2014

  26. [34]

    H. Xu, D. A. Dikin, C. Burkhart, W. Chen, Descriptor-based methodology for statistical characterization and 3D reconstruction of microstructural materials, Comput. Mater. Sci. 85 (2014) 206–216

    2014

  27. [35]

    H.Deng,Q.Zhao,T.Liu,Z.Zhang,Z.Zhu,Y.Zhou,H.-X.Peng,W.Yang,H.Gao,H.Zhou, Beyondunitcells:Programmablemorphogenetic design of irregular architected materials, Proc. Natl. Acad. Sci. U.S.A. 123 (2026) e2606412123

    2026

  28. [36]

    X.Zheng,T.-T.Chen,X.Jiang,M.Naito,I.Watanabe, Deep-learning-basedinversedesignofthree-dimensionalarchitectedcellularmaterials with the target porosity and stiffness using voxelized voronoi lattices, Sci. Technol. Adv. Mater. 24 (2023) 2157682. Page 26 of 28

    2023

  29. [37]

    L. He, M. Zhao, J. P. Y. Cheung, T. Zhang, X. Ren, Gaussian random field-based characterization and reconstruction of cancellous bone microstructure considering the constraint of correlation structure, J. Mech. Behav. Biomed. Mater. 152 (2024) 106443

    2024

  30. [38]

    Kumar, S

    S. Kumar, S. Tan, L. Zheng, D. M. Kochmann, Inverse-designed spinodoid metamaterials, npj Comput. Mater. 6 (2020) 73

    2020

  31. [39]

    Zheng, S

    L. Zheng, S. Kumar, D. M. Kochmann, Data-driven topology optimization of spinodoid metamaterials with seamlessly tunable anisotropy, Comput. Methods Appl. Mech. Eng. 383 (2021) 113894

    2021

  32. [40]

    F.V.Senhora,E.D.Sanders,G.H.Paulino, Optimally-Tailoredspinodalarchitectedmaterialsformultiscaledesignandmanufacturing, Adv. Mater. 34 (2022) 2109304

    2022

  33. [41]

    S. Deng, D. Lee, A. Chandrasekhar, S. Knapik, L. Wang, H. D. Espinosa, W. Chen, Data-driven topology optimization for multiscale biomimetic spinodal design, arXiv (2025) 2506.23420

    2025

  34. [42]

    C. M. Portela, A. Vidyasagar, S. Krödel, T. Weissenbach, D. W. Yee, J. R. Greer, D. M. Kochmann, Extreme mechanical resilience of self-assembled nanolabyrinthine materials, Proc. Natl. Acad. Sci. U.S.A. 117 (2020) 5686–5693

    2020

  35. [43]

    Y. Jia, K. Liu, X. S. Zhang, Unstructured growth of irregular architectures for optimized metastructures, J. Mech. Phys. Solids 192 (2024) 105787

    2024

  36. [44]

    Methods Appl

    L.Wang,Y.-C.Chan,F.Ahmed,Z.Liu,P.Zhu,W.Chen,Deepgenerativemodelingformechanistic-basedlearninganddesignofmetamaterial systems, Comput. Methods Appl. Mech. Eng. 372 (2020) 113377

    2020

  37. [45]

    Xiang, J

    Y. Xiang, J. Hou, X. Chen, K. Tang, X. Wang, Decoupled design of hybrid mechanical metamaterials via ensembled deep learning, Int. J. Mech. Sci. 301 (2025) 110514

    2025

  38. [46]

    S. C. Shen, M. J. Buehler, Nature-inspired architected materials using unsupervised deep learning, Commun. Eng. 1 (2022) 37

    2022

  39. [47]

    Y. Mao, Q. He, X. Zhao, Designing complex architectured materials with generative adversarial networks, Sci. Adv. 6 (2020) eaaz4169

    2020

  40. [48]

    5 (2023) 1466–1475

    J.-H.Bastek,D.M.Kochmann, Inversedesignofnonlinearmechanicalmetamaterialsviavideodenoisingdiffusionmodels, Nat.Mach.Intell. 5 (2023) 1466–1475

    2023

  41. [49]

    R.Guo,M.Alves,M.Mehdikhani,C.Breite,Y.Swolfs, Synthesisingrealistic2dmicrostructuresofunidirectionalfibre-reinforcedcomposites with a generative adversarial network, Compos. Sci. Technol. 250 (2024) 110539

    2024

  42. [50]

    J. P. Quesada-Molina, H. Mofatteh, A. Akbarzadeh, S. Mariani, Navigating beyond the training set: A deep learning framework for inverse design of architected composite materials, Eng. Appl. Artif. Intell. 150 (2025) 110473

    2025

  43. [51]

    A. M. Turing, The chemical basis of morphogenesis, Philos. Trans. R. Soc. Lond. B Biol. Sci. 237 (1952) 37–72

    1952

  44. [52]

    S.Kondo,T.Miura, Reaction-Diffusionmodelasaframeworkforunderstandingbiologicalpatternformation, Science329(2010)1616–1620

    2010

  45. [53]

    A.Runions,M.Fuhrer,B.Lane,P.Federl,A.-G.Rolland-Lagan,P.Prusinkiewicz, Modelingandvisualizationofleafvenationpatterns, ACM Transactions on Graphics 24 (2005) 702–711

    2005

  46. [54]

    Mordvintsev, E

    A. Mordvintsev, E. Randazzo, E. Niklasson, M. Levin, Growing neural cellular automata, Distill 5 (2020) e23

    2020

  47. [55]

    Seibert, A

    P. Seibert, A. Raßloff, Y. Zhang, K. Kalina, P. Reck, D. Peterseim, M. Kästner, Reconstructing microstructures from statistical descriptors using neural cellular automata, Integr. Mater. Manuf. Innov. 13 (2024) 272–287

    2024

  48. [56]

    Pajouheshgar, Y

    E. Pajouheshgar, Y. Xu, A. Abbasi, A. Mordvintsev, W. Jakob, S. Süsstrunk, Neural cellular automata: From cells to pixels, arXiv (2026) 2506.22899

    2026 arXiv

  49. [57]

    Andreassen, C

    E. Andreassen, C. S. Andreasen, How to determine composite material properties using numerical homogenization, Comput. Mater. Sci. 83 (2014) 488–495

    2014

  50. [58]

    103 (2016) 89–102

    R.Bostanabad,A.T.Bui,W.Xie,D.W.Apley,W.Chen,Stochasticmicrostructurecharacterizationandreconstructionviasupervisedlearning, Acta Mater. 103 (2016) 89–102

    2016

  51. [59]

    R.Bostanabad,Y.Zhang,X.Li,T.Kearney,L.C.Brinson,D.W.Apley,W.K.Liu,W.Chen, Computationalmicrostructurecharacterization and reconstruction: Review of the state-of-the-art techniques, Prog. Mater. Sci. 95 (2018) 1–41

    2018

  52. [60]

    Simonyan, A

    K. Simonyan, A. Zisserman, Very deep convolutional networks for large-scale image recognition, in: Proc. Int. Conf. Learn. Represent., pp. 1–14

  53. [61]

    Y. Jiao, F. H. Stillinger, S. Torquato, Modeling heterogeneous materials via two-point correlation functions: I. basic principles, Phys. Rev. E 76 (2007) 031110

    2007

  54. [62]

    Loseille, F

    A. Loseille, F. Alauzet, Continuous mesh framework. part i: Well-posed continuous interpolation error, SIAM J. Numer. Anal. 49 (2011) 38–60

    2011

  55. [63]

    A. P. McAnena, T. McClennen, H. Zheng, Patient-specific 3-Dimensional-printed orthopedic implants and surgical devices are potential alternatives to conventional technology but require additional characterization, Clin. Orthop. Surg. 17 (2025) 1–15

    2025

  56. [64]

    Knöppel, K

    F. Knöppel, K. Crane, U. Pinkall, P. Schröder, Globally optimal direction fields, ACM Transactions on Graphics 32 (2013) 59:1–59:10

    2013

  57. [65]

    Y. Yang, S. Liu, H. Pan, Y. Liu, X. Tong, PFCNN: Convolutional neural networks on 3D surfaces using parallel frames, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 13578–13587

  58. [66]

    J.Masci,D.Boscaini,M.M.Bronstein,P.Vandergheynst, GeodesicconvolutionalneuralnetworksonRiemannianmanifolds, in:Proceedings of the IEEE International Conference on Computer Vision Workshops (ICCVW), pp. 37–45

  59. [67]

    F. V. Senhora, E. Sanders, G. Paulino, Unbiased mechanical cloaks, Zenodo (2025) 15271218

    2025

  60. [68]

    X.-Y.Zhang,G.Fang,J.Zhou, Additivelymanufacturedscaffoldsforbonetissueengineeringandthepredictionoftheirmechanicalbehavior: A review, Materials 10 (2017) 50

    2017

  61. [69]

    Cheikho, J

    K. Cheikho, J. F. Ganghoffer, A. Baldit, E. Labbé, S. Alix, H. Kerdjoudj, C. Mauprivez, A. Lebée, C. Laurent, A flexible design framework to design graded porous bone scaffolds with adjustable anisotropic properties, J. Mech. Behav. Biomed. Mater. 140 (2023) 105727

    2023

  62. [70]

    Babazadeh-Naseri, G

    A. Babazadeh-Naseri, G. Li, M. S. Shourijeh, J. E. Akin, C. F. Higgs, B. J. Fregly, N. J. Dunbar, Stress-shielding resistant design of custom pelvic prostheses using lattice-based topology optimization, Med. Eng. Phys. 121 (2023) 104012

    2023

  63. [71]

    P. S. P. Poh, D. Valainis, K. Bhattacharya, M. van Griensven, P. Dondl, Optimization of bone scaffold porosity distributions, Sci. Rep. 9 (2019) 9170. Page 27 of 28

    2019

  64. [72]

    Schaffner, J

    M. Schaffner, J. A. Faber, L. Pianegonda, P. A. Rühs, F. Coulter, A. R. Studart, 3d printing of robotic soft actuators with programmable bioinspired architectures, Nat. Commun. 9 (2018) 878

    2018

  65. [73]

    114 (2017) 51–56

    F.Connolly,C.J.Walsh,K.Bertoldi, Automaticdesignoffiber-reinforcedsoftactuatorsfortrajectorymatching, Proc.Natl.Acad.Sci.U.S.A. 114 (2017) 51–56

    2017

  66. [74]

    X.Ni,H.Luan,J.-T.Kim,S.I.Rogge,Y.Bai,J.W.Kwak,S.Liu,D.S.Yang,S.Li,S.Li,Z.Li,Y.Zhang,C.Wu,X.Ni,Y.Huang,H.Wang, J. A. Rogers, Soft shape-programmable surfaces by fast electromagnetic actuation of liquid metal networks, Nat. Commun. 13 (2022) 5576. Page 28 of 28

    2022

Pith tools

Reviewed August 2, 2026 · model on record in the stance chip above.