Pith. sign in

REVIEW 3 major objections 4 minor 54 references

Efficient strain-space hyperreduction in large-deformation solid mechanics

T0 review · 3 major / 4 minor · reviewed 2026-08-01 · deepseek-v4-flash

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

Pith's one-line read This paper claims that strain-space hyperreduction, previously limited to periodic-cell homogenisation, can be lifted to general large-deformation solid problems with arbitrary Dirichlet boundary conditions, delivering speedups up to 100,00

desk verdict A credible, well-executed extension of strain-space hyperreduction beyond RVEs, but the 'arbitrary Dirichlet BC' claim outruns the validation: only scalar and proportional loads are tested. read the letter →

arxiv 2607.19330 v2 pith:JILYK3IX submitted 2026-07-21 cs.CE

Erik Faust , Lisa Scheunemann This is my paper · ORCID
classification cs.CE
keywords modelorderreductionhyperreductionstrain-spaceEmpiricalCubatureMethodE3CEMSLDirichletboundaryconditionslarge-deformationhyperelasticity
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

Strain-space model order reduction has been confined to computational homogenisation of periodic unit cells, but this paper shows how to extend it to arbitrary solid-mechanics problems with prescribed boundary displacements. The key move is a lifting: boundary-consistent strain fields are precomputed once, scaled by the boundary values, and subtracted so that the remaining fluctuation field satisfies homogeneous boundary conditions and can be reduced efficiently. On two hyperelastic plate-with-holes benchmarks, the lifted strain-space versions of ECM, E3C, and EMSL all outperform displacement-space ECSW in the accuracy-versus-runtime tradeoff. In particular, E3C and EMSL achieve 10,000- and 100,000-fold speedups while retaining high accuracy, suggesting that strain-space hyperreduction is a practical tool for the broader class of large-deformation solid problems.

What carries the argument

The load-bearing machinery is the lifting-field split: for each Dirichlet boundary, a unit boundary-value solution is computed offline at reference material parameters, then arbitrary boundary values scale and superpose these fields, and the solver works on the fluctuation field, which is zero on all boundaries. This converts all inhomogeneous Dirichlet conditions into homogeneous ones, so POD modes can be built from integration-point strain snapshots and reduced models operate directly on displacement-gradient values, bypassing element-level assembly entirely. On top of this split, hyperreduction is performed either by selecting sparse integration points with optimised weights (strain-space

What would settle it

Run the same reduced models on a large-deformation problem with pronounced strain localisation (e.g., a nearly incompressible block, a necking bar, or a structure with multiple interacting nonzero Dirichlet boundaries under large relative displacements) and measure whether the fluctuation field remains low-rank; if the scaled lifting fails to absorb boundary effects, the reduced-model error will rise sharply even at high integration-point counts, while the full-order model stays accurate.

Watch

Extended reading notes

Core claim

The central discovery is that the whole machinery of strain-space hyperreduction transfers from periodic unit-cell problems to general solids once the displacement-gradient field is split into a boundary-consistent lifting field and a fluctuation field. With this split, Dirichlet boundary conditions are satisfied by construction, strain-space POD modes vanish on the boundary, and reduced cubature methods (ECM, E3C) as well as cluster-wise material-linearisation methods (EMSL) become directly applicable. In the reported experiments, strain-space methods need far fewer integration points than displacement-space ECSW for a given accuracy; for example, in the second benchmark E3C reaches below 0

Load-bearing premise

The precomputed lifting fields are computed once at a central material parameter set and simply scaled by boundary values, relying on linear superposition of boundary effects to remain a good approximation across the full parameter and deformation range of a nonlinear problem.

Editorial extensions

If this is right

  • Strain-space hyperreduction now applies to general 3D solid-mechanics problems with arbitrarily prescribed Dirichlet boundary values, not just periodic unit cells.
  • Because the reduced models operate on strains and stresses at integration points, training requires only displacement, strain, and stress snapshots plus a material routine, making deployment with black-box solvers practical.
  • On the two hyperelastic benchmarks, strain-space methods Pareto-dominate displacement-space ECSW: EMSL is the best choice when online and offline runtimes are extremely limited, while E3C yields the highest accuracy when a moderate runtime budget is acceptable.
  • The reported 10,000- and 100,000-fold speedups point toward making inverse parameter estimation, optimisation, and repeated simulation of large-deformation components computationally feasible.
  • Offline costs differ significantly across the methods, with E3C requiring the most expensive training (around three and a half hours in the second example) and EMSL the cheapest (about 30 seconds), which matters when reduced models must be built quickly.

Reading between the lines

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

  • The lifting approach is linear by construction, so for strongly nonlinear, path-dependent problems the precomputed boundary-consistent fields may need to be supplemented or recomputed during the simulation; the paper notes alternative lifting strategies are deferred to future work.
  • The very large speedups are demonstrated on moderately nonlinear hyperelastic problems whose solution manifold is low-dimensional (roughly 10-15 POD modes suffice); on more strongly nonlinear or history-dependent problems the advantage over displacement-space ECSW is likely smaller but may remain substantial.
  • Because the reduced solver never touches element-level details, a natural testable extension is to transplant the same reduced operators across different element types, mesh resolutions, or FE codes without retraining the core POD basis.
  • The authors note that a material model could be inferred from stress-strain pairs; if combined with the lifting approach, this would yield a fully non-intrusive pipeline that only needs simulation data, which could broaden the applicability to legacy or industrial solvers.
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

3 major / 4 minor

Summary. The paper generalises strain-space model order reduction (MOR) techniques, previously developed for computational homogenisation, to general large-deformation hyperelastic problems with non-periodic geometries. The key ingredient is a lifting strategy: BC-consistent displacement-gradient fields are computed offline at unit boundary values and then scaled/superposed to satisfy arbitrary Dirichlet boundary values, so that the reduced solve is performed on a homogeneous fluctuation field. On this basis the authors formulate strain-space versions of ECM, E3C, and EMSL, and compare them with displacement-space ECSW on two hyperelastic plate-with-two-holes benchmarks with parameterised material behaviour and loading. They report that the strain-space methods dominate ECSW in the accuracy/runtime tradeoff, with EMSL and E3C achieving very large speedups (up to ~10^4-10^5).

Significance. If the numerical results are taken at face value, the paper makes a useful methodological contribution: it broadens the applicability of strain-space hyperreduction from RVE/periodic settings to ordinary solid-mechanical boundary value problems, and it does so in a relatively non-intrusive manner. The derivations in Sections 4-5 are clear, and the validation protocol is above average for this literature: 25 independent training and 25 validation material samples, 750 validation snapshots in the harder benchmark, and systematic sweeps over basis size d and integration-point count |H|. The paper also reports offline costs, which is commendable. The main weakness is that the central claim of handling 'arbitrary-valued, parameterised Dirichlet BCs' is not actually exercised by the numerical experiments, because the load paths are fixed curves in the boundary-value space.

major comments (3)
  1. [Abstract and §6.2] The abstract claims 'arbitrary-valued, parameterised Dirichlet boundary conditions are satisfied by construction'. The construction in Eq. (12)-(13)/(27)-(29) indeed satisfies the BCs kinematically for any boundary values, but the accuracy of the reduced model for arbitrary combinations of boundary values is not demonstrated. In §6.2 the three load cases are: tension d1∈[0,50], shear d2∈[0,25], and a single proportional mixed path d1∈[0,35], d2∈[0,20]. The two Dirichlet values are never varied independently, so the claim that the method handles multiple nonzero, independently valued Dirichlet BCs is untested. This is a load-bearing gap for the paper's central generalization claim.
  2. [§2.3, Eq. (12)-(13), footnote 1] The lifting fields are computed once at reference material parameters and at unit boundary values, then scaled linearly with d_j. For finite-deformation hyperelasticity, linear scaling and superposition of these fields is not an equilibrium solution for arbitrary d_j; all nonlinear coupling must be absorbed by the fluctuation field. The authors acknowledge this (footnote 1, §3.1) and defer a detailed study, but no experiment in §6 actually probes a regime where the linear-superposition assumption is stressed, e.g., d1=50,d2=0 versus d1=0,d2=25 versus d1=50,d2=25, or random combinations within the ranges. A validation set with independently varied d1,d2 would confirm that the fluctuation manifold remains low-rank across the intended BC parameter space.
  3. [§5, Eq. (42)-(43)] The EMSL predictor M is a linear map fitted to training snapshots along the specific load paths used in training. For the method to be said to generalise to 'arbitrary-valued' Dirichlet BCs, one should test M on boundary-value combinations not lying on those paths. As written, M may extrapolate poorly for new (d1,d2) pairs, and the paper explicitly leaves more advanced inference to future work. This is not a flaw in the EMSL derivation, but it means the scope of the numerical claim should be stated more narrowly, or an additional experiment with unseen BC combinations should be included.
minor comments (4)
  1. [Tables 1-8] The symbol '×' appears in many table entries but is never defined. If it denotes failed/non-converged training or simulation runs, this should be stated; if it denotes parameter combinations that were not attempted, that should also be explicit.
  2. [§6.1/§6.2] The reported speedups are relative to a Python FE implementation on laptop hardware, and the authors appropriately caution about absolute runtimes. It would be helpful to state the total online runtime of the full-order model per snapshot (or per load step) so that the speedup numbers can be interpreted independently of the particular Python implementation.
  3. [Fig. 18] The offline costs are reported, but the Pareto plots in Figs. 13 and 17 include only online runtime. Since the paper discusses 'method of choice when online and offline runtime budgets are very limited', it would be clearer to also mark offline cost on the Pareto plots or discuss the offline/online tradeoff explicitly in the text.
  4. [Throughout] There are occasional typos and formatting artefacts (e.g., 'displacment' in Fig. 11, 'UMPACK' in §6.1, inconsistent spacing around 'd' in tables). A careful proofread would improve presentation.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: strain-space ROM derivation is self-contained and validated against independent full-order simulations.

full rationale

The paper's derivation chain does not reduce to its inputs. The lifting split u = \bar U d + \tilde u (Eqs. 12-13) and H_g = \bar H_g d + \tilde H_g (Eqs. 27-29) is a kinematically exact change of variables; BC-consistent fields are computed once at reference parameters, and the fluctuation field is solved from the projected weak form (Eqs. 30-31). The hyperreduction weights (ECM NNLS, E3C L-BFGS, EMSL cluster preprocessing) are fitted to training snapshots, but all accuracy claims are evaluated on independent validation samples against the full-order model, with error metric Eq. (55). EMSL's predictor M in Eq. (42) is a least-squares fit from training snapshots, but it is only used to select the linearisation point; the final reduced solution is obtained by solving the affine residual equation r + T y + s = 0, not by the fit itself. Self-citations to [16,17,38] are historical and methodological; EMSL is fully re-derived in Section 5, and the numerical benchmarks are self-contained. The limitations the paper acknowledges (footnote 1 on scaling lifting fields, Section 3.1 and Section 5 deferring alternative lifting/inference strategies, and the untested case of independently varied Dirichlet values in Section 6.2) concern approximation quality and generalisation, not circularity: no equation or fitted parameter is renamed as a prediction.

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

The central claims rest on a standard MOR pipeline: POD modes, fitted hyperreduction weights, and (for EMSL) a fitted parameter-to-state map. These are trained on snapshots of the same problem class as the validation set. The paper's contribution—the lifting-based BC handling—introduces no new physics, but the validity of the scaling/superposition of lifting fields in finite deformation is a load-bearing modeling assumption not independently verified.

free parameters (6)
  • Reference material parameters for BC-consistent lifting fields (c̄1, c̄2, λ̄) = Center of sampled range: c1,c2 ∈ [100,1000], λ ∈ [400,10000] (exact values not given)
    Chosen by hand (Sections 6.1/6.2) to compute ū_j once; the paper states this choice promotes basis quality and was not systematically varied.
  • EMSL parameter-to-reduced-variable map M = Least-squares matrix from Eq. (42): M = Y p_sᵀ (p_s p_sᵀ)⁻¹
    Fitted to training snapshots; used to predict ŷ for cluster-wise linearization. EMSL's accuracy and speedup claims depend on the quality of this surrogate.
  • ECM integration weights ξ_c = Nonnegative weights selected by sparse NNLS (Eq. 36) with volume constraint (Eq. 37)
    Fitted to strain-space stress snapshots; determines accuracy of reduced cubature and the reported error/runtime tradeoff.
  • E3C integration point modes φ_c and weights ξ_c = Cluster centroids (Eqs. 39-40) plus L-BFGS-optimized positions/weights
    Fitted to minimize snapshot residuals; allows off-mesh integration points, a key source of E3C's accuracy advantage.
  • EMSL cluster assignments and centroids φ_c = Lloyd's algorithm on Gauss-point mode vectors (Eq. 41)
    Clusters determine the linearization subdomains; the number of clusters |H| is a scanned hyperparameter.
  • Reduced basis size d and reduced integration point count |H| = Scanned over d∈[5,30], |H|∈[1,100] for strain-space methods; |E_m|∈[1,100] for ECSW
    Tunable hyperparameters; the headline speedup factors correspond to selected points on the Pareto front, not to a parameter-free prediction.
assumptions (6)
  • domain assumption Mooney-Rivlin hyperelastic model (Eq. 2) faithfully represents the materials in the target problem class.
    The entire benchmark and claim are built on this constitutive law; results may not extend to elastoplasticity or other path-dependent behavior (the paper itself lists this as future work).
  • domain assumption The solution manifold of the parameterized problem is close to a low-dimensional linear subspace, so d≤30 POD modes suffice.
    Assumed for all MOR methods; the paper provides empirical evidence (tables, footnote 5) but no a priori bound. The EMSL outlier suggests the assumption can fail locally.
  • domain assumption The fluctuation strain field (after subtracting scaled lifting fields) is well-represented by the POD subspace and vanishes on all Dirichlet boundaries.
    Section 4.1: 'Because the modes computed from the fluctuation field also vanish at these boundaries, Dirichlet BCs are fulfilled by construction.' This requires the lifting to be BC-consistent; the paper notes alternative lifting strategies were not studied.
  • ad hoc to paper The lifting decomposition Eq. (12)-(13) and (27)-(29) (linear superposition and scaling of BC-consistent fields) is valid in finite deformation.
    The paper acknowledges in footnote 1 that for large deformations it can be helpful to compute the lifting for a larger d*_j and scale down; this is a linearization-like approximation whose error is not quantified.
  • domain assumption Cluster-wise linearization of the material law (Eq. 46) leaves a small second-order remainder over the whole load path.
    EMSL's accuracy relies on the linear approximation around the predicted cluster-average strain; no error bound is given, and the outlier at d=30, |H|=100 shows the approximation can fail.
  • standard math Newton-Raphson and reduced solvers converge for all validation parameters (except acknowledged outlier).
    The paper states all 750 validation simulations converged for |H|>d except the EMSL outlier; this is an empirical observation, not a proof.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Efficient strain-space hyperreduction in large-deformation solid mechanics." pith.science (2026). https://pith.science/paper/JILYK3IX

@misc{pith2026260719330,
  author       = {Pith},
  title        = {Pith review of: Efficient strain-space hyperreduction in large-deformation solid mechanics},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/JILYK3IX}},
  note         = {Machine review of arXiv:2607.19330}
}
read the original abstract

Strain-space model order reduction (MOR) techniques have recently been shown to achieve exceptional performance in terms of the tradeoff between runtime and accuracy achieved in computational homogenisation problems. In this article, we generalise such techniques to problems in large-deformation solid mechanics beyond the context of computational homogenisation. Arbitrary-valued, parameterised Dirichlet boundary conditions are satisfied by construction using a lifting with boundary-consistent fields computed offline. This allows us to pose a version of the Empirical Cubature Method (ECM) [24,25] in strain space and generalise the Empirically Corrected Cluster Cubature (E3C) [46,48,49] as well as Empirical Material Sampling and Linearisation (EMSL) [17] beyond computational homogenisation problems. The strain-space versions of EMSL, ECM, and E3C are compared against each other and a standard displacement-space formulation of Energy Conserving Weighting and Sampling (ECSW) [15]. On two hyperelastic example problems with parameterised material behaviour and deformation, the strain-space methods outperform the displacement-space alternative in the tradeoff between runtime and accuracy. E3C and EMSL in particular facilitate 10,000 and 100,000-fold speedups, respectively, while retaining high levels of accuracy. EMSL is shown to be the method of choice when online and offline runtime budgets are very limited, while E3C yields exceptional levels of accuracy when slightly more runtime is acceptable.

Figures

Figures reproduced from arXiv: 2607.19330 by the authors.

Figure 1
Figure 1. Illustration of boundary handling on a minimal example. Degrees of free [PITH_FULL_IMAGE:figures/full_fig_p005_1.png] view at source ↗
Figure 2
Figure 2. Illustration of lifting for inhomogeneous Dirichlet boundary conditions on a [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
Figure 3
Figure 3. Illustration of a δ = 1-dimensional solution manifold (black line) in a D = 3- dimensional solution space. A d = 2-dimensional POD model (gray plane) obtained from snapshots (black dots) is also shown. Figure by the authors, originally used in [17]. and the stiffness matrix is given by Kred = ∂g red ∂y = ψ T ∂g ∂u˜ ∂u˜ ∂y = ψ TKψ , (19) such that the Newton-Raphson scheme, when projected onto the reduced space, can … view at source ↗
Figures from the paper (16 more)
Figure 4
Figure 4. Figure 4: Schematic visualisation of reduced cubature: integration point modes allow for [PITH_FULL_IMAGE:figures/full_fig_p015_4.png]
Figure 12
Figure 12. Figure 12: The centroids φ c = 1 ξ c X g∈Cc φ gV g . (41) can then be used as bases for the expected average strain in each cluster. In each load step, we can estimate the reduced variables from the parameter p using a simple least-squares model with M = Y sp sT (p sp sT ) −1 , …
Figure 5
Figure 5. Figure 5: Schematic visualisation of EMSL: firstly, an estimate of the reduced solution [PITH_FULL_IMAGE:figures/full_fig_p016_5.png]
Figure 6
Figure 6. Figure 6: Undeformed configuration of plate with two holes and deformation and stress [PITH_FULL_IMAGE:figures/full_fig_p019_6.png]
Figure 7
Figure 7. Figure 7: Training (blue plus signs) and validation (red circles) samples in material pa [PITH_FULL_IMAGE:figures/full_fig_p020_7.png]
Figure 8
Figure 8. Figure 8: Deformation at the end of tensile load path, compared to BC-consistent field, [PITH_FULL_IMAGE:figures/full_fig_p020_8.png]
Figure 9
Figure 9. Figure 9: Mean relative error over number of integration points [PITH_FULL_IMAGE:figures/full_fig_p021_9.png]
Figure 10
Figure 10. Figure 10: Relative runtime over number of integration points [PITH_FULL_IMAGE:figures/full_fig_p023_10.png]
Figure 11
Figure 11. Figure 11: Elements selected by displacement-space ECSW (shaded in red) and Gauss [PITH_FULL_IMAGE:figures/full_fig_p023_11.png]
Figure 12
Figure 12. Figure 12: Cluster association of all Gauss point, as chosen by EMSL, for [PITH_FULL_IMAGE:figures/full_fig_p024_12.png]
Figure 13
Figure 13. Figure 13: Relative error over relative runtime for displacement-space ECSW (black cir [PITH_FULL_IMAGE:figures/full_fig_p025_13.png]
Figure 14
Figure 14. Figure 14: Undeformed configuration of plate with two holes and deformation and stress [PITH_FULL_IMAGE:figures/full_fig_p026_14.png]
Figure 15
Figure 15. Figure 15: Mean relative error over number of integration points [PITH_FULL_IMAGE:figures/full_fig_p028_15.png]
Figure 16
Figure 16. Figure 16: Relative runtime over number of integration points [PITH_FULL_IMAGE:figures/full_fig_p031_16.png]
Figure 17
Figure 17. Figure 17: Relative error over relative runtime for displacement-space ECSW (black cir [PITH_FULL_IMAGE:figures/full_fig_p032_17.png]
Figure 18
Figure 18. Figure 18: Runtimes required for offline snapshot computation and offline training phases [PITH_FULL_IMAGE:figures/full_fig_p034_18.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

54 extracted references · 2 canonical work pages

  1. [1]

    Alkayas, Anup Teejo Mathew, Daniel Feliu-Talegon, Ping Deng, Thomas George Thuruthel, and Federico Renda

    Abdulaziz Y. Alkayas, Anup Teejo Mathew, Daniel Feliu-Talegon, Ping Deng, Thomas George Thuruthel, and Federico Renda. Soft synergies: Model order re- duction of hybrid soft-rigid robots via optimal strain parameterization.IEEE Trans- actions on Robotics, 41:1118–1137, 2025. doi: 10.1109/TRO.2024.3522182

    arXiv 2025

  2. [2]

    Optimizing cubature for efficient integration of subspace deformations.ACM transactions on graphics (TOG), 27(5): 1–10, 2008

    Steven S An, Theodore Kim, and Doug L James. Optimizing cubature for efficient integration of subspace deformations.ACM transactions on graphics (TOG), 27(5): 1–10, 2008. doi: 10.1145/1409060.1409118

    arXiv 2008

  3. [3]

    Neural-network-augmented projection-based model order reduction for mitigating the Kolmogorov barrier to reducibility.Journal of Computational Physics, 492:112420, November 2023

    Joshua Barnett, Charbel Farhat, and Yvon Maday. Neural-network-augmented projection-based model order reduction for mitigating the Kolmogorov barrier to reducibility.Journal of Computational Physics, 492:112420, November 2023. ISSN 00219991. doi: 10.1016/j.jcp.2023.112420

    arXiv 2023

  4. [4]

    A Survey of Projection-Based Model Reduction Methods for Parametric Dynamical Systems.SIAM Review, 57(4): 483–531, January 2015

    Peter Benner, Serkan Gugercin, and Karen Willcox. A Survey of Projection-Based Model Reduction Methods for Parametric Dynamical Systems.SIAM Review, 57(4): 483–531, January 2015. ISSN 0036-1445, 1095-7200. doi: 10.1137/130932715

    doi:10.1137/130932715 2015

  5. [5]

    Pod– galerkin reduced order methods for combined navier–stokes transport equations based on a hybrid fv-fe solver.Computers & Mathematics with Applications, 79(2):256–273, 2020

    Saray Busto, Giovanni Stabile, Gianluigi Rozza, and M Elena V´ azquez-Cend´ on. Pod– galerkin reduced order methods for combined navier–stokes transport equations based on a hybrid fv-fe solver.Computers & Mathematics with Applications, 79(2):256–273, 2020

    2020

  6. [6]

    Galerkin v

    Kevin Carlberg, Matthew Barone, and Harbir Antil. Galerkin v. least-squares Petrov– Galerkin projection in nonlinear model reduction.Journal of Computational Physics, 330:693–734, February 2017. ISSN 00219991. doi: 10.1016/j.jcp.2016.10.033

    doi:10.1016/j.jcp.2016.10.033 2017

  7. [7]

    Sorensen

    Saifon Chaturantabut and Danny C. Sorensen. Nonlinear Model Reduction via Dis- crete Empirical Interpolation.SIAM Journal on Scientific Computing, 32(5):2737– 2764, January 2010. ISSN 1064-8275, 1095-7197. doi: 10.1137/090766498

    doi:10.1137/090766498 2010

  8. [8]

    L1-Based Reduced Over Collo- cation and Hyper Reduction for Steady State and Time-Dependent Nonlinear Equa- tions.Journal of Scientific Computing, 87(1):10, April 2021

    Yanlai Chen, Lijie Ji, Akil Narayan, and Zhenli Xu. L1-Based Reduced Over Collo- cation and Hyper Reduction for Steady State and Time-Dependent Nonlinear Equa- tions.Journal of Scientific Computing, 87(1):10, April 2021. ISSN 0885-7474, 1573-

    2021

Show all 54 references
  1. [9]

    Model order reduction assisted by deep neural networks (ROM-net).Advanced Modeling and Simulation in Engineering Sciences, 7(1):16, December 2020

    Thomas Daniel, Fabien Casenave, Nissrine Akkari, and David Ryckelynck. Model order reduction assisted by deep neural networks (ROM-net).Advanced Modeling and Simulation in Engineering Sciences, 7(1):16, December 2020. ISSN 2213-7467. doi: 10.1186/s40323-020-00153-6

    2020 doi

  2. [10]

    Physics-informed cluster analysis and a priori efficiency criterion for the construction of local reduced-order bases.Journal of Computational Physics, 458: 111120, June 2022

    Thomas Daniel, Fabien Casenave, Nissrine Akkari, Ali Ketata, and David Ryck- elynck. Physics-informed cluster analysis and a priori efficiency criterion for the construction of local reduced-order bases.Journal of Computational Physics, 458: 111120, June 2022. ISSN 00219991. d...

    2022

  3. [11]

    Algorithm 832: Umfpack v4

    Timothy A Davis. Algorithm 832: Umfpack v4. 3—an unsymmetric-pattern mul- tifrontal method.ACM Transactions on Mathematical Software (TOMS), 30(2): 196–199, 2004

    2004

  4. [12]

    Nonlinear finite element analysis of solids and structures

    Ren´ e De Borst, Mike A Crisfield, Joris JC Remmers, and Clemens V Verhoosel. Nonlinear finite element analysis of solids and structures. John Wiley & Sons, 2012

    2012

  5. [13]

    Everson and L

    R. Everson and L. Sirovich. Karhunen–Lo` eve procedure for gappy data.Journal of the Optical Society of America A, 12(8):1657, August 1995. ISSN 1084-7529, 1520-

    1995

  6. [14]

    Mostafa Faghih Shojaei and Arash Yavari. Compatible-strain mixed finite ele- ment methods for 3d compressible and incompressible nonlinear elasticity.Com- puter Methods in Applied Mechanics and Engineering, 357:112610, 2019. ISSN 0045-7825. doi: https://doi.org/10.1016/j.cma.2...

    2019

  7. [15]

    Charbel Farhat, Philip Avery, Todd Chapman, and Julien Cortial. Dimensional re- duction of nonlinear finite element dynamic models with finite rotations and energy- based mesh sampling and weighting for computational efficiency.International Jour- nal for Numerical Methods in ...

    2014 doi

  8. [16]

    A hyperreduced manifold learning approach to nonlinear model order reduction for the homogenisation of hyperelastic rves.arXiv preprint arXiv:2508.21527, 2025

    Erik Faust and Lisa Scheunemann. A hyperreduced manifold learning approach to nonlinear model order reduction for the homogenisation of hyperelastic rves.arXiv preprint arXiv:2508.21527, 2025

    2025 arXiv

  9. [17]

    Erik Faust and Lisa Scheunemann. Empirical material sampling and linearisation– a simple and efficient strain-space model order reduction approach for com- putational homogenisation in large-deformation hyperelasticity.arXiv preprint arXiv:2604.27179, 2026. doi: https://doi.or...

    2026 doi arXiv

  10. [18]

    Discovering plasticity models without stress data.npj Computational Materials, 8(1):91, 2022

    Moritz Flaschel, Siddhant Kumar, and Laura De Lorenzis. Discovering plasticity models without stress data.npj Computational Materials, 8(1):91, 2022

    2022

  11. [19]

    Automated discovery of generalized standard material models with euclid.Computer Methods in Applied Mechanics and Engineering, 405:115867, 2023

    Moritz Flaschel, Siddhant Kumar, and Laura De Lorenzis. Automated discovery of generalized standard material models with euclid.Computer Methods in Applied Mechanics and Engineering, 405:115867, 2023

    2023

  12. [20]

    Two-stage data-driven homogenization for nonlinear solids using a reduced order model.European Journal of Mechanics - A/Solids, 69: 201–220, May 2018

    Felix Fritzen and Oliver Kunc. Two-stage data-driven homogenization for nonlinear solids using a reduced order model.European Journal of Mechanics - A/Solids, 69: 201–220, May 2018. ISSN 09977538. doi: 10.1016/j.euromechsol.2017.11.007. Efficient strain-space MOR38

    2018 doi

  13. [21]

    Reduced basis hybrid computational homog- enization based on a mixed incremental formulation.Computer Methods in Ap- plied Mechanics and Engineering, 260:143–154, June 2013

    Felix Fritzen and Matthias Leuschner. Reduced basis hybrid computational homog- enization based on a mixed incremental formulation.Computer Methods in Ap- plied Mechanics and Engineering, 260:143–154, June 2013. ISSN 00457825. doi: 10.1016/j.cma.2013.03.007

    2013 doi

  14. [22]

    A pod-galerkin reduced order model for a les filtering approach.Journal of Computational Physics, 436: 110260, 2021

    Michele Girfoglio, Annalisa Quaini, and Gianluigi Rozza. A pod-galerkin reduced order model for a les filtering approach.Journal of Computational Physics, 436: 110260, 2021

    2021

  15. [23]

    Sequential subspace mode adaptation for the reduced-order homogenization of dissipative microstructures using e3c hyper- reduction.arXiv preprint arXiv:2606.02089, 2026

    Hauke Goldbeck and Stephan Wulfinghoff. Sequential subspace mode adaptation for the reduced-order homogenization of dissipative microstructures using e3c hyper- reduction.arXiv preprint arXiv:2606.02089, 2026

    2026 arXiv

  16. [24]

    A multiscale method for periodic structures using domain decompo- sition and ecm-hyperreduction.Computer Methods in Applied Mechanics and Engi- neering, 368:113192, 2020

    JA Hern´ andez. A multiscale method for periodic structures using domain decompo- sition and ecm-hyperreduction.Computer Methods in Applied Mechanics and Engi- neering, 368:113192, 2020

    2020

  17. [25]

    Hern´ andez, M.A

    J.A. Hern´ andez, M.A. Caicedo, and A. Ferrer. Dimensional hyper-reduction of nonlinear finite element models via empirical cubature.Computer Methods in Ap- plied Mechanics and Engineering, 313:687–722, January 2017. ISSN 00457825. doi: 10.1016/j.cma.2016.10.022

    2017 doi

  18. [26]

    JA Hern´ andez, JR Bravo, and S Ares de Parga. Cecm: A continuous empirical cu- bature method with application to the dimensional hyperreduction of parameterized finite element models.Computer Methods in Applied Mechanics and Engineering, 418:116552, 2024

    2024

  19. [27]

    A complement to neural networks for anisotropic inelasticity at finite strains.Computer Methods in Applied Mechanics and Engineer- ing, 450:118612, 2026

    Hagen Holthusen and Ellen Kuhl. A complement to neural networks for anisotropic inelasticity at finite strains.Computer Methods in Applied Mechanics and Engineer- ing, 450:118612, 2026

    2026

  20. [28]

    Holzapfel

    Gerhard A. Holzapfel. Nonlinear Solid Mechanics: A Continuum Approach for En- gineering Science.Meccanica, 37(4):489–490, July 2002. ISSN 1572-9648. doi: 10.1023/A:1020843529530

    2002 doi

  21. [29]

    Finite Strain Homogenization Using a Reduced Basis and Efficient Sampling.Mathematical and Computational Applications, 24(2):56, May 2019

    Oliver Kunc and Felix Fritzen. Finite Strain Homogenization Using a Reduced Basis and Efficient Sampling.Mathematical and Computational Applications, 24(2):56, May 2019. ISSN 2297-8747. doi: 10.3390/mca24020056

    2019 doi

  22. [30]

    Bessa, and Wing Kam Liu

    Zeliang Liu, M.A. Bessa, and Wing Kam Liu. Self-consistent clustering analysis: An efficient multi-scale scheme for inelastic heterogeneous materials.Computer Methods in Applied Mechanics and Engineering, 306:319–341, July 2016. ISSN 00457825. doi: 10.1016/j.cma.2016.04.004

    2016 doi

  23. [31]

    An efficient mixed variational reduced-order model formulation for nonlinear analyses of elastic shells

    D Magisano, Ke Liang, G Garcea, L Leonetti, and Martin Ruess. An efficient mixed variational reduced-order model formulation for nonlinear analyses of elastic shells. International Journal for Numerical Methods in Engineering, 113(4):634–655, 2018

    2018

  24. [32]

    Michel and P

    J.C. Michel and P. Suquet. Nonuniform transformation field analysis.International Journal of Solids and Structures, 40(25):6937–6955, December 2003. ISSN 00207683. doi: 10.1016/S0020-7683(03)00346-9. Erik Faust & Lisa Scheunemann39

    2003 doi

  25. [33]

    Karl Pearson. LIII.On lines and planes of closest fit to systems of points in space.The London, Edinburgh, and Dublin Philosophical Magazine and Jour- nal of Science, 2(11):559–572, November 1901. ISSN 1941-5982, 1941-5990. doi: 10.1080/14786440109462720

    1901 doi

  26. [34]

    Breaking the Kolmogorov Barrier with Nonlinear Model Reduction.Notices of the American Mathematical Society, 69(05):1, May 2022

    Benjamin Peherstorfer. Breaking the Kolmogorov Barrier with Nonlinear Model Reduction.Notices of the American Mathematical Society, 69(05):1, May 2022. ISSN 0002-9920, 1088-9477. doi: 10.1090/noti2475

    2022 doi

  27. [35]

    Model reduction in elastoplasticity: Proper orthogonal decomposition combined with adaptive sub-structuring.Computational Mechanics, 54(3):677–687, September 2014

    Annika Radermacher and Stefanie Reese. Model reduction in elastoplasticity: Proper orthogonal decomposition combined with adaptive sub-structuring.Computational Mechanics, 54(3):677–687, September 2014. ISSN 0178-7675, 1432-0924. doi: 10. 1007/s00466-014-1020-6

    2014

  28. [36]

    A clustering-enhanced potential-based reduced order homogenization framework for nonlinear heterogeneous materials.European Journal of Mechanics - A/Solids, 103:105190, 2024

    Hongshi Ruan, Xiaozhe Ju, Junjun Chen, Lihua Liang, and Yangjian Xu. A clustering-enhanced potential-based reduced order homogenization framework for nonlinear heterogeneous materials.European Journal of Mechanics - A/Solids, 103:105190, 2024. ISSN 0997-7538. doi: https://doi....

    2024

  29. [37]

    Aspects of Computational Homoge- nization at Finite Deformations: A Unifying Review From Reuss’ to Voigt’s Bound

    Saba Saeb, Paul Steinmann, and Ali Javili. Aspects of Computational Homoge- nization at Finite Deformations: A Unifying Review From Reuss’ to Voigt’s Bound. Applied Mechanics Reviews, 68(5):050801, September 2016. ISSN 0003-6900, 2379-

    2016

  30. [38]

    Lisa Scheunemann and Erik Faust. A manifold learning approach to nonlinear model order reduction of quasi-static problems in solid mechanics: proof of concept and parameter study.Computational Mechanics, pages 1–41, 2025

    2025

  31. [39]

    Habilitation, Universit¨ at Stuffgart, 2000

    J¨ org Schr¨ oder.Homogenisierungsmethoden Der Nichtlinearen Kontinuumsmechanik Unter Beachtung von Stabilit¨ atsproblemen. Habilitation, Universit¨ at Stuffgart, 2000

    2000

  32. [40]

    A Numerical Two-Scale Homogenization Scheme: The FE2-Method

    J¨ org Schr¨ oder. A Numerical Two-Scale Homogenization Scheme: The FE2-Method. InPlasticity and beyond: Microstructures, Crystal-Plasticity and Phase Transitions, number Vol. 550 in Courses and Lectures, pages 1–64. Springer, Wien, 2014. ISBN 978-3-7091-1624-1

    2014

  33. [41]

    Giovanni Stabile, Saddam Hijazi, Andrea Mola, Stefano Lorenzi, Gianluigi Rozza, et al. Pod-galerkin reduced order methods for cfd using finite volume discretisa- tion: vortex shedding around a circular cylinder.Communications in Applied and Industrial Mathematics, 8(1):210–236, 2017

    2017

  34. [42]

    Prakash Thakolkaran, Yaqi Guo, Shivam Saini, Mathias Peirlinck, Benjamin Al- heit, and Siddhant Kumar. Can kan cans? input-convex kolmogorov-arnold networks (kans) as hyperelastic constitutive artificial neural networks (cans).Computer Meth- ods in Applied Mechanics and Engine...

    2025

  35. [43]

    A Tutorial on the Proper Orthogonal Decomposition

    Julien Weiss. A Tutorial on the Proper Orthogonal Decomposition. InAIAA Aviation 2019 Forum, page 21, Dallas, Texas, June 2019. American Institute of Aeronautics and Astronautics. ISBN 978-1-62410-589-0. doi: 10.2514/6.2019-3333. Efficient strain-space MOR40

    2019 doi

  36. [44]

    Unsteady flow sensing and estimation via the gappy proper orthogonal decomposition

    K Willcox. Unsteady flow sensing and estimation via the gappy proper orthogonal decomposition. 2006

    2006

  37. [45]

    Wriggers.Nonlinear Finite Element Methods

    P. Wriggers.Nonlinear Finite Element Methods. Springer, Berlin, 2008. ISBN 978- 3-540-71000-4 978-3-540-71001-1

    2008

  38. [46]

    Empirically corrected cluster cubature (e3c).Computer Meth- ods in Applied Mechanics and Engineering, 437:117779, 2025

    Stephan Wulfinghoff. Empirically corrected cluster cubature (e3c).Computer Meth- ods in Applied Mechanics and Engineering, 437:117779, 2025

    2025

  39. [47]

    Stephan Wulfinghoff. Computational crystal plasticity homogenization using empiri- cally corrected cluster cubature (e3c) hyper-reduction.Computer Methods in Applied Mechanics and Engineering, 452:118681, 2026

    2026

  40. [48]

    E3c hyper-reduction for computational ho- mogenization in nonlinear mechanics.Computational Mechanics, 76(6):1563–1572, 2025

    Stephan Wulfinghoff and Jan Hauck. E3c hyper-reduction for computational ho- mogenization in nonlinear mechanics.Computational Mechanics, 76(6):1563–1572, 2025

    2025

  41. [49]

    E3C for Computational Homogenization in Nonlinear Mechanics, January 2025

    Stephan Wulfinghoff and Jan Hauck. E3C for Computational Homogenization in Nonlinear Mechanics, January 2025

    2025

  42. [50]

    Kafka, and Wing Kam Liu

    Cheng Yu, Orion L. Kafka, and Wing Kam Liu. Self-consistent clustering analysis for multiscale modeling at finite strains.Computer Methods in Applied Mechanics and Engineering, 349:339–359, 2019. ISSN 0045-7825. doi: https://doi.org/10.1016/j. cma.2019.02.027. URLhttps://www.s...

    2019 doi

  43. [51]

    Springer, 2018

    Tarek Zohdi.Finite element primer for beginners. Springer, 2018. A Timing data |Em| 1 2 5 10 20 30 50 75 100 d 5 0.006084× ×0.05071 0.1004 0.1509 0.2502 0.3671 0.4881 7× ×0.02802 0.05159 0.1002 0.1511 0.2508 0.3752 0.4965 10× ×0.02735 0.05216 0.1021 0.1515 0.2504 0.3752 0.5421...

    2018

  44. [407]

    doi: 10.1115/1.4034024

    doi

  45. [7691]

    Erik Faust & Lisa Scheunemann37

    doi: 10.1007/s10915-021-01416-z. Erik Faust & Lisa Scheunemann37

    doi

  46. [8532]

    doi: 10.1364/JOSAA.12.001657

    doi

Pith tools

Reviewed August 1, 2026 · model on record in the stance chip above.