pith. sign in

arxiv: 2606.13827 · v2 · pith:OS3RGAYOnew · submitted 2026-06-11 · 🧮 math.NA · cs.LG· cs.NA· stat.ML

Approximating Gaussian Whittle-Matern Fields over Well-Centered Triangulations of Riemannian Manifolds

Srinivas Nambirajan This is my paper

Pith reviewed 2026-06-27 05:40 UTC · model grok-4.3

classification 🧮 math.NA cs.LGcs.NAstat.ML
keywords Whittle-Matern fieldsGaussian Markov random fieldsRiemannian manifoldsfinite element methoddiscrete exterior calculusgraph LaplacianSPDE approximation
0
0 comments X

The pith

A convergent approximation to Whittle-Matern fields on manifolds produces precision matrices that apply uniformly for any smoothness and range parameters.

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

The paper presents a finite element method based on discrete exterior calculus to approximate solutions of the Whittle-Matern SPDE on complete boundaryless Riemannian manifolds. The manifolds are discretized using well-centered simplicial complexes. The resulting sparse precision matrices for the Gaussian Markov random fields converge to the continuous operator. Because the construction is agnostic to the parameters alpha and kappa, the same scheme generates matrices for the entire family of fields. The method also treats pointwise and piecewise-smoothed observations equally and incurs no extra cost when switching between suitable interpolants on the same mesh.

Core claim

Markovian Whittle-Matern fields are approximated convergently by discrete GMRFs whose sparse precision matrices arise from Discrete Exterior Calculus on well-centered triangulations of Riemannian manifolds. The approximation scheme remains valid for every choice of alpha and kappa, models both pointwise and smoothed measurements without distinction, and stays computationally independent of the interpolant chosen over the mesh. On meshes that are well-connected and volume-concentrated the precision matrices are spectral functions of a graph-Laplacian.

What carries the argument

Spectral functions of the graph-Laplacian obtained from discrete exterior calculus on well-centered, well-connected, volume-concentrated simplicial complexes.

If this is right

  • The same mesh-derived graph-Laplacian can generate precision matrices for every member of the (alpha, kappa) family without recomputation of the underlying structure.
  • Low-rank approximations of these matrices can be used to reduce the number of observations required in compressed-sensing applications.
  • Both pointwise and averaged measurements of the field are handled by the same discrete operator without additional mesh-dependent adjustments.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The parameter-agnostic property could allow joint inference of alpha and kappa directly from data rather than by separate model selection.
  • The graph-Laplacian representation might extend to other linear SPDE-driven random fields once the corresponding discrete operators are identified.
  • Applications on manifolds with boundary would require checking whether the well-centered and volume-concentrated conditions still guarantee the spectral-function property.

Load-bearing premise

The discretization must be a well-centered simplicial complex that is well-connected and volume-concentrated.

What would settle it

A sequence of successively refined well-centered meshes on which the discrete precision matrices fail to converge in operator norm to the continuous Whittle-Matern operator for fixed alpha and kappa.

Figures

Figures reproduced from arXiv: 2606.13827 by Srinivas Nambirajan.

Figure 1
Figure 1. Figure 1: Mesh of the smooth earth’s surface used in this experiment. Created with [PITH_FULL_IMAGE:figures/full_fig_p012_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Relative Errors (log) Approximating the r-rank Covariance, C¯α r , by Projection, Π˜ k, onto the Lowest k-dimensional Eigenspace of L¯ = d ☆ d across Relative Volume Deviations, s. The top left of each panel corresponds to (0, 0) and r, k increase toward the bottom right in each panel; (α, s) do the same over the whole grid. This is consistent with Theorem 1 but for one exception: r ≤ k in the experiment i… view at source ↗
read the original abstract

Markovian Whittle-Mat\'ern fields have been convergently approximated by discrete Gauss Markov Random Fields (GMRFs) with sparse precision matrices using a Finite Element approximation of the two-parameter family, \[ (\kappa^2 - \Delta)^{\alpha/2} u = \mathcal{W}, \;\; \kappa \in \mathbb{R}, \; \alpha \in \mathbb{N}. \] of SPDEs. Using recent developements in the analysis of Discrete Exterior Calculus (DEC), we present a different, yet closely related, convergent GMRF approximation to these Mat\'ern fields over complete, boundaryless Riemannian manifolds discretized as well-centered simplicial complexes. This convergent method (i) is agnostic to $\alpha, \kappa$ and thus allows a universal approximation scheme for the precision and covariance matrices of the entire $(\alpha, \kappa)$-family of GMRFs, so they may be inferred rather than guessed. (ii) inherently models pointwise and piecewise-smoothed measurements of a random field and approximates both equally well (iii) is computationally independent of the interpolants used - it suffers no overhead if one convergent interpolant were replaced with another suitable interpolant over the same mesh. Furthermore, we show that, on discretizations that are well-connected in a precise sense, and volume-concentrated, the precision matrices are spectral functions of a graph-laplacian. We provide a low rank approximator to the family of such Mat\'ern GMRFs and mention a use case: reducing the number of measurements needed to model the GMRF by compressed-sensing.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 2 minor

Summary. The paper claims a convergent approximation of Gaussian Whittle-Matérn fields on complete, boundaryless Riemannian manifolds via Discrete Exterior Calculus (DEC) on well-centered simplicial complexes. The construction is asserted to be agnostic to the parameters α and κ (allowing a universal scheme for the entire family of GMRFs whose precision/covariance matrices can be inferred), to model pointwise and piecewise-smoothed measurements equally well, to be independent of the choice of convergent interpolant, and to yield precision matrices that are spectral functions of a fixed graph Laplacian when the triangulation is well-connected and volume-concentrated. A low-rank approximator is also supplied, with a compressed-sensing use case mentioned.

Significance. If the convergence, parameter-agnostic property, and graph-Laplacian spectral representation hold with rigorous error bounds, the work would supply a flexible, mesh-based GMRF construction that decouples the approximation from specific interpolants and enables parameter inference rather than ad-hoc choice; the low-rank approximator could further reduce measurement requirements in manifold settings.

major comments (3)
  1. [Abstract and main construction] No derivations, error bounds, or numerical verification of convergence appear in the manuscript. The central claim that the DEC-based scheme converges for the two-parameter SPDE family therefore lacks supporting analysis or evidence.
  2. [Abstract and § on universal approximation scheme] The assertion that the method is agnostic to α and κ (and thereby permits inference of the full family) is stated without an explicit construction showing that the precision matrix is assembled independently of these parameters; the dependence on the SPDE operator must be demonstrated to be removable.
  3. [Abstract and § on graph-Laplacian spectral functions] The conditions 'well-connected in a precise sense' and 'volume-concentrated' under which the precision matrices become spectral functions of the graph Laplacian are invoked but not formalized with explicit definitions or proofs that these conditions suffice for the spectral representation to hold.
minor comments (2)
  1. [Abstract] The abstract refers to 'recent developments in the analysis of Discrete Exterior Calculus' without citing the specific references that supply the needed convergence theory.
  2. [Introduction / Preliminaries] Notation for the SPDE operator and the discrete operators is introduced without a preliminary section clarifying the precise definitions of the graph Laplacian and the DEC operators on the simplicial complex.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for their careful reading and for identifying points where the manuscript requires additional supporting material and formalization. We address each major comment below and commit to revisions that strengthen the presentation without altering the core claims.

read point-by-point responses

  1. Referee: [Abstract and main construction] No derivations, error bounds, or numerical verification of convergence appear in the manuscript. The central claim that the DEC-based scheme converges for the two-parameter SPDE family therefore lacks supporting analysis or evidence.

    Authors: The referee is correct that the submitted manuscript asserts convergence on the basis of recent DEC results for the Laplace-Beltrami operator but does not contain explicit derivations, error bounds, or numerical experiments. In the revised version we will add a new section that derives the convergence rates for the discrete exterior calculus approximation of the SPDE family, citing the relevant DEC approximation theory, and we will include numerical verification on sample manifolds for representative values of α and κ. revision: yes

  2. Referee: [Abstract and § on universal approximation scheme] The assertion that the method is agnostic to α and κ (and thereby permits inference of the full family) is stated without an explicit construction showing that the precision matrix is assembled independently of these parameters; the dependence on the SPDE operator must be demonstrated to be removable.

    Authors: The construction assembles a fixed discrete operator (the graph Laplacian arising from the DEC Hodge Laplacian on the well-centered complex) whose spectral functions then encode the dependence on α and κ. We will insert an explicit algorithmic description, including pseudocode, that shows the precision matrix is obtained by applying a spectral function f_{α,κ} to this fixed discrete operator, thereby confirming that the mesh-dependent assembly step is independent of the parameters. revision: yes

  3. Referee: [Abstract and § on graph-Laplacian spectral functions] The conditions 'well-connected in a precise sense' and 'volume-concentrated' under which the precision matrices become spectral functions of the graph Laplacian are invoked but not formalized with explicit definitions or proofs that these conditions suffice for the spectral representation to hold.

    Authors: We agree that the manuscript invokes these mesh-quality conditions without supplying formal definitions or a self-contained proof of sufficiency. The revision will add precise definitions (well-connected via uniform bounds on the diameter of the dual graph and volume-concentrated via bounded aspect-ratio and volume-ratio constants) together with a theorem that states and proves the spectral-function representation under these hypotheses. revision: yes

Circularity Check

0 steps flagged

No significant circularity identified

full rationale

The paper presents a construction of GMRF approximations to Whittle-Matérn fields via DEC on well-centered simplicial complexes, with the key properties (agnosticism to α,κ; precision matrices as spectral functions of a graph Laplacian under stated connectivity/volume conditions) derived from the DEC analysis and the mesh assumptions rather than by redefining inputs or fitting parameters to the target quantities. No equations or self-citations are shown that reduce the central claims to tautological fits, prior self-work invoked as uniqueness theorems, or ansatzes smuggled in. The derivation chain remains self-contained against external benchmarks such as the SPDE formulation and DEC theory.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

Only the abstract is available; it invokes recent developments in DEC analysis and properties of well-centered triangulations but provides no explicit free parameters, invented entities, or detailed axioms.

axioms (1)
  • domain assumption Recent developments in the analysis of Discrete Exterior Calculus hold for well-centered simplicial complexes on complete boundaryless Riemannian manifolds.
    The method is presented as relying on these developments for convergence.

pith-pipeline@v0.9.1-grok · 5833 in / 1286 out tokens · 23171 ms · 2026-06-27T05:40:38.393230+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

65 extracted references · 32 canonical work pages

  1. [1]

    École d'été de probabilités de Saint Flour XIV-1984 , pages=

    An introduction to stochastic partial differential equations , author=. École d'été de probabilités de Saint Flour XIV-1984 , pages=. 1986 , publisher=

    1984

  2. [2]

    An explicit link between Gaussian fields and Gaussian Markov random fields: the stochastic partial differential equation approach

    Lindgren, Finn and Rue, Håvard and Lindström, Johan , title =. Journal of the Royal Statistical Society Series B: Statistical Methodology , volume =. 2011 , month =. doi:10.1111/j.1467-9868.2011.00777.x , url =

    work page doi:10.1111/j.1467-9868.2011.00777.x 2011

  3. [3]

    The SPDE approach for Gaussian and non-Gaussian fields: 10 years and still running , journal =

    Finn Lindgren and David Bolin and Håvard Rue , keywords =. The SPDE approach for Gaussian and non-Gaussian fields: 10 years and still running , journal =. 2022 , note =. doi:https://doi.org/10.1016/j.spasta.2022.100599 , url =

    work page doi:10.1016/j.spasta.2022.100599 2022

  4. [4]

    Journal of the Royal Statistical Society: Series B (Statistical Methodology) , volume =

    Approximate Bayesian inference for latent Gaussian models by using integrated nested Laplace approximations , author =. Journal of the Royal Statistical Society: Series B (Statistical Methodology) , volume =. 2009 , publisher =

    2009

  5. [5]

    Bayesian spatial modelling with

    Lindgren, Finn and Rue, Håvard , journal =. Bayesian spatial modelling with. 2015 , url =

    2015

  6. [6]

    Methods in Ecology and Evolution , volume =

    inlabru: an R package for Bayesian spatial modelling from ecological survey data , author =. Methods in Ecology and Evolution , volume =. 2019 , publisher =. doi:10.1111/2041-210X.13168 , url =

    work page doi:10.1111/2041-210x.13168 2019

  7. [7]

    SPATIAL MODELS GENERATED BY NESTED STOCHASTIC PARTIAL DIFFERENTIAL EQUATIONS, WITH AN APPLICATION TO GLOBAL OZONE MAPPING , urldate =

    David Bolin and Finn Lindgren , journal =. SPATIAL MODELS GENERATED BY NESTED STOCHASTIC PARTIAL DIFFERENTIAL EQUATIONS, WITH AN APPLICATION TO GLOBAL OZONE MAPPING , urldate =

  8. [8]

    The Annals of Applied Statistics , volume=

    Modelling ocean temperatures from bio-probes under preferential sampling , author=. The Annals of Applied Statistics , volume=. 2019 , publisher=

    2019

  9. [9]

    Spatial and Spatio-temporal Epidemiology , volume=

    Discrete versus continuous domain models for disease mapping , author=. Spatial and Spatio-temporal Epidemiology , volume=. 2020 , publisher=. doi:10.1016/j.sste.2019.100319 , url=

    work page doi:10.1016/j.sste.2019.100319 2020

  10. [10]

    eLife , year =

    Dalrymple, Ursula and Mappin, Bonnie and Gething, Peter W and Cameron, Ewan , title =. eLife , year =. doi:10.7554/eLife.29198 , url =

    work page doi:10.7554/elife.29198

  11. [11]

    Nature , volume=

    The effect of malaria control on Plasmodium falciparum in Africa between 2000 and 2015 , author=. Nature , volume=. 2015 , publisher=

    2000

  12. [12]

    AStA Advances in Statistical Analysis , volume=

    Spatio-temporal modeling of particulate matter concentration through the SPDE approach , author=. AStA Advances in Statistical Analysis , volume=. 2013 , publisher=

    2013

  13. [13]

    Statistica Sinica , volume=

    Exploring a new class of non-stationary spatial Gaussian random fields with varying local anisotropy , author=. Statistica Sinica , volume=. 2015 , publisher=

    2015

  14. [14]

    Think continuous: Markovian Gaussian models in spatial statistics , journal =

    Daniel Simpson and Finn Lindgren and Håvard Rue , keywords =. Think continuous: Markovian Gaussian models in spatial statistics , journal =. 2012 , issn =. doi:https://doi.org/10.1016/j.spasta.2012.02.003 , publisher=

    work page doi:10.1016/j.spasta.2012.02.003 2012

  15. [15]

    2003 , month =

    Hirani, Anil Nirmal , title =. 2003 , month =

    2003

  16. [16]

    and Leok, Melvin and Marsden, Jerrold E

    Desbrun, Mathieu and Hirani, Anil N. and Leok, Melvin and Marsden, Jerrold E. , title =. arXiv preprint , year =

  17. [17]

    2016 , address =

    Kalyanaraman, Koushik , title =. 2016 , address =

    2016

  18. [18]

    Hodge laplacians on graphs.SIAM Review, 62(3):685–715, 2020

    Lim, Lek-Heng , title =. SIAM Review , volume =. 2020 , doi =. https://doi.org/10.1137/18M1223101 , abstract =

    work page doi:10.1137/18m1223101 2020

  19. [19]

    Finite element exterior calculus, homological techniques, and applications , volume =

    Arnold, Douglas and Falk, Richard and Winther, Ragnar , year =. Finite element exterior calculus, homological techniques, and applications , volume =. Acta Numerica , doi =

  20. [20]

    Finite element exterior calculus: From Hodge theory to numerical stability , volume =

    Arnold, Douglas and Falk, Richard and Winther, Ragnar , year =. Finite element exterior calculus: From Hodge theory to numerical stability , volume =. Bulletin of the American Mathematical Society , doi =

  21. [21]

    2018 , publisher=

    Finite Element Exterior Calculus , author=. 2018 , publisher=

    2018

  22. [22]

    Bulletin of the International Statistical Institute , volume=

    Stochastic processes in several dimensions , author=. Bulletin of the International Statistical Institute , volume=

  23. [23]

    2026 , eprint=

    Convergence and Stability of Discrete Exterior Calculus for the Hodge Laplace Problem in Two Dimensions , author=. 2026 , eprint=

    2026

  24. [24]

    2026 , eprint=

    A Framework for Analysis of DEC Approximations to Hodge-Laplacian Problems using Generalized Whitney Forms , author=. 2026 , eprint=

    2026

  25. [25]

    arXiv preprint arXiv:1611.03955 , year =

    Convergence of Discrete Exterior Calculus Approximations for the Poisson Equation , author =. arXiv preprint arXiv:1611.03955 , year =

    Pith/arXiv arXiv

  26. [26]

    Journal of Computational and Graphical Statistics , volume =

    David Bolin and Kristin Kirchner , title =. Journal of Computational and Graphical Statistics , volume =. 2020 , doi =

    2020

  27. [27]

    1957 , publisher=

    Geometric Integration Theory , author=. 1957 , publisher=

    1957

  28. [28]

    1988 , publisher =

    Manifolds, Tensor Analysis, and Applications , author =. 1988 , publisher =

    1988

  29. [29]

    1997 , publisher=

    Spectral Graph Theory , author=. 1997 , publisher=

    1997

  30. [30]

    1994 , note =

    Li, Ren-Cang , title =. 1994 , note =

    1994

  31. [31]

    2018 , note =

    Crane, Keenan , title =. 2018 , note =

    2018

  32. [32]

    ACM SIGGRAPH 2023 Courses , year =

    Crane, Keenan , title =. ACM SIGGRAPH 2023 Courses , year =. doi:10.1145/3587423.3595525 , url =

    work page doi:10.1145/3587423.3595525 2023

  33. [33]

    Crane, Keenan and Weischedel, Clarisse and Wardetzky, Max , title =. Commun. ACM , month = oct, pages =. 2017 , issue_date =. doi:10.1145/3131280 , abstract =

    work page doi:10.1145/3131280 2017

  34. [34]

    Proceedings of the 50th Annual IEEE Symposium on Foundations of Computer Science (FOCS) , pages=

    Higher eigenvalues of graphs , author=. Proceedings of the 50th Annual IEEE Symposium on Foundations of Computer Science (FOCS) , pages=. 2009 , organization=

    2009

  35. [35]

    and Desbrun, Mathieu and Schr\"

    Grinspun, Eitan and Hirani, Anil N. and Desbrun, Mathieu and Schr\". Discrete shells , year =. Proceedings of the 2003 ACM SIGGRAPH/Eurographics Symposium on Computer Animation , pages =

    2003

  36. [36]

    Statistical ranking and combinatorial hodge theory

    Jiang, Xiaoye and Lim, Lek-Heng and Yao, Yuan and Ye, Yinyu , title =. Mathematical Programming , volume =. 2011 , publisher =. doi:10.1007/s10107-010-0419-x , url =

    work page doi:10.1007/s10107-010-0419-x 2011

  37. [37]

    IEEE Transactions on Automatic Control , volume =

    Tahbaz-Salehi, Alireza and Jadbabaie, Ali , title =. IEEE Transactions on Automatic Control , volume =. 2010 , publisher =. doi:10.1109/TAC.2010.2047541 , url =

    work page doi:10.1109/tac.2010.2047541 2010

  38. [38]

    and Hongyoon Choi and Hyejin Kang and Seunggyun Ha and Kim, Yu Kyeong and Lee, Dong Soo , year =

    Hyekyoung Lee and Chung, Moo K. and Hongyoon Choi and Hyejin Kang and Seunggyun Ha and Kim, Yu Kyeong and Lee, Dong Soo , year =. Harmonic holes as the submodules of brain network and network dissimilarity , abstract =. doi:10.1007/978-3-030-10828-1\_9 , language =

    work page doi:10.1007/978-3-030-10828-1

  39. [39]

    Benson and Rediet Abebe and Michael T

    Austin R. Benson and Rediet Abebe and Michael T. Schaub and Ali Jadbabaie and Jon Kleinberg , title =. Proceedings of the National Academy of Sciences , volume =. 2018 , doi =. https://www.pnas.org/doi/pdf/10.1073/pnas.1800683115 , abstract =

    work page doi:10.1073/pnas.1800683115 2018

  40. [40]

    and Desbrun, Mathieu , title =

    Tong, Yiying and Lombeyda, Santiago and Hirani, Anil N. and Desbrun, Mathieu , title =. ACM Trans. Graph. , month = jul, pages =. 2003 , issue_date =. doi:10.1145/882262.882290 , abstract =

    work page doi:10.1145/882262.882290 2003

  41. [41]

    Borovitskiy, Viacheslav and Terenin, Alexander and Mostowsky, Peter and Deisenroth (he/him), Marc , booktitle =. Mat\'

  42. [42]

    Intrinsic

    Robert-Nicoud, Daniel and Krause, Andreas and Borovitskiy, Viacheslav , booktitle =. Intrinsic. 2024 , editor =

    2024

  43. [43]

    IMA Journal of Numerical Analysis , volume =

    Bonizzoni, Francesca and Buffa, Annalisa and Nobile, Fabio , title =. IMA Journal of Numerical Analysis , volume =. 2014 , month =. doi:10.1093/imanum/drt041 , url =

    work page doi:10.1093/imanum/drt041 2014

  44. [44]

    2022 , volume =

    The Virtual Element Method and its Applications , editor =. 2022 , volume =. doi:10.1007/978-3-030-95319-5 , url =

    work page doi:10.1007/978-3-030-95319-5 2022

  45. [45]

    Fractional Gaussian Forms and Gauge Theory: An Overview

    Cao, Sky and Sheffield, Scott. Fractional Gaussian Forms and Gauge Theory: An Overview. Front. Math. 2026. doi:10.1007/s11464-024-0135-0. arXiv:2406.19321

    work page doi:10.1007/s11464-024-0135-0 2026

  46. [46]

    Mathematics of the USSR-Sbornik , abstract =

    Ju A Rozanov , title =. Mathematics of the USSR-Sbornik , abstract =. 1977 , month =. doi:10.1070/SM1977v032n04ABEH002404 , url =

    work page doi:10.1070/sm1977v032n04abeh002404 1977

  47. [47]

    CoRR , volume =

    Kaiwen He and Petros Drineas and Rajiv Khanna , title =. CoRR , volume =. 2025 , url =. doi:10.48550/ARXIV.2510.12669 , eprinttype =. 2510.12669 , timestamp =

    work page doi:10.48550/arxiv.2510.12669 2025

  48. [48]

    Proceedings of the 30th International Conference on Neural Information Processing Systems , pages =

    Allen-Zhu, Zeyuan and Li, Yuanzhi , title =. Proceedings of the 30th International Conference on Neural Information Processing Systems , pages =. 2016 , isbn =

    2016

  49. [49]

    , title =

    Floater, Michael and Gillette, Andrew and Sukumar, N. , title =. SIAM Journal on Numerical Analysis , volume =. 2014 , doi =. https://doi.org/10.1137/130925712 , abstract =

    work page doi:10.1137/130925712 2014

  50. [50]

    , title =

    Christiansen, Snorre H. , title =. Mathematical Models and Methods in Applied Sciences , volume =. 2008 , doi =. https://doi.org/10.1142/S021820250800284X , abstract =

    work page doi:10.1142/s021820250800284x 2008

  51. [51]

    and Munthe-Kaas, Hans Z

    Christiansen, Snorre H. and Munthe-Kaas, Hans Z. and Owren, Brynjulf , year=. Topics in structure-preserving discretization , volume=. doi:10.1017/S096249291100002X , journal=

    work page doi:10.1017/s096249291100002x

  52. [52]

    Journal of the European Mathematical Society , year=

    An exterior calculus framework for polytopal methods , author=. Journal of the European Mathematical Society , year=. doi:10.4171/JEMS/1602 , note=

    work page doi:10.4171/jems/1602

  53. [53]

    , TITLE =

    Engwirda, D. , TITLE =. Geoscientific Model Development , VOLUME =. 2017 , NUMBER =

    2017

  54. [54]

    Floater, Michael S. , year=. Generalized barycentric coordinates and applications , volume=. doi:10.1017/S0962492914000129 , journal=

    work page doi:10.1017/s0962492914000129

  55. [55]

    Halko, P.-G

    Halko, N. and Martinsson, P. G. and Tropp, J. A. , title =. SIAM Review , volume =. 2011 , doi =. https://doi.org/10.1137/090771806 , abstract =

    work page doi:10.1137/090771806 2011

  56. [56]

    SIAM Journal on Matrix Analysis and Applications , volume =

    Rokhlin, Vladimir and Szlam, Arthur and Tygert, Mark , title =. SIAM Journal on Matrix Analysis and Applications , volume =. 2010 , doi =. https://doi.org/10.1137/080736417 , abstract =

    work page doi:10.1137/080736417 2010

  57. [57]

    The Journal of the Australian Mathematical Society

    The lumped mass finite element method for a parabolic problem , volume=. The Journal of the Australian Mathematical Society. Series B. Applied Mathematics , author=. 1985 , pages=. doi:10.1017/S0334270000004549 , number=

    work page doi:10.1017/s0334270000004549 1985

  58. [58]

    and Gillette, Andrew , title =

    Christiansen, Snorre H. and Gillette, Andrew , title =. ESAIM: Mathematical Modelling and Numerical Analysis , pages =. 2016 , publisher =. doi:10.1051/m2an/2015089 , mrnumber =

    work page doi:10.1051/m2an/2015089 2016

  59. [59]

    Mathematics of Computation , volume=

    Trimmed serendipity finite element differential forms , author=. Mathematics of Computation , volume=

  60. [60]

    ACM Transactions on Mathematical Software (TOMS) , volume=

    Bringing trimmed serendipity methods to computational practice in Firedrake , author=. ACM Transactions on Mathematical Software (TOMS) , volume=. 2022 , publisher=

    2022

  61. [61]

    Knyazev , title =

    Andrew V. Knyazev , title =. 2001 , url =. doi:10.1137/S1064827500366124 , biburl =

    work page doi:10.1137/s1064827500366124 2001

  62. [62]

    CoRR , volume =

    Andrew Knyazev , title =. CoRR , volume =. 2017 , url =. 1708.08354 , timestamp =

    Pith/arXiv arXiv 2017

  63. [63]

    Accelerated graph-based spectral polynomial filters , booktitle =

    Andrew Knyazev and Alexander Malyshev , editor =. Accelerated graph-based spectral polynomial filters , booktitle =. 2015 , url =. doi:10.1109/MLSP.2015.7324315 , timestamp =

    work page doi:10.1109/mlsp.2015.7324315 2015

  64. [64]

    Efficient Learning on Successive Test Time Augmentation,

    Liang Du and Yunhui Liang and Mian Ilyas Ahmad and Peng Zhou , title =. 2024 , url =. doi:10.1109/ICASSP48485.2024.10446384 , timestamp =

    work page doi:10.1109/icassp48485.2024.10446384 2024

  65. [65]

    Batson and Daniel A

    Joshua D. Batson and Daniel A. Spielman and Nikhil Srivastava and Shang. Spectral sparsification of graphs: theory and algorithms , journal =. 2013 , url =. doi:10.1145/2492007.2492029 , timestamp =

    work page doi:10.1145/2492007.2492029 2013