Bivariate polynomial histopolation techniques on Padua, Fekete and Leja triangles

Federico Nudo; Francesco Dell'Accio; Ludovico Bruni Bruno; Wolfgang Erb

REVIEW 3 major objections 5 minor 42 references

Bivariate polynomial histopolation techniques on Padua, Fekete and Leja triangles

T0 review · 3 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read On sufficiently fine regular meshes, reconstructing a bivariate function from triangle averages is as stable as interpolating at Padua points, with a Lebesgue constant bounded by a constant times $(\ln m)^2$.

desk verdict A practical histopolation toolbox, but the proof of the central theorem rests on a false equivalence between histopolation and mean-value interpolation. read the letter →

arxiv 2506.03025 v1 pith:A4Q3XNAQ submitted 2025-06-03 math.NA cs.NA

Ludovico Bruni Bruno , Francesco Dell'Accio , Wolfgang Erb , Federico Nudo This is my paper

classification math.NAcs.NA MSC 33F0541A0541A10

keywords polynomialhistopolationhistopolation-regressionPaduapointstrianglesLebesgueconstantdiscreteLejasequencesapproximateFeketetriangulation

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

The paper asks whether a polynomial can be recovered from cell averages over a triangulation, when only the average of the function on each triangle is known. It proposes selecting a subset of triangles so that the equal-average conditions are both solvable and well-conditioned, and shows that on the so-called Padua triangles this holds under a mesh-fineness condition: the histopolation problem is unisolvent and its Lebesgue constant grows at most like the square of a logarithm. The same construction is combined with a least-squares regression on the remaining triangles, and compared with two greedy alternatives, approximate Fekete and discrete Leja triangles. The significance is that average-based reconstruction, which is the natural data format in many applications, can be made nearly as stable as point interpolation at the well-known Padua nodes.

What carries the argument

The load-bearing object is the Padua triangle set $T_M^{\mathrm{Pad}}$, obtained by assigning each of the $M$ Padua points to the first triangle of the triangulation that contains it. Its injectivity is guaranteed by Proposition 1 via an explicit comparison of $h_{\max}$ with the minimal Padua-point distance. The argument then runs through the mean value theorem, which rewrites every histopolation condition $\frac{1}{|t_i|}\int_{t_i} p\,dxdy = \mu_i(f)$ as $p(\xi_i)=f(\xi_i)$ for an unknown $\xi_i\in t_i$; with $\|x^{\mathrm{Pad}}_i-\xi_i\|_2 \le h_{\max}$, norming-set stability for perturbed nodes converts the Padua points' near-optimal conditioning into unisolvence and the logarithmic-squared Lebesgue bound. For the regression variant, the additional machinery is the constrained least-squares system (13) with Lagrange multipliers, whose direct elimination gives the operator-norm bound in Proposition 3. The greedy alternatives replace the geometric attribution by pivoted QR (approximate Fekete triangles) or pivoted LU (discrete Leja triangles) on the collocation matrix $W_{i,j}=\mu_i(p_j)$.

What would settle it

Compute $L_m(T_M^{\mathrm{Pad}})$ numerically for regular Delaunay meshes with $n$ satisfying (7) and growing $m$: if $L_m(T_M^{\mathrm{Pad}})/(\ln m)^2$ is unbounded, the claimed $K(1-\alpha)^{-1}(\ln m)^2$ bound is false as stated. A direct check of Proposition 1 is to run Algorithm 1 on a random mesh of about 2000 triangles with $m=7$; the paper itself reports that only 25 of the required 36 triangles are found, so the histopolator is undefined.

Watch

Extended reading notes

Core claim

The central claim is that bivariate polynomial histopolation can be built on triangles that contain the Padua points, provided the mesh is fine enough for the attribution map from Padua points to triangles to be injective. Under the explicit condition on the maximal edge length $h_{\max}$ of Proposition 1, the set of Padua triangles has exactly $M=(m+1)(m+2)/2$ elements and is unisolvent for histopolation in $\mathbb{P}_m(\mathbb{R}^2)$ (Proposition 2). The key mechanism is the mean value theorem: each average condition over a triangle is equivalent to a point evaluation at some unknown node inside that triangle, so histopolation becomes a perturbation of interpolation at the Padua points. Because the perturbation is controlled by $h_{\max}$, stability results for perturbed norming sets give a Lebesgue constant bound $L_m(T_M^{\mathrm{Pad}}) \le K(1-\alpha)^{-1}(\ln m)^2$ (Proposition 5), matching the Padua points up to the factor $(1-\alpha)^{-1}$. The paper also shows numerically that the Padua-triangle histopolator converges for all three test functions, while the Fekete and Leja greedy selections suffer conditioning peaks on the same meshes.

Load-bearing premise

The method needs a mesh so fine that every Padua point lies in its own private triangle, and that fails on many irregular meshes: about 30% of randomized structured attempts and almost all completely random Delaunay attempts in the paper's tests.

Editorial extensions

If this is right

If the bound in Proposition 5 holds, triangle-average data on fine regular meshes can be used with the same confidence as point values at Padua nodes, including the logarithmic-squared stability that prevents Runge-type blow-up.
The histopolation-regression operator is a linear projector, so the Lebesgue inequality applies and the extra non-selected triangle averages can only improve the approximation constant; the numerics show faster convergence for smooth functions and removal of instability peaks for rough ones.
For irregular meshes the Fekete and Leja greedy extractions remain well defined where Padua attribution fails, but their Lebesgue constants show isolated peaks and their histopolators do not converge for all test functions, so they are fallback strategies rather than equivalent replacements.
The construction is not tied to squares: the same framework extends to general polygonal domains for Fekete and Leja triangles, and the Padua argument carries over to other low-Lebesgue point sets and to higher-dimensional simplices, as the paper's Remark 5 states.

Reading between the lines

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

A practical cost not emphasized in the paper: on the regular Delaunay mesh used in the experiments, the injectivity condition (7) forces the triangle count to grow roughly like $m^4$ for degree $m$, so the near-optimal conditioning is purchased with a fine mesh; applications needing high degree on coarse meshes would have to rely on the Fekete and Leja fallback.
The same mean-value perturbation argument should work for any point set whose Lebesgue constant grows slowly, not only Padua points; testing other low-Lebesgue node sets on the same triangulations would isolate whether the logarithmic-squared bound is a property of the node set or of the histopolation mechanism.
The 30% failure rate on randomized structured meshes suggests an adaptive strategy: try Padua attribution first, and when it fails, switch to a greedy triangle extraction; the paper's experiments indicate such a hybrid would keep stability on regular regions and robustness on irregular ones.

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.

Desk Editor's Note

read the letter

This is a useful paper for anyone working on reconstruction from cell-averaged data. The Padua-triangle construction is new, the combined histopolation-regression framework is a natural extension of the authors' segmental work, and the comparison with Fekete/Leja triangles is informative. The experiments are honest: they show convergence, but also report that Padua triangles often fail on random meshes.

The weak point is the theory. Proposition 2's proof says the histopolation condition is equivalent to interpolation at an unknown mean-value point. That is not right. The MVT gives separate points for the function and the polynomial, and equality of averages does not imply equality at a common point. The 'perturbed node' depends on f, so the stability result from [35] does not apply. The same gap propagates to Proposition 5, so the logarithmic Lebesgue bound is not established as written. I suspect the result is repairable with a direct estimate of the averaging functional as a perturbation of point evaluation, but the current proof is not that.

There is also a mismatch between the sufficient condition h_max = O((m ln m)^-2) and the mesh sizes in the experiments, so the numerical convergence is not covered by the theorems. Proposition 4 is referenced to an unpublished preprint, and the numerics would benefit from code and error bars. These are secondary.

Who should read it: numerical analysts working on histopolation, cell-average data, or mesh-based approximation. It deserves a serious referee, but revision should be major: fix the proof of Proposition 2 and restate what is actually proven, or restrict the claims. I would not cite the main theorem in its current form.

Referee Report

3 major / 5 minor

Summary. This paper develops polynomial histopolation and combined histopolation-regression on triangulated planar domains. It selects M triangles from a triangulation by association with Padua points, approximate Fekete selection, or discrete Leja sequences, and solves for a bivariate polynomial of total degree m whose averages over the selected triangles match given cell averages, optionally with a least-squares regression on the remaining triangles. The main theoretical results are Proposition 1 (injectivity of the Padua attribution map under a mesh-size condition), Proposition 2 (unisolvence of Padua triangles for histopolation under a stronger mesh-size condition), and Proposition 5 (a logarithmic-squared Lebesgue constant bound). Numerical experiments compare the three triangle-selection strategies on Friedrichs-Keller and random meshes.

Significance. If the theoretical results were established, Proposition 5 would show that histopolation from cell averages on Padua triangles is as well conditioned as Padua interpolation, and the combination with regression would give a practical scheme for average data. The paper contributes useful algorithms (Algorithms 1-4) and a careful numerical comparison, including the observation that Fekete and Leja triangles remain viable on irregular meshes where the Padua attribution map frequently fails. The numerical evidence in Figures 8-13 is informative. However, the central perturbation argument is not currently valid, and the sufficient condition in Proposition 2 is far too restrictive to cover the experiments; the theoretical claims therefore need substantial revision before the paper can be accepted.

major comments (3)

[§2.3 and §2.6, Propositions 2 and 5] The proofs of Propositions 2 and 5 rest on the claim that a histopolation condition on a triangle t is equivalent to an interpolation condition at a point p in t that depends only on f. This is not correct: the mean value theorem gives mu_t(f)=f(p_f) and mu_t(q)=q(p_q) with, in general, p_f different from p_q; the equality mu_t(q)=mu_t(f) does not imply q(p_f)=f(p_f). A one-dimensional instance suffices to see the flaw: on t=[0,1], take f(x)=x and q(x)=6(x-1/2)^2. Both have average 1/2; the unique mean-value point for f is p_f=1/2, and q(p_f)=0, while f(p_f)=1/2. Thus the histopolation operator is not a nodal interpolation operator at f-dependent points, and the cited perturbation result [35, Proposition 1] is not applicable as written. The same gap invalidates the proof of the bound L_m(T_Pad_M) <= K/(1-alpha) (ln m)^2 in Proposition 5.
[§2.3, Eq. (9), and §4.1] The sufficient condition (9) is not satisfied by the Friedrichs-Keller experiments reported in Figures 4, 6, and 8-11. For T_reg_N one has h_max=4/sqrt(N)=2*sqrt(2)/n, while the largest admissible degree in (7) grows like m ~ (pi/2) sqrt(n); hence h_max is of order n^{-1}, whereas the right-hand side alpha/(2K(m ln m)^2) is of order n^{-1}(ln n)^{-2}. The actual h_max therefore exceeds the Proposition 2 threshold by a factor of order (ln n)^2, so Propositions 2 and 5 do not explain the observed convergence and Lebesgue constants. The authors should either prove a less restrictive sufficient condition or explicitly state that the numerical results lie outside the scope of the stated sufficient condition.
[§2.5, Proposition 3 and Eq. (19)] In the proof of Proposition 3, the vector b has N entries, not D entries, so the inequality ||b(f)||_1 <= D ||f||_infinity is false when N > D. It should read ||b(f)||_1 <= N ||f||_infinity, and the first term in the definition of eta_d should be N rather than D. Since the numerical bound in Figure 3 is based on these quantities, the displayed formulas need correction and the figure should be recomputed if the code used D. As stated, Eq. (19) is not a valid upper bound for the operator norm of the histopolation-regression operator.

minor comments (5)

[§2.2, proof of Proposition 1] The sentence 'we must guarantee that the distance between two Padua points is less than h_max' should read 'greater than h_max'; otherwise it contradicts the displayed inequality that follows.
[Figures 8-11] The legend 'Padova triangles' should be 'Padua triangles' to match the terminology used in the text.
[§4.3.1] The text reports that about 30% of attempts failed to produce a Padua histopolator, but Figure 13 plots errors for the Padua method; the authors should state how unsuccessful runs are treated, for instance whether the plotted errors are conditional means over successful realizations.
[Algorithm 4 and Section 4] The numerical evaluation relies on a Gaussian quadrature rule for the triangle averages, but no order or implementation detail is given; specifying it would improve reproducibility.
[Figure 3] The relation zeta_d + eta_d approximately n^{3.64} is an empirical fit to computed values, not a proved estimate; the text should label it as such.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the Padua-triangle histopolation bounds are derived from external perturbation estimates and the Padua Lebesgue bound, not from the paper's own fitted or assumed values.

full rationale

The central derivation chain (Propositions 2 and 5) is not circular. Unisolvence of the Padua triangles is proved by bounding the distance between each Padua point and the mean-value point inside the corresponding triangle, then invoking the external perturbation result [35, Proposition 1] on stability of norming sets; the logarithmic Lebesgue bound uses [35, Corollary 1] and the external Padua Lebesgue-constant estimate of [5]. These are parameter-free stability estimates with explicit constants, and none of them is fitted to the numerical data of this paper. The Padua-triangle set is defined by containment of Padua points, but unisolvence and conditioning are not assumed as inputs; they are shown as consequences of the external estimates. Proposition 4, which identifies the operator norm of a histopolation projector with the quantity L_m(T_M), is cited from [12], a preprint with one overlapping author; however, this is a general functional-analytic characterization and not equivalent to the paper's target logarithmic bound, which rests on [5] and [35]. The remaining self-citations [11,13,14] are used for motivation, analogy, or the one-dimensional segmental predecessor and are not load-bearing inputs to the bivariate derivation. The proof of Proposition 2 contains an MVT-based reduction of histopolation to interpolation at f-dependent points that is questionable as a mathematical statement; this is a correctness risk, not a circular reduction, since histopolation is not defined in terms of those points. No fitted parameter is renamed as a prediction, and no result is forced by self-citation. Therefore the appropriate circularity finding is no significant circularity.

Assumptions & free parameters 1 free parameters · 5 assumptions · 0 invented entities

The central claims rest on standard results (mean value theorem, Padua-point theory, perturbation stability) plus one unpublished preprint identity. No new particles, forces, or unmeasured constants are introduced. The only fitted quantity is the observed growth exponent in a numerical trend, which is not used as a premise.

free parameters (1)

Growth exponent of the zeta_d + eta_d bound trend = approximately 3.64
In Section 2.5 the authors observe numerically that zeta_d + eta_d grows like n^3.64 (Figure 3). This exponent is fitted to the plotted data and is reported as an observation, not used in any theorem. It is not load-bearing for the central claims.

assumptions (5)

standard math Mean value theorem for integrals over a triangle: the average of a continuous function equals its value at some point in the triangle.
Used in Propositions 2, 3, and 5 to convert histopolation conditions into point-evaluation conditions at unknown nodes inside each triangle.
standard math Stability of norming sets under small perturbations (Piazzon and Vianello, ref [35], Proposition 1 and Corollary 1).
External published theorem used in Propositions 2 and 5 to guarantee unisolvence and to control the Lebesgue constant when the averaging points are perturbed away from the Padua nodes.
domain assumption Lebesgue constant identity: the operator norm of a histopolation operator on a set of disjoint triangles equals the corresponding generalized Lebesgue constant L_m(T_M).
Proposition 4 states this identity is derived from ref [12], an unpublished preprint by Bruni Bruno and Elefante. The result is not proved in the paper, yet it is used to interpret the numerically computed constants.
standard math Unisolvence of the Padua points for P_m and the bound L_m(X_Pad^M) <= K (ln m)^2.
External known results from refs [5, 7], used in Propositions 2 and 5 as the base case for the perturbation argument.
domain assumption Gaussian quadrature rules can integrate the basis polynomials over triangles accurately enough to set up the collocation matrices.
Algorithm 4 calls for computing the matrices W and C using a Gaussian quadrature rule; no convergence or accuracy analysis of the quadrature is provided, but this is standard practice in numerical analysis.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Bivariate polynomial histopolation techniques on Padua, Fekete and Leja triangles." pith.science (2026). https://pith.science/paper/A4Q3XNAQ

@misc{pith2026250603025,
  author       = {Pith},
  title        = {Pith review of: Bivariate polynomial histopolation techniques on Padua, Fekete and Leja triangles},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/A4Q3XNAQ}},
  note         = {Machine review of arXiv:2506.03025}
}

abstract

This paper explores the reconstruction of a real-valued function $f$ defined over a domain $\Omega \subset \mathbb{R}^2$ using bivariate polynomials that satisfy triangular histopolation conditions. More precisely, we assume that only the averages of $f$ over a given triangulation $\mathcal{T}_N$ of $\Omega$ are available and seek a bivariate polynomial that approximates $f$ using a histopolation approach, potentially flanked by an additional regression technique. This methodology relies on the selection of a subset of triangles $\mathcal{T}_M \subset \mathcal{T}_N$ for histopolation, ensuring both the solvability and the well-conditioning of the problem. The remaining triangles can potentially be used to enhance the accuracy of the polynomial approximation through a simultaneous regression. We will introduce histopolation and combined histopolation-regression methods using the Padua points, discrete Leja sequences, and approximate Fekete nodes. The proposed algorithms are implemented and evaluated through numerical experiments that demonstrate their effectiveness in function approximation.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

42 extracted references · 42 canonical work pages

[35]

Dolomite s Res

Piazzon, F., Vianello, M.: Stability inequalities for L ebesgue constants via Markov-like inequalities. Dolomite s Res. Notes Approx. 11, 1–9 (2018)

work page 2018
[1]

Alonso Rodr ´ ıguez, A., Rapetti, F.: On a generalization o f the Lebesgue’s constant. J. Comput. Phys. 428, 109964 (2021)

work page 2021
[2]

Atkinson, K., Han, W.: Theoretical numerical analysis, v ol. 39. Springer (2005)

work page 2005
[3]

SIAM (199 6)

Bj¨ orck,˚ A.: Numerical methods for least squares problems. SIAM (199 6)

work page
[4]

Indagationes Mathematicae 34(2), 274–293 (2023)

Bos, L.: On Fekete points for a real simplex. Indagationes Mathematicae 34(2), 274–293 (2023)

work page 2023
[5]

Bos, L., Caliari, M., De Marchi, S., Vianello, M., Xu, Y.: B ivariate Lagrange interpolation at the Padua points: the generating curve approach. J. Approx. Theory 143, 15–25 (2006)

work page 2006
[6]

Bos, L., De Marchi, S., Sommariva, A., Vianello, M.: Compu ting multivariate Fekete and Leja points by numerical linear algebra. SIAM J. Numer. Anal. 48, 1984–1999 (2010)

work page 2010
[7]

Bos, L., De Marchi, S., Vianello, M., Xu, Y.: Bivariate Lag range interpolation at the Padua points: the ideal theory approach. Numer. Math. 108, 43–57 (2007)

work page 2007

Show all 42 references

[8]

Cambridge University press (2018)

Boyd, S., Vandenberghe, L.: Introduction to applied line ar algebra: vectors, matrices, and least squares. Cambridge University press (2018)

2018
[9]

Briani, M., Sommariva, A., Vianello, M.: Computing Feket e and Lebesgue points: simplex, square, disk. J. Comput. Appl. Math. 236(9), 2477–2486 (2012)

2012
[10]

Brudnyi, A., Yomdin, Y.: Norming sets and related Remez- type inequalities. J. Aust. Math. Soc. 100(2), 163–181 (2016)

2016
[11]

Preprint available at https://arxiv.org/abs/2407.16825

Bruni Bruno, L., Dell’Accio, F., Erb, W., Nudo, F.: Polyn omial histopolation on mock-Chebyshev segments (2024). Preprint available at https://arxiv.org/abs/2407.16825

2024 arXiv
[12]

preprint (2024)

Bruni Bruno, L., Elefante, G.: Polynomial reconstructi on from integral data. preprint (2024)

2024
[13]

Bruni Bruno, L., Erb, W.: Polynomial interpolation of fu nction averages on interval segments. SIAM J. Numer. Anal. 62, 1759–1781 (2024)

2024
[14]

BIT Numer

Bruni Bruno, L., Erb, W.: The Fekete problem in segmental polynomial interpolation. BIT Numer. Math. 65(1), Paper No. 3 (2025)

2025
[15]

Caliari, M., De Marchi, S., Vianello, M.: Bivariate poly nomial interpolation on the square at new nodal sets. Appl. Math. Comput. 165(2), 261–274 (2005)

2005
[16]

Caliari, M., De Marchi, S., Vianello, M.: Bivariate Lagr ange interpolation at the Padua points: computational aspects. J. Comput. Appl. Math. 221(2), 284–292 (2008)

2008
[17]

Dover Pu blications, Inc., New York (1975)

Davis, P.J.: Interpolation and Approximation. Dover Pu blications, Inc., New York (1975)

1975
[18]

De Marchi, S.: On Leja sequences: some results and applic ations. Appl. Math. Comput. 152(3), 621–647 (2004)

2004
[19]

Dolomites Res

De Marchi, S., Dell’Accio, F., Nudo, F.: A mixed interpol ation-regression approximation operator on the triangle. Dolomites Res. Notes Approx. 17, 33–44 (2024)

2024
[20]

De Marchi, S., Dell’Accio, F., Mazza, M.: On the constrai ned mock-Chebyshev least-squares. J. Comput. Appl. Math. 280, 94–109 (2015)

2015
[21]

Dell’Accio, F., Di Tommaso, F., Nudo, F.: Generalizatio ns of the constrained mock-Chebyshev least squares in two variables: Tensor product vs total degree polynomial interpolation. Appl. Math. Lett. 125, 107732 (2022)

2022
[22]

Dell’Accio, F., Marcell´ an, F., Nudo, F.: An extension o f a mixed interpolation–regression method using zeros of orthogonal polynomials. J. Comput. Appl. Math. 450, 116010 (2024)

2024
[23]

submitted (2025)

Dell’Accio, F., Mezzanotte, D., Nudo, F., Occorsio, D.: On the error bound of the constrained mock-Chebyshev least squares operator and its generalization to quasi-equ ispaced grids. submitted (2025)

2025
[24]

Dencker, P., Erb, W.: Multivariate polynomial interpol ation on Lissajous-Chebyshev nodes. J. Approx. Theory 219, 15–45 (2017)

2017
[25]

Dencker, P., Erb, W., Kolomoitsev, Y., Lomako, T.: Lebes gue constants for polyhedral sets and polynomial interpolation on Lissajous–Chebyshev nodes. J. Complex. 43, 1–27 (2017)

2017
[26]

Dubiner, M.: The theory of multi-dimensional polynomia l approximation. J. Anal. Math. 67, 39–116 (1995) Bivariate polynomial histopolation techniques on Padua, F ekete and Leja triangles 24

1995
[27]

Edelsbrunner, H.: Geometry and topology for mesh genera tion, Cambridge Monographs on Applied and Com- putational Mathematics , vol. 7. Cambridge University Press, Cambridge (2001)

2001
[28]

Appl Math Comput

Erb, W.: Bivariate Lagrange interpolation at the node po ints of Lissajous curves—the degenerate case. Appl Math Comput. 289, 409–425 (2016)

2016
[29]

Erb, W., Kaethner, C., Ahlborg, M., Buzug, T.M.: Bivaria te Lagrange interpolation at the node points of non-degenerate Lissajous curves. Numer. Math. 133(4), 685–705 (2016)

2016
[30]

In: Recent Advances in Orthogonal Polynomials, Special Functions, and Their Applications, Contemporary Mathematics, vol

Harris, L.A.: Bivariate Lagrange interpolation at the g eronimus nodes. In: Recent Advances in Orthogonal Polynomials, Special Functions, and Their Applications, Contemporary Mathematics, vol. 591, pp. 135–147. American Mathematical Society (2013)

2013
[31]

Hormann, K., Agathos, A.: The point in polygon problem fo r arbitrary polygons. Comput. Geom. 20(3), 131–144 (2001)

2001
[32]

Ibrahimoglu, B.A.: Lebesgue functions and Lebesgue con stants in polynomial interpolation. J. Inequal. Appl. pp. 1–15 (2016)

2016
[33]

Springer (2003)

Knabner, P., Angermann, L., Rupp, A.: Numerical methods for elliptic and parabolic partial diﬀerential equations. Springer (2003)

2003
[34]

Knowl.-Based Syst

Lin, J.Y., Guo, S., Xie, L., Du, R., Xu, G.: A discrete meth od for the initialization of semi-discrete optimal transport problem. Knowl.-Based Syst. 212, 106608 (2021)

2021
[36]

Rahla, R.I., Serra-Capizzano, S., Tablino-Possio, C.: Spectral analysis of Pk ﬁnite element matrices in the case of Friedrichs–Keller triangulations via generalized locally Toeplitz technology. Numer. Linear Algebra Appl. 27(4), e2302 (2020)

2020
[37]

Robidoux, N.: Polynomial histopolation, superconverg ent degrees of freedom, and pseudospectral discrete hodge operators (2006)

2006
[38]

Roth, M.J.: Nodal conﬁgurations and Voronoi tessellati ons for triangular spectral elements. Ph.D. thesis, University of Victoria (2005)

2005
[39]

In: Spline fu nctions and approximation theory (Proc

Schoenberg, I.J.: Splines and histograms. In: Spline fu nctions and approximation theory (Proc. Sympos., Univ. Alberta, Edmonton, Alta., 1972), Internat. Ser. Nume r. Math., Vol. 21, pp. 277–327. Birkh¨ auser Verlag, Basel-Stuttgart (1973)

1973
[40]

Beirao da Veiga, L., Lipnikov, K., Manzini, G.: The mimet ic ﬁnite diﬀerence method for elliptic problems, vol. 11. Springer (2014)

2014
[41]

Xu, Y.: Lagrange interpolation on Chebyshev points of tw o variables. J. Approx. Theory 87(2), 220–238 (1996)

1996
[42]

II, second edn

Zorich, V.A.: Mathematical analysis. II, second edn. Un iversitext. Springer, Heidelberg (2016)

2016

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.

[1] [35]

Dolomite s Res

Piazzon, F., Vianello, M.: Stability inequalities for L ebesgue constants via Markov-like inequalities. Dolomite s Res. Notes Approx. 11, 1–9 (2018)

work page 2018

[2] [1]

Alonso Rodr ´ ıguez, A., Rapetti, F.: On a generalization o f the Lebesgue’s constant. J. Comput. Phys. 428, 109964 (2021)

work page 2021

[3] [2]

Atkinson, K., Han, W.: Theoretical numerical analysis, v ol. 39. Springer (2005)

work page 2005

[4] [3]

SIAM (199 6)

Bj¨ orck,˚ A.: Numerical methods for least squares problems. SIAM (199 6)

work page

[5] [4]

Indagationes Mathematicae 34(2), 274–293 (2023)

Bos, L.: On Fekete points for a real simplex. Indagationes Mathematicae 34(2), 274–293 (2023)

work page 2023

[6] [5]

Bos, L., Caliari, M., De Marchi, S., Vianello, M., Xu, Y.: B ivariate Lagrange interpolation at the Padua points: the generating curve approach. J. Approx. Theory 143, 15–25 (2006)

work page 2006

[7] [6]

Bos, L., De Marchi, S., Sommariva, A., Vianello, M.: Compu ting multivariate Fekete and Leja points by numerical linear algebra. SIAM J. Numer. Anal. 48, 1984–1999 (2010)

work page 2010

[8] [7]

Bos, L., De Marchi, S., Vianello, M., Xu, Y.: Bivariate Lag range interpolation at the Padua points: the ideal theory approach. Numer. Math. 108, 43–57 (2007)

work page 2007