Abstract
We construct approximate Fekete point sets for kernel-based interpolation by maximising the determinant of a kernel Gram matrix obtained via truncation of an orthonormal expansion of the kernel. Uniform error estimates are proved for kernel interpolants at the resulting points. If the kernel is Gaussian, we show that the approximate Fekete points in one dimension are the solution to a convex optimisation problem and that the interpolants converge with a super-exponential rate. Numerical examples are provided for the Gaussian kernel.
Similar content being viewed by others
Change history
05 December 2020
A Correction to this paper has been published: https://doi.org/10.1007/s11075-020-01034-0
Notes
Wendland [38, p. 192] goes as far as describing these results “almost pointless” for kernels, such as the Gaussian, that are associated with exponential rates of convergence.
Observe that in this section we begin indexing of the expansion from zero to simplify notation.
In one dimension, the convergence occurs for any points and most commonly used infinitely smooth radial kernels but in higher dimensions the Gaussian kernel is special in that it is the only known kernel for which convergence to a polynomial interpolant, of minimal degree in a certain sense, occurs for every point set.
This is the constant C in Theorem 6.1 of [24]. To derive the claimed bound, observe that this constant is given as C = 𝜖B/4 for \(B \leq \min \limits \{(b-a)/6, 1\}\) in their proof of Theorem 4.5. On p. 120, they show that 𝜖 = 1/2 if the kernel is Gaussian.
References
Arcangéli, R., de Silanes, M. C. L., Torrens, J. J.: An extension of a bound for functions in Sobolev spaces, with applications to (m,s)-spline interpolation and smoothing. Numer. Math. 107(2), 181–211 (2007)
Beatson, R.: Error bounds for anisotropic RBF interpolation. J. Approx. Theory 162(3), 512–527 (2010)
Belhadji, A., Bardenet, R., Chainais, P.: Kernel quadrature with DPPs. Adv. Neural Inf. Process. Syst. 32, 12907–12917 (2019)
Berlinet, A., Thomas-agnan, C.: Reproducing Kernel Hilbert Spaces in Probability and Statistics. Springer, Berlin (2004)
Bos, L., De Marchi, S.: On optimal points for interpolation by univariate exponential functions. Dol. Res. Notes Approx. 4, 8–12 (2011)
Bos, L., De Marchi, S., Sommariva, A., Vianello, M.: Computing multivariate Fekete and Leja points by numerical linear algebra. SIAM J. Numer. Anal. 48(5), 1984–1999 (2010)
Bos, L. P., Maier, U.: On the asymptotics of Fekete-type points for univariate radial basis interpolation. J. Approx. Theory 119(2), 252–270 (2002)
Briani, M., Sommariva, A., Vianello, M.: Computing Fekete and Lebesgue points: simplex, square, disk. J. Comput. Appl. Math. 236(9), 2477–2486 (2012)
De Marchi, S., Schaback, R.: Nonstandard kernels and their applications. Dol. Res. Notes Approx. 2, 16–43 (2009)
De Marchi, S., Schaback, R.: Stability of kernel-based interpolation. Adv. Comput. Math. 32, 155–161 (2010)
De Marchi, S., Schaback, R., Wendland, H.: Near-optimal data-independent point locations for radial basis function interpolation. Adv. Comput. Math. 23(3), 317–330 (2005)
Fasshauer, G., Hickernell, F., Woźniakowski, H.: On dimension-independent rates of convergence for function approximation with Gaussian kernels. SIAM J. Numer. Anal. 50(1), 247–271 (2012)
Fasshauer, G., McCourt, M.: Kernel-based Approximation Methods Using MATLAB. Number 19 in Interdisciplinary Mathematical Sciences. World Scientific Publishing (2015)
Fasshauer, G. E.: Meshfree Approximation Methods with MATLAB. Number 6 in Interdisciplinary Mathematical Sciences. World Scientific Publishing (2007)
Fasshauer, G. E., McCourt, M. J.: Stable evaluation of Gaussian radial basis function interpolants. SIAM J. Sci. Comput. 34(2), A737–A762 (2012)
Gautier, G., Bardenet, R., Valko, M.: On two ways to use determinantal point processes for Monte Carlo integration. Adv. Neural Inf. Process. Syst. 32, 7768–7777 (2019)
Johansson, F., et al.: mpmath: a Python library for arbitrary-precision floating-point arithmetic (version 1.10). http://mpmath.org/ (2018)
Karvonen, T., Särkkä, S.: Worst-case optimal approximation with increasingly flat Gaussian kernels. Advances in Computational Mathematics. Published online https://doi.org/10.1007/s10444-020-09767-1 (2020)
Lee, Y. J., Yoon, G. J., Yoon, J.: Convergence of increasingly flat radial basis interpolants to polynomial interpolants. SIAM J. Math. Anal. 39 (2), 537–553 (2007)
Minh, H. Q.: Some properties of Gaussian reproducing kernel Hilbert spaces and their implications for function approximation and learning theory. Constr. Approx. 32(2), 307–338 (2010)
Müller, S.: Komplexität und Stabilität von kernbasierten Rekonstruktionsmethoden. PhD Thesis, Institut für Numerische und Angewandte Mathematik, Georg-August-Universität Göttingen (2009)
Narcowich, F. J., Ward, J. D., Wendland, H.: Sobolev error estimates and a Bernstein inequality for scattered data interpolation via radial basis functions. Constr. Approx. 24(2), 175–186 (2006)
Paulsen, V. I., Raghupathi, M.: An Introduction to the Theory of Reproducing Kernel Hilbert Spaces. Number 152 in Cambridge Studies in Advanced Mathematics. Cambridge University Press (2016)
Rieger, C., Zwicknagl, B.: Sampling inequalities for infinitely smooth functions, with applications to interpolation and machine learning. Adv. Comput. Math. 32, 103–129 (2010)
Rieger, C., Zwicknagl, B.: Improved exponential convergence rates by oversampling near the boundary. Constr. Approx. 39(2), 323–341 (2014)
Robbins, H.: A remark on Stirling’s formula. Am. Math. Mon. 62(1), 26–29 (1955)
Santin, G., Haasdonk, B.: Convergence rate of the data-independent P-greedy algorithm in kernel-based approximation. Dol. Res. Notes Approx. 10, 68–78 (2017)
Schaback, R.: Improved error bounds for scattered data interpolation by radial basis functions. Math. Comput. 68(225), 201–216 (1999)
Schaback, R.: A unified theory of radial basis functions: native Hilbert spaces for radial basis functions II. J. Comput. Appl. Math. 121(1–2), 165–177 (2000)
Schaback, R.: Multivariate interpolation by polynomials and radial basis functions. Constr. Approx. 21(3), 293–317 (2005)
Schaback, R.: Superconvergence of kernel-based interpolation. J. Approx. Theory 235, 1–19 (2018)
Schaback, R., Wendland, H.: Adaptive greedy techniques for approximate solution of large RBF systems. Numer. Algorithm. 24(3), 239–254 (2000)
Sloan, I. H., Woźniakowski, H.: Multivariate approximation for analytic functions with Gaussian kernels. J. Complex. 45, 1–21 (2018)
Steinwart, I., Hush, D., Scovel, C.: An explicit description of the reproducing kernel Hilbert spaces of Gaussian RBF kernels. IEEE Trans. Inf. Theory 52(10), 4635–4643 (2006)
Sun, H.: Mercer theorem for RKHS on noncompact sets. J. Complex. 21(3), 337–349 (2005)
Tanaka, K.: Generation of point sets by convex optimization for interpolation in reproducing kernel Hilbert spaces. Numer. Algorithm. 84, 1049–1079 (2020)
Tanaka, K., Sugihara, M.: Design of accurate formulas for approximating functions in weighted Hardy spaces by discrete energy minimization. IMA J. Numer. Anal. 39(4), 1957–1984 (2019)
Wendland, H.: Scattered Data Approximation. Number 17 in Cambridge Monographs on Applied and Computational Mathematics. Cambridge University Press (2005)
Wendland, H., Rieger, C.: Approximate interpolation with applications to selecting smoothing parameters. Numer. Math. 101(4), 729–748 (2005)
Wirtz, D., Haasdonk, B.: A vectorial kernel orthogonal greedy algorithm. Dol. Res. Notes Approx. 6, 83–100 (2013)
Funding
T. Karvonen was supported by the Aalto ELEC Doctoral School and the Lloyd’s Register Foundation programme on data-centric engineering at the Alan Turing Institute, UK. This research was partially carried out while he was visiting the University of Tokyo, funded by the Finnish Foundation for Technology Promotion and Oskar Öflunds Stiftelse. S. Särkkä was supported by the Academy of Finland. K. Tanaka was supported by the grant-in-aid of Japan Society of the Promotion of Science with KAKENHI Grant Number 17K14241.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
The original online version of this article was revised: The corrected equations in Sections 2.1, 2.2, 3.2, 3.3 and 4.3.
Rights and permissions
About this article
Cite this article
Karvonen, T., Särkkä, S. & Tanaka, K. Kernel-based interpolation at approximate Fekete points. Numer Algor 87, 445–468 (2021). https://doi.org/10.1007/s11075-020-00973-y
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11075-020-00973-y