Optimal approximation of infinite-dimensional holomorphic functions. II: Recovery from i.i.d. pointwise samples
The paper continues previous work, e.g., [\textit{B. Adcock}, et al. Calcolo 61, No.~1, Paper No.~12, 45 p.\ (2024; Zbl 1557.65055)], in which infinite dimensional functions are estimated using \(m\) (adaptive or random) samples. In this case the functions map \(\boldsymbol{y}\in\mathcal{U}:=[-1,1]^{\mathbb{N}}\) into a Hilbert space \(\mathcal{V}\). The fuctions are assumed to be holomorphic in some Bernstein polyellipse (with foci in \(\pm1\), parametrized by \(\boldsymbol{\rho}\in[1,\infty)^{\mathbb{N}}\) and weighted by \(\boldsymbol{b}\in[0,\infty)^{\mathbb{N}}\)). The images \(f(\boldsymbol{y})\) are measured by a Banach-norm which is an \(L^p\)-type norm for \(\|f(\boldsymbol{y})\|_{\mathcal{V}}\) weighted by \(\varrho(\boldsymbol{y})\) where \(\varrho\) is a probability measure on \(\mathcal{U}\). In this paper in particular \(\varrho\) is taken to be a tensor product of \((\alpha_i,\beta_i)\)-Jacobi measures. This means that the theory relates to Jacobi polynomial expansions, generalizing Legendre or Chebyshev cases considered previously. Like in the previous paper, also here, two reconstruction maps are considered: either based on least squares or on compressed sensing which are effective for either known and unknown relative importance of the variables, respectively.\N\NThe results are probabilistic error bounds for the approximation error, as well as uniform upper bounds for the whole class of functions considered.\N\NFor the practically important case of pointwise evaluations that are drawn independently from the identical underlying probability, the optimal algebraic convergence rate is the same as obtained in their previous paper, up to a small factor. This means that i.i.d.\ samples are nearly optimal for this class of holomorphic functions.
- A new upper bound for sampling numbers
- A note on sampling recovery of multivariate functions in the uniform norm
- A problem concerning orthogonal polynomials.
- Analytic regularity and nonlinear approximation of a class of parametric semilinear elliptic PDEs
- Approximation of high-dimensional parametric PDEs
- Breaking the curse of dimensionality in sparse polynomial approximation of parametric PDEs
- Constructive deep ReLU neural network approximation
- Constructive subsampling of finite frames with applications in optimal function recovery
- Convergence rates of best \(N\)-term Galerkin approximations for a class of elliptic SPDEs
- Deep learning in high dimension: neural network expression rates for generalized polynomial chaos expansions in UQ
- Function values are enough for \(L_2\)-approximation
- Function values are enough for \(L_2\)-approximation. II
- Generalized Jacobi polynomials/functions and their applications
- Interpolation via weighted \(\ell_{1}\) minimization
- Introduction to uncertainty quantification
- Karhunen-Loève approximation of random fields by generalized fast multipole methods
- Neural and spectral operator surrogates: unified construction and expression rate bounds
- On efficient algorithms for computing near-best polynomial approximations to high-dimensional, Hilbert-valued functions from limited samples
- On optimal recovery in L₂
- On sampling discretization in \(L_2\)
- On the optimal polynomial approximation of stochastic PDEs by Galerkin and collocation methods
- On the stability and accuracy of least squares approximations
- Optimal approximation of infinite-dimensional holomorphic functions
- Optimal pointwise sampling for \(L^2\) approximation
- Random points are optimal for the approximation of Sobolev functions
- Randomized least-squares with minimal oversampling and interpolation in general spaces
- Sparse Polynomial Approximation of High-Dimensional Functions
- Sparse polynomial approximation of parametric elliptic PDEs. Part I: affine coefficients
- Sparse recovery in bounded Riesz systems with applications to numerical methods for PDEs
- Stochastic finite element methods for partial differential equations with random input data
- Tractability of multivariate problems. Volume I: Linear information
- Tractability of multivariate problems. Volume II: Standard information for functionals.
- Uncertainty quantification. Theory, implementation, and applications
- User-friendly tail bounds for sums of random matrices
This page was built for publication: Optimal approximation of infinite-dimensional holomorphic functions. II: Recovery from i.i.d. pointwise samples
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6988790)
