Abstract:We propose a new approach for approximating functions in $C([0,1]^d)$ via Kolmogorov superposition theorem (KST) based on the linear spline approximation of the K-outer function in Kolmogorov superposition representation. We improve the results in \cite{LaiShenKST21} by showing that the optimal approximation rate based on our proposed approach is $\mathcal{O}(\frac{1}{n^2})$, with $n$ being the number of knots over $[0,1]$, and the approximation constant increases linearly in $d$. We show that there is a dense subclass in $C([0,1]^d)$ whose approximation can achieve such optimal rate, and the number of parameters needed in such approximation is at most $\mathcal{O}(nd)$. Moreover, for $d\geq 4$, we apply the tensor product spline denoising technique to smooth the KB-splines and get the corresponding LKB-splines. We use those LKB-splines as the basis to approximate functions for the cases when $d=4$ and $d=6$, which extends the results in \cite{LaiShenKST21} for $d=2$ and $d=3$. Based on the idea of pivotal data locations introduced in \cite{LaiShenKST21}, we validate via numerical experiments that fewer than $\mathcal{O}(nd)$ function values are enough to achieve the approximation rates such as $\mathcal{O}(\frac{1}{n})$ or $\mathcal{O}(\frac{1}{n^2})$ based on the smoothness of the K-outer function. Finally, we demonstrate that our approach can be applied to numerically solving partial differential equation such as the Poisson equation with accurate approximation results.

Computing best discrete least-squares approximations by first-degree splines with free knots

Discrete approximation by first-degree splines with free knots

Approximate T-Spline Surface Skinning

The best quadrature based on given Hermite information for the Sobolev class KW r [ a , b ]

Knot Calculation for Spline Fitting Via Sparse Optimization

Best free knot linear spline approximation and its application to neural networks

Estimation of Knots in Linear Spline Models

B-spline surface fitting by iterative geometric interpolation/approximation algorithms

Smoothing splines for discontinuous signals

The Optimal Rate for Linear KB-splines and LKB-splines Approximation of High Dimensional Continuous Functions and its Application

Spline regression with automatic knot selection

High-Degree Splines from Discrete Fourier Transforms: Robust Methods to Obtain the Boundary Conditions

KNOT PLACEMENT FOR B-SPLINE CURVE APPROXIMATION VIA L∞,1-Norm AND DIFFERENTIAL EVOLUTION ALGORITHM

Approximation of splines in Wasserstein spaces

Knot Calculation for Spline Fitting Based on the Unimodality Property.

Compact Finite Differences and Cubic Splines

Accelerating Accurate B-spline Free-form Deformation of Polygonal Objects

The Unimodality of Initial B-Spline Approximations in Spline Fitting

Free knot linear interpolation and the Milstein scheme for stochastic differential equations

Multivariate Splines and Their Applications

Computational Topology for Approximations of Knots