Abstract:High-dimensional interpolation problems appear in various applications of uncertainty quantification, stochastic optimization and machine learning. Such problems are computationally expensive and request the use of adaptive grid generation strategies like anisotropic sparse grids to mitigate the curse of dimensionality. However, it is well known that the standard dimension-adaptive sparse grid method converges very slowly or even fails in the case of non-smooth functions. For piecewise smooth functions with kinks, we construct two novel hp-adaptive sparse grid collocation algorithms that combine low-order basis functions with local support in parts of the domain with less regularity and variable-order basis functions elsewhere. Spatial refinement is realized by means of a hierarchical multivariate knot tree which allows the construction of localised hierarchical basis functions with varying order. Hierarchical surplus is used as an error indicator to automatically detect the non-smooth region and adaptively refine the collocation points there. The local polynomial degrees are optionally selected by a greedy approach or a kink detection procedure. Three numerical benchmark examples with different dimensions are discussed and comparison with locally linear and highest degree basis functions are given to show the efficiency and accuracy of the proposed methods.

What problem does this paper attempt to address?

This paper attempts to solve the problem that in high - dimensional interpolation problems, when the function is not smooth in some regions (such as having discontinuities or inflection points), the traditional sparse grid method has a slow convergence rate or even fails. Specifically: 1. **Background of high - dimensional interpolation problems**: - High - dimensional interpolation problems widely occur in fields such as uncertainty quantification, stochastic optimization, and machine learning. - The traditional full tensor - product grid method has a too - high computational cost due to the curse of dimensionality, that is, as the dimension increases, the number of required grid points grows exponentially. 2. **Limitations of existing methods**: - Although the sparse grid method can effectively reduce the amount of computation, it performs poorly when dealing with non - smooth functions, especially for piecewise - smooth functions with discontinuities or inflection points, and its convergence rate is very slow or even fails to converge. 3. **Methods proposed in the paper**: - The paper proposes two new adaptive hp - polynomial sparse grid collocation algorithms, aiming to combine low - order locally - supported basis functions and variable - order basis functions to improve the resolution of non - smooth regions. - These methods achieve spatial refinement through hierarchical multivariate knot trees and use the hierarchical residual as an error indicator to automatically detect non - smooth regions and adaptively adjust the collocation points. 4. **Specific technical means**: - **Local polynomial basis functions**: Adjust the order of the polynomial according to local regularity. - **Greedy algorithm**: Calculate scores for all possible polynomial orders and select the order with the smallest interpolation error. - **Inflection point detection**: Identify regions where the first - order derivative is discontinuous based on polynomial annihilation techniques and use piecewise - linear basis functions in these regions. 5. **Verification and comparison**: - The paper verifies the effectiveness and accuracy of the proposed methods through three numerical benchmark examples of different dimensions and compares them with the local - linear and highest - order basis function methods. In summary, this paper aims to improve the sparse grid method so that it can efficiently handle high - dimensional piecewise - smooth functions with discontinuities or inflection points, thereby improving interpolation accuracy and computational efficiency.

Adaptive hp-Polynomial Based Sparse Grid Collocation Algorithms for Piecewise Smooth Functions with Kinks

Coarsened C1-continuous Shape Functions Using B-spline Patches for Simulating Nonlinear Heterogeneous Structures

An adaptive hierarchical sparse grid collocation algorithm for the solution of stochastic differential equations

B-Splines for Sparse Grids: Algorithms and Application to Higher-Dimensional Optimization

Greedy Adaptive Local Recovery of Functions in Sobolev Spaces

High-order, multidimensional, and conservative coarse–fine interpolation for adaptive mesh refinement

Regular polynomial interpolation and approximation of global solutions of linear partial differential equations

Two-grid $hp$-version discontinuous Galerkin finite element methods for quasilinear elliptic PDEs on agglomerated coarse meshes

An Adaptive Multi‐Grid Solver for Applications in Computer Graphics

Adaptive Sparse Polynomial Chaos Expansions via Leja Interpolation

A Discrete Adapted Hierarchical Basis Solver For Radial Basis Function Interpolation

Efficient Computation of Smoothing Splines Via Adaptive Basis Sampling

A discontinuous piecewise polynomial generalized moving least squares scheme for robust finite element analysis on arbitrary grids

Product kernels are efficient and flexible tools for high-dimensional scattered data interpolation

Low-Rank Kernel Matrix Approximation Using Skeletonized Interpolation With Endo- or Exo-Vertices

The Sparse-Grid-Based Adaptive Spectral Koopman Method

Polyharmonic homogenization, rough polyharmonic splines and sparse super-localization

Classical multivariate Hermite coordinate interpolation on n-dimensional grids

Isogeometric collocation with smooth mixed degree splines over planar multi-patch domains

Adaptive meshfree approximation for linear elliptic partial differential equations with PDE-greedy kernel methods

Coarse Spaces Based on Higher-Order Interpolation for Schwarz Preconditioners for Helmholtz Problems