Emergent Mind

Abstract

Vandermonde matrices are usually exponentially ill-conditioned and often result in unstable approximations. In this paper, we introduce and analyze the \textit{multivariate Vandermonde with Arnoldi (V+A) method}, which is based on least-squares approximation together with a Stieltjes orthogonalization process, for approximating continuous, multivariate functions on $d$-dimensional irregular domains. The V+A method addresses the ill-conditioning of the Vandermonde approximation by creating a set of discrete orthogonal basis with respect to a discrete measure. The V+A method is simple and general. It relies only on the sample points from the domain and requires no prior knowledge of the domain. In this paper, we first analyze the sample complexity of the V+A approximation. In particular, we show that, for a large class of domains, the V+A method gives a well-conditioned and near-optimal $N$-dimensional least-squares approximation using $M=\mathcal{O}(N2)$ equispaced sample points or $M=\mathcal{O}(N2\log N)$ random sample points, independently of $d$. We also give a comprehensive analysis of the error estimates and rate of convergence of the V+A approximation. Based on the multivariate V+A approximation, we propose a new variant of the weighted V+A least-squares algorithm that uses only $M=\mathcal{O}(N\log N)$ sample points to give a near-optimal approximation. Our numerical results confirm that the (weighted) V+A method gives a more accurate approximation than the standard orthogonalization method for high-degree approximation using the Vandermonde matrix.

We're not able to analyze this paper right now due to high demand.

Please check back later (sorry!).

Generate a summary of this paper on our Pro plan:

We ran into a problem analyzing this paper.

Newsletter

Get summaries of trending comp sci papers delivered straight to your inbox:

Unsubscribe anytime.