Tucker Tensor analysis of Matern functions in spatial statistics

TL;DR

This paper introduces tensor-based numerical methods to efficiently approximate covariance functions in spatial statistics, significantly reducing computational costs for large datasets and enabling advanced statistical analyses.

Contribution

It develops and proves the exponential convergence of Tucker and canonical tensor decompositions for Matern functions, facilitating scalable statistical computations in high dimensions.

Findings

Exponential convergence of tensor approximations demonstrated numerically.

Storage cost reduced from exponential to linear in data size.

Efficient computation of covariance-related operations using low-rank tensors.

Abstract

In this work, we describe advanced numerical tools for working with multivariate functions and for the analysis of large data sets. These tools will drastically reduce the required computing time and the storage cost, and, therefore, will allow us to consider much larger data sets or finer meshes. Covariance matrices are crucial in spatio-temporal statistical tasks, but are often very expensive to compute and store, especially in 3D. Therefore, we approximate covariance functions by cheap surrogates in a low-rank tensor format. We apply the Tucker and canonical tensor decompositions to a family of Matern- and Slater-type functions with varying parameters and demonstrate numerically that their approximations exhibit exponentially fast convergence. We prove the exponential convergence of the Tucker and canonical approximations in tensor rank parameters. Several statistical operations are performed in this low-rank tensor format, including evaluating the conditional covariance matrix, spatially averaged estimation variance, computing a quadratic form, determinant, trace, loglikelihood, inverse, and Cholesky decomposition of a large covariance matrix. Low-rank tensor approximations reduce the computing and storage costs essentially. For example, the storage cost is reduced from an exponential $\mathcal{O}(n^d)$ to a linear scaling $\mathcal{O}(drn)$, where $d$ is the spatial dimension, $n$ is the number of mesh points in one direction, and $r$ is the tensor rank. Prerequisites for applicability of the proposed techniques are the assumptions that the data, locations, and measurements lie on a tensor (axes-parallel) grid and that the covariance function depends on a distance, $\Vert x-y \Vert$.