Conference Program Home
  My Program

All Times EDT

Abstract Details

Activity Number: 73 - SPEED: Statistical Computing and Statistics in Genomics Part 2
Type: Contributed
Date/Time: Sunday, August 7, 2022 : 5:05 PM to 5:50 PM
Sponsor: Section on Statistical Computing
Abstract #323731
Title: High-Dimensional Nonlinear Spatio-Temporal Filtering Using Hierarchical Sparse Cholesky Factors
Author(s): Anirban Chakraborty* and Matthias Katzfuss
Companies: Texas A&M University and Texas A&M University
Keywords: cholesky factor; computational statistics; nonlinear operator; parallel computing
Abstract:

Spatial statistics often involves Cholesky decomposition of covariance matrices. To ensure scalability to high dimensions, several recent approximations have assumed sparse Cholesky factors. However, extensions to nonlinear filtering require calculation of the gradient matrix of the nonlinear operator, which is computationally extensive and not possible in some situations. We propose an algorithm based on a hierarchical sparse Cholesky factor that is updated by compressing it into a smaller dense matrix, directly applying the nonlinear operator on the compressed Cholesky factor, and subsequently decompressing. We demonstrate the advantage of our method in several numerical comparisons.


Authors who are presenting talks have a * after their name.

Back to the full JSM 2022 program