English

Lower bounds for trace estimation via Block Krylov and other methods

Statistics Theory 2025-07-01 v1 Data Structures and Algorithms Machine Learning Numerical Analysis Numerical Analysis Statistics Theory

Abstract

This paper studies theoretical lower bounds for estimating the trace of a matrix function, tr(f(A))\text{tr}(f(A)), focusing on methods that use Hutchinson's method along with Block Krylov techniques. These methods work by approximating matrix-vector products like f(A)Vf(A)V using a Block Krylov subspace. This is closely related to approximating functions with polynomials. We derive theoretical upper bounds on how many Krylov steps are needed for functions such as A1/2A^{-1/2} and A1A^{-1} by analyzing the upper bounds from the polynomial approximation of their scalar equivalent. In addition, we also develop lower limits on the number of queries needed for trace estimation, specifically for tr(Wp)\text{tr}(W^{-p}) where WW is a Wishart matrix. Our study clarifies the connection between the number of steps in Block Krylov methods and the degree of the polynomial used for approximation. This links the total cost of trace estimation to basic limits in polynomial approximation and how much information is needed for the computation.

Keywords

Cite

@article{arxiv.2506.22701,
  title  = {Lower bounds for trace estimation via Block Krylov and other methods},
  author = {Shi Jie Yu},
  journal= {arXiv preprint arXiv:2506.22701},
  year   = {2025}
}
R2 v1 2026-07-01T03:37:29.061Z