On the variance of Schatten p-norm estimation with Gaussian sketching matrices
-
Lior Horesh
and
Tomasz Nowicki
Abstract
Monte Carlo matrix trace estimation is a popular randomized technique to estimate the trace of implicitly-defined matrices via averaging quadratic forms across several observations of a random vector. The most common approach to analyze the quality of such estimators is to consider the variance over the total number of observations. In this paper we present a procedure to compute the variance of the estimator proposed in [W. Kong and G. Valiant, Spectrum estimation from samples, Ann. Statist. 45 2017, 5, 2218–2247] for the case of Gaussian random vectors and provide a sharper bound than previously available.
References
[1] Y.-C. Chu and A. Cortinovis, Improved bounds for randomized Schatten norm estimation of numerically low-rank matrices, preprint (2024), https://arxiv.org/abs/2408.17414. Search in Google Scholar
[2] E. Dudley, A. K. Saibaba and A. Alexanderian, Monte Carlo estimators for the Schatten p-norm of symmetric positive semidefinite matrices, Electron. Trans. Numer. Anal. 55 (2022), 213–241. 10.1553/etna_vol55s213Search in Google Scholar
[3] V. Kalantzis, S. Ubaru, C. W. Wu, G. Kollias and L. Horesh, Asynchronous randomized trace estimation, Proc. Mach. Learn. Res. (PMLR) 238 (2024), 4294–4302. Search in Google Scholar
[4] W. Kong and G. Valiant, Spectrum estimation from samples, Ann. Statist. 45 (2017), no. 5, 2218–2247. 10.1214/16-AOS1525Search in Google Scholar
[5] Y. Li, H. L. Nguyen and D. P. Woodruff, On sketching matrix norms and the top singular vector, Proceedings of the Twenty-Fifth Annual ACM-SIAM Symposium on Discrete Algorithms, ACM, New York (2014), 1562–1581. 10.1137/1.9781611973402.114Search in Google Scholar
[6] P.-G. Martinsson and J. A. Tropp, Randomized numerical linear algebra: foundations and algorithms, Acta Numer. 29 (2020), 403–572. 10.1017/S0962492920000021Search in Google Scholar
[7] Y. Xie, S. Gu, Y. Liu, W. Zuo, W. Zhang and L. Zhang, Weighted Schatten p-norm minimization for image denoising and background subtraction, IEEE Trans. Image Process. 25 (2016), no. 10, 4842–4857. 10.1109/TIP.2016.2599290Search in Google Scholar
[8] R. Zhang, S. Frei and P. L. Bartlett, Trained transformers learn linear models in-context, J. Mach. Learn. Res. 25 (2024), 1–55. Search in Google Scholar
© 2025 Walter de Gruyter GmbH, Berlin/Boston
Articles in the same Issue
- Frontmatter
- Impact of psychrometry on the aerosol distribution pattern in human lungs
- Bayesian inference of traffic intensity in M/M/1 queue under symmetric and asymmetric loss functions
- On the variance of Schatten p-norm estimation with Gaussian sketching matrices
- A meshfree Random Walk on Boundary algorithm with iterative refinement
- Combining randomized and deterministic iterative algorithms for high accuracy solution of large linear systems and boundary integral equations
- Preservation of structural properties of the CIR model by θ-Milstein schemes
Articles in the same Issue
- Frontmatter
- Impact of psychrometry on the aerosol distribution pattern in human lungs
- Bayesian inference of traffic intensity in M/M/1 queue under symmetric and asymmetric loss functions
- On the variance of Schatten p-norm estimation with Gaussian sketching matrices
- A meshfree Random Walk on Boundary algorithm with iterative refinement
- Combining randomized and deterministic iterative algorithms for high accuracy solution of large linear systems and boundary integral equations
- Preservation of structural properties of the CIR model by θ-Milstein schemes
