Skip to main content

Complexity of zigzag sampling algorithm for strongly log-concave distributions

Publication ,  Journal Article
Lu, J; Wang, L
Published in: Stat Comput
December 20, 2020

We study the computational complexity of zigzag sampling algorithm for strongly log-concave distributions. The zigzag process has the advantage of not requiring time discretization for implementation, and that each proposed bouncing event requires only one evaluation of partial derivative of the potential, while its convergence rate is dimension independent. Using these properties, we prove that the zigzag sampling algorithm achieves $\varepsilon$ error in chi-square divergence with a computational cost equivalent to $O\bigl(\kappa^2 d^\frac{1}{2}(\log\frac{1}{\varepsilon})^{\frac{3}{2}}\bigr)$ gradient evaluations in the regime $\kappa \ll \frac{d}{\log d}$ under a warm start assumption, where $\kappa$ is the condition number and $d$ is the dimension.

Duke Scholars

Published In

Stat Comput

Publication Date

December 20, 2020

Volume

32

Start / End Page

48
 

Citation

APA
Chicago
ICMJE
MLA
NLM
Lu, Jianfeng, and Lihan Wang. “Complexity of zigzag sampling algorithm for strongly log-concave distributions.” Stat Comput 32 (December 20, 2020): 48.
Lu, Jianfeng, and Lihan Wang. “Complexity of zigzag sampling algorithm for strongly log-concave distributions.” Stat Comput, vol. 32, Dec. 2020, p. 48.

Published In

Stat Comput

Publication Date

December 20, 2020

Volume

32

Start / End Page

48