We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

stat.ML

Change to browse by:

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Mathematics > Classical Analysis and ODEs

Title: On $L^q$ Convergence of the Hamiltonian Monte Carlo

Abstract: We establish $L_q$ convergence for Hamiltonian Monte Carlo algorithms. More specifically, under mild conditions for the associated Hamiltonian motion, we show that the outputs of the algorithms converge (strongly for $2\le q<\infty$ and weakly for $1<q<2$) to the desired target distribution.
Subjects: Classical Analysis and ODEs (math.CA); Data Structures and Algorithms (cs.DS); Machine Learning (cs.LG); Functional Analysis (math.FA); Machine Learning (stat.ML)
Cite as: arXiv:2101.08688 [math.CA]
  (or arXiv:2101.08688v2 [math.CA] for this version)

Submission history

From: Yingdong Lu [view email]
[v1] Thu, 21 Jan 2021 15:59:32 GMT (8kb)
[v2] Sun, 5 Jun 2022 00:53:19 GMT (39kb)

Link back to: arXiv, form interface, contact.