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

Download:

Current browse context:

math.OC

Change to browse by:

References & Citations

Bookmark

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

Mathematics > Optimization and Control

Title: A direct proof of convergence of Davis-Yin splitting algorithm allowing larger stepsizes

Abstract: This note is devoted to the splitting algorithm proposed by Davis and Yin in 2017 for computing a zero of the sum of three maximally monotone operators, with one of them being cocoercive. We provide a direct proof that guarantees its convergence when the stepsizes are smaller than four times the cocoercivity constant, thus doubling the size of the interval established by Davis and Yin. As a by-product, the same conclusion applies to the forward-backward splitting algorithm. Further, we use the notion of "strengthening" of a set-valued operator to derive a new splitting algorithm for computing the resolvent of the sum. Last but not least, we provide some numerical experiments illustrating the importance of appropriately choosing the stepsize and relaxation parameters of the algorithms.
Subjects: Optimization and Control (math.OC)
MSC classes: 47H05, 90C30, 65K05
Cite as: arXiv:2108.01318 [math.OC]
  (or arXiv:2108.01318v2 [math.OC] for this version)

Submission history

From: David Torregrosa-Belén [view email]
[v1] Tue, 3 Aug 2021 06:33:30 GMT (470kb,D)
[v2] Mon, 17 Jan 2022 10:01:16 GMT (1014kb,D)

Link back to: arXiv, form interface, contact.