eprintid: 2226
rev_number: 6
eprint_status: archive
userid: 6
dir: disk0/00/00/22/26
datestamp: 2014-07-01 11:13:05
lastmod: 2014-07-01 11:13:05
status_changed: 2014-07-01 11:13:05
type: book_section
metadata_visibility: show
creators_name: Patrinos, Panagiotis
creators_name: Bemporad, Alberto
creators_id: panagiotis.patrinos@imtlucca.it
creators_id: alberto.bemporad@imtlucca.it
title: Proximal Newton methods for convex composite optimization
ispublished: pub
subjects: QA75
divisions: CSA
full_text_status: none
keywords: Approximation algorithms, Approximation methods, Convergence, Gradient methods, Radio frequency, Signal processing algorithms
abstract: This paper proposes two proximal Newton methods for convex nonsmooth optimization problems in composite form. The algorithms are based on a new continuously differentiable exact penalty function, namely the Composite Moreau Envelope. The first algorithm is based on a standard line search strategy, whereas the second one combines the global efficiency estimates of the corresponding first-order methods, while achieving fast asymptotic convergence rates. Furthermore, they are computationally attractive since each Newton iteration requires the solution of a linear system of usually small dimension.
date: 2013-12
date_type: published
publisher: IEEE
pagerange: 2358-2363
event_title: Decision and Control (CDC), 2013 IEEE 52nd Annual Conference on
id_number: 10.1109/CDC.2013.6760233
refereed: TRUE
isbn: 978-1-4673-5714-2 
book_title: 52nd IEEE Conference on Decision and Control
official_url: http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=6760233&isnumber=6759837
citation:   Patrinos, Panagiotis and Bemporad, Alberto  Proximal Newton methods for convex composite optimization.   In:  52nd IEEE Conference on Decision and Control.   IEEE, pp. 2358-2363.  ISBN 978-1-4673-5714-2       (2013)