<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE root>
<article xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns:ali="http://www.niso.org/schemas/ali/1.0/" article-type="research-article" dtd-version="1.2" xml:lang="en"><front><journal-meta><journal-id journal-id-type="publisher-id">Доклады Академии наук</journal-id><journal-title-group><journal-title xml:lang="en">Доклады Академии наук</journal-title><trans-title-group xml:lang="ru"><trans-title>Доклады Академии наук</trans-title></trans-title-group></journal-title-group><issn publication-format="print">0869-5652</issn><publisher><publisher-name xml:lang="en">The Russian Academy of Sciences</publisher-name></publisher></journal-meta><article-meta><article-id pub-id-type="publisher-id">12803</article-id><article-id pub-id-type="doi">10.31857/S0869-5652485115-18</article-id><article-categories><subj-group subj-group-type="toc-heading" xml:lang="en"><subject>Mathematics</subject></subj-group><subj-group subj-group-type="toc-heading" xml:lang="ru"><subject>Математика</subject></subj-group><subj-group subj-group-type="article-type"><subject>Research Article</subject></subj-group></article-categories><title-group><article-title xml:lang="en">Primal-dual accelerated gradient descent with line search for convex and nonconvex optimization problems</article-title><trans-title-group xml:lang="ru"><trans-title>Прямо-двойственный ускоренный градиентный метод с одномерным поиском для выпуклых, невыпуклых и негладких задач оптимизации</trans-title></trans-title-group></title-group><contrib-group><contrib contrib-type="author"><name-alternatives><name xml:lang="en"><surname>Guminov</surname><given-names>S. V.</given-names></name><name xml:lang="ru"><surname>Гуминов</surname><given-names>С. В.</given-names></name></name-alternatives><address><country country="RU">Russian Federation</country></address><email>sergey.guminov@phystech.edu</email><xref ref-type="aff" rid="aff1"/><xref ref-type="aff" rid="aff2"/></contrib><contrib contrib-type="author"><name-alternatives><name xml:lang="en"><surname>Nesterov</surname><given-names>Yu. E.</given-names></name><name xml:lang="ru"><surname>Нестеров</surname><given-names>Ю. Е.</given-names></name></name-alternatives><address><country country="RU">Russian Federation</country></address><email>sergey.guminov@phystech.edu</email><xref ref-type="aff" rid="aff3"/><xref ref-type="aff" rid="aff4"/></contrib><contrib contrib-type="author"><name-alternatives><name xml:lang="en"><surname>Dvurechensky</surname><given-names>P. E.</given-names></name><name xml:lang="ru"><surname>Двуреченский</surname><given-names>П. Е.</given-names></name></name-alternatives><address><country country="RU">Russian Federation</country></address><email>sergey.guminov@phystech.edu</email><xref ref-type="aff" rid="aff2"/><xref ref-type="aff" rid="aff5"/></contrib><contrib contrib-type="author"><name-alternatives><name xml:lang="en"><surname>Gasnikov</surname><given-names>A. V.</given-names></name><name xml:lang="ru"><surname>Гасников</surname><given-names>А. В.</given-names></name></name-alternatives><address><country country="RU">Russian Federation</country></address><email>sergey.guminov@phystech.edu</email><xref ref-type="aff" rid="aff1"/><xref ref-type="aff" rid="aff2"/></contrib></contrib-group><aff-alternatives id="aff1"><aff><institution xml:lang="en">Moscow Institute of Physics and Technology</institution></aff><aff><institution xml:lang="ru">Московский физико-технический институт (государственный университет)</institution></aff></aff-alternatives><aff-alternatives id="aff2"><aff><institution xml:lang="en">Institute for Information Transmission Problems of the Russian Academy of Sciences</institution></aff><aff><institution xml:lang="ru">Институт проблем передачи информации Российской Академии наук</institution></aff></aff-alternatives><aff id="aff3"><institution>Center for Operations Research and Econometrics (CORE), Catholic University of Louvain</institution></aff><aff-alternatives id="aff4"><aff><institution xml:lang="en">Higher School of Economics</institution></aff><aff><institution xml:lang="ru">Национальный исследовательский университет “Высшая школа экономики”</institution></aff></aff-alternatives><aff id="aff5"><institution>Weierstrass Institute for Applied Analysis and Stochastics</institution></aff><pub-date date-type="pub" iso-8601-date="2019-05-19" publication-format="electronic"><day>19</day><month>05</month><year>2019</year></pub-date><volume>485</volume><issue>1</issue><issue-title xml:lang="en"/><issue-title xml:lang="ru"/><fpage>15</fpage><lpage>18</lpage><history><date date-type="received" iso-8601-date="2019-05-22"><day>22</day><month>05</month><year>2019</year></date><date date-type="accepted" iso-8601-date="2019-05-22"><day>22</day><month>05</month><year>2019</year></date></history><permissions><copyright-statement xml:lang="en">Copyright ©; 2019, Russian academy of sciences</copyright-statement><copyright-statement xml:lang="ru">Copyright ©; 2019, Российская академия наук</copyright-statement><copyright-year>2019</copyright-year><copyright-holder xml:lang="en">Russian academy of sciences</copyright-holder><copyright-holder xml:lang="ru">Российская академия наук</copyright-holder></permissions><self-uri xlink:href="https://journals.eco-vector.com/0869-5652/article/view/12803">https://journals.eco-vector.com/0869-5652/article/view/12803</self-uri><abstract xml:lang="en"><p>In this paper a new variant of accelerated gradient descent is proposed. The proposed method does not require any information about the objective function, uses exact line search for the practical accelerations of convergence, converges according to the well-known lower bounds for both convex and non-convex objective functions and possesses primal-dual properties. We also provide a universal version of said method, which converges according to the known lower bounds for both smooth and non-smooth problems.</p></abstract><trans-abstract xml:lang="ru"><p>Представлен новый ускоренный градиентный метод оптимизации. Данный метод не требует никакой априорной информации о целевой функции, использует процедуру одномерного поиска для ускорения сходимости на практике, сходится согласно известным нижним оценкам как для выпуклых, так и невыпуклых целевых функций и обладает свойством прямо-двойственности. Также представлена универсальная версия данного метода.</p></trans-abstract><kwd-group xml:lang="en"><kwd>accelerated gradient descent</kwd><kwd>line-search</kwd><kwd>primal-dual methods</kwd><kwd>convex optimization</kwd><kwd>nonconvex optimization</kwd></kwd-group><kwd-group xml:lang="ru"><kwd>ускоренный градиентный спуск</kwd><kwd>одномерный поиск</kwd><kwd>прямо-двойственные методы</kwd><kwd>выпуклая оптимизация</kwd><kwd>невыпуклая оптимизация</kwd></kwd-group><funding-group><award-group><award-id></award-id></award-group><funding-statement xml:lang="ru">Исследование выполнено за счёт гранта российского научного фонда (проект 18–71–10108).</funding-statement></funding-group></article-meta></front><body></body><back><ref-list><ref id="B1"><label>1.</label><mixed-citation>Нестеров Ю.Е. Эффективные методы в нелиней- ном программировании. М.: радио и связь, 1989.</mixed-citation></ref><ref id="B2"><label>2.</label><mixed-citation>Waltz R. A., Morales J.L., Nocedal J., Orban D. An Interior Algorithm for Nonlinear Optimization that Combines Line Search and Trust Region Steps //Math. Progr. 2006. V. 107. № 3. P. 391–408.</mixed-citation></ref><ref id="B3"><label>3.</label><mixed-citation>Narkiss G., Zibulevsky M. Sequential Subspace Optimization Method for Large-Scale Unconstrained Problems. Tech. Report CCIT № 559. Haifa: EE Dept.Technion, 2005.</mixed-citation></ref><ref id="B4"><label>4.</label><mixed-citation>Nesterov Yu. Smooth Minimization of Non-Smooth Functions // Math. Progr. 2005. V. 103. № 1. P. 127–152.</mixed-citation></ref><ref id="B5"><label>5.</label><mixed-citation>Nesterov Yu. Primal-Dual Subgradient Methods for Convex Problems // Math. Progr. 2009. V. 120. № 1. P. 221–259.</mixed-citation></ref><ref id="B6"><label>6.</label><mixed-citation>Dvurechensky P., Gasnikov A., Kroshnin A. Computational Optimal Transport: Complexity by Accelerated Gradient Descent Is Better Than by Sinkhorn’s Algorithm. Proc. of the 35th International Conference on Machine Learning. Stockholm: PMLR, 2018. P. 1367–1376.</mixed-citation></ref><ref id="B7"><label>7.</label><mixed-citation>Dvurechensky P., Gasnikov A., Matsievsky S., Rodomanov A., Usik I. Primal-Dual Method for Searching Equilibrium in Hierarchical Congestion Population Games CEUR-WS // Supplementary Proc. the 9th In- tern. Conf. on Discrete Optimization and Operations Research and Scientific School (DOOR 2016). B.: Springer, 2016. P. 584–595. http://ceurws.org/Vol-1623/</mixed-citation></ref><ref id="B8"><label>8.</label><mixed-citation>Chernov A., Dvurechensky P., Gasnikov A. Fast Primal- Dual Gradient Method for Strongly Convex Minimiza- tion Problems with Linear Constraints // Proceeding of the 9th Intern. Conf. on Discrete Optimization and Operations Research (DOOR 2016). B.: Springer, 2016. P. 391–403.</mixed-citation></ref><ref id="B9"><label>9.</label><mixed-citation>Нестеров Ю.Е. Метод решения задач выпуклого программирования с трудоемкостью O(1/k2) // ДАН. 1983. Т. 269. № 3. С. 543–547.</mixed-citation></ref><ref id="B10"><label>10.</label><mixed-citation>Нестеров Ю.Е. Введение в выпуклую оптимиза- цию. М.: МЦНМО, 2010. 280 с.</mixed-citation></ref><ref id="B11"><label>11.</label><mixed-citation>Allen-Zhu Z., Orecchia L. Linear Coupling: An Ulti- mate Unification of Gradient and Mirror Descent. Proc. of the 8th Innovations in Theoretical Computer Science. Saarbrücken: Schloss Dagstuhl, 2017.</mixed-citation></ref><ref id="B12"><label>12.</label><mixed-citation>Nesterov Yu. Universal Gradient Methods for Convex Optimization Problems // Math. Progr. 2015. V. 152. № 1/2. P. 381–404.</mixed-citation></ref><ref id="B13"><label>13.</label><mixed-citation>Guminov S., Gasnikov A., Anikin A., Gornov A. A Uni- versal Modification of the Linear Coupling Method // Optim. Met. and Soft. 2018. DOI: 10.1080/10556788. 2018.1517158.</mixed-citation></ref></ref-list></back></article>
