<?xml version="1.0" encoding="UTF-8"?>
<article article-type="research-article" dtd-version="1.3" xml:lang="ru" xmlns:xlink="http://www.w3.org/1999/xlink" xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:noNamespaceSchemaLocation="https://metafora.rcsi.science/xsd_files/journal3.xsd">
  <front>
    <journal-meta>
      <journal-id journal-id-type="publisher-id">moitvivt</journal-id>
      <journal-title-group>
        <journal-title xml:lang="ru">Моделирование, оптимизация и информационные технологии</journal-title>
        <trans-title-group xml:lang="en">
          <trans-title>Modeling, Optimization and Information Technology</trans-title>
        </trans-title-group>
      </journal-title-group>
      <issn pub-type="epub">2310-6018</issn>
      <publisher>
        <publisher-name>Издательство</publisher-name>
      </publisher>
    </journal-meta>
    <article-meta>
      <article-id pub-id-type="doi">10.26102/2310-6018/2026.57.6.007</article-id>
      <article-id pub-id-type="custom" custom-type="elpub">2283</article-id>
      <title-group>
        <article-title xml:lang="ru">Применение мультиагентных систем в задачах распределенной оптимизации</article-title>
        <trans-title-group xml:lang="en">
          <trans-title>Application of multi-agent systems in distributed optimization problems</trans-title>
        </trans-title-group>
      </title-group>
      <contrib-group>
        <contrib contrib-type="author" corresp="yes">
          <name-alternatives>
            <name name-style="eastern" xml:lang="ru">
              <surname>Азарнова</surname>
              <given-names>Татьяна Васильевна</given-names>
            </name>
            <name name-style="western" xml:lang="en">
              <surname>Azarnova</surname>
              <given-names>Тatyana Vasilevna</given-names>
            </name>
          </name-alternatives>
          <email>ivdas92@mail.ru</email>
          <xref ref-type="aff">aff-1</xref>
        </contrib>
        <contrib contrib-type="author">
          <name-alternatives>
            <name name-style="eastern" xml:lang="ru">
              <surname>Калишкин</surname>
              <given-names>Евгений Олегович</given-names>
            </name>
            <name name-style="western" xml:lang="en">
              <surname>Kalishkin</surname>
              <given-names>Evgeny Olegovich</given-names>
            </name>
          </name-alternatives>
          <email>e2697517@gmail.com</email>
          <xref ref-type="aff">aff-2</xref>
        </contrib>
      </contrib-group>
      <aff-alternatives id="aff-1">
        <aff xml:lang="ru">Воронежский государственный университет</aff>
        <aff xml:lang="en">Voronezh State University</aff>
      </aff-alternatives>
      <aff-alternatives id="aff-2">
        <aff xml:lang="ru">Воронежский государственный университет</aff>
        <aff xml:lang="en">Voronezh State University</aff>
      </aff-alternatives>
      <pub-date pub-type="epub">
        <day>01</day>
        <month>01</month>
        <year>2026</year>
      </pub-date>
      <volume>1</volume>
      <issue>1</issue>
      <elocation-id>10.26102/2310-6018/2026.57.6.007</elocation-id>
      <permissions>
        <copyright-statement>Copyright © Авторы, 2026</copyright-statement>
        <copyright-year>2026</copyright-year>
        <license license-type="creative-commons-attribution" xlink:href="https://creativecommons.org/licenses/by/4.0/">
          <license-p>This work is licensed under a Creative Commons Attribution 4.0 International License</license-p>
        </license>
      </permissions>
      <self-uri xlink:href="https://moitvivt.ru/ru/journal/article?id=2283"/>
      <abstract xml:lang="ru">
        <p>В данной статье проводится систематизация и детальный анализ современных методов распределенной оптимизации в мультиагентных системах (МАС). Мультиагентные подходы, определяемые как совокупность взаимодействующих автономных вычислительных сущностей, становятся критически востребованными в условиях, когда централизованная обработка данных невозможна из-за масштабов задач, жестких требований к скорости реакции или необходимости обеспечения приватности локальной информации. Цель работы заключается в комплексном исследовании ключевых подходов к децентрализованной оптимизации и выявлении фундаментальных факторов, определяющих их вычислительную эффективность, устойчивость к сбоям и практическую применимость. В рамках исследования подробно рассмотрены пять основных классов алгоритмов: консенсусный градиентный спуск (DGD), методы отслеживания градиента (Gradient Tracking), распределенный метод чередующихся направлений множителей (ADMM), а также современные стохастические и коммуникационно-эффективные подходы, включая Local SGD и FedAvg. В статье детально проанализированы системные ограничения, накладываемые топологией графа связности, алгоритмами компрессии данных и строгими требованиями дифференциальной приватности. Раскрыты ключевые теоретические и практические аспекты построения оптимальных процедур на основе сбалансированного сочетания архитектуры сети, характера локальных функций стоимости и доступной пропускной способности каналов связи. Сформулированы рекомендации по выбору конкретных алгоритмических решений в зависимости от специфики прикладной среды.</p>
      </abstract>
      <trans-abstract xml:lang="en">
        <p>This paper systematizes and provides a detailed analysis of modern distributed optimization methods in multi-agent systems (MAS). Multi-agent approaches, defined as a set of interacting autonomous computing entities, are becoming critically in demand when centralized data processing is impossible due to the scale of problems, strict response time requirements, or the need to ensure the privacy of local information. The goal of the work is to comprehensively investigate key approaches to decentralized optimization and identify the fundamental factors that determine their computational efficiency, fault tolerance, and practical applicability. Within the framework of the study, five main classes of algorithms are considered in detail: consensus gradient descent (DGD), gradient tracking methods, the distributed alternating direction method of multipliers (ADMM), as well as modern stochastic and communication-efficient approaches, including Local SGD and FedAvg. The paper thoroughly analyzes systemic limitations imposed by the connectivity graph topology, data compression algorithms, and strict differential privacy requirements. Key theoretical and practical aspects of constructing optimal procedures based on a balanced combination of network architecture, the nature of local cost functions, and available communication channel bandwidth are revealed. Recommendations for choosing specific algorithmic solutions depending on the specifics of the application environment are formulated.</p>
      </trans-abstract>
      <kwd-group xml:lang="ru">
        <kwd>мультиагентные системы</kwd>
        <kwd>распределенная оптимизация</kwd>
        <kwd>консенсусный градиентный спуск</kwd>
        <kwd>отслеживание градиента</kwd>
        <kwd>коммуникационная эффективность</kwd>
        <kwd>топология графа</kwd>
        <kwd>дифференциальная приватность</kwd>
      </kwd-group>
      <kwd-group xml:lang="en">
        <kwd>multi-agent systems</kwd>
        <kwd>distributed optimization</kwd>
        <kwd>consensus gradient descent</kwd>
        <kwd>gradient tracking</kwd>
        <kwd>communication efficiency</kwd>
        <kwd>graph topology</kwd>
        <kwd>differential privacy</kwd>
      </kwd-group>
      <funding-group>
        <funding-statement xml:lang="ru">Исследование выполнено без спонсорской поддержки.</funding-statement>
        <funding-statement xml:lang="en">The study was performed without external funding.</funding-statement>
      </funding-group>
    </article-meta>
  </front>
  <back>
    <ref-list>
      <title>References</title>
      <ref id="cit1">
        <label>1</label>
        <mixed-citation xml:lang="ru">Yang T., Yi X., Wu J., et al. A survey of distributed optimization. Annual Reviews in Control. 2019;47:278–305. https://doi.org/10.1016/j.arcontrol.2019.05.006</mixed-citation>
      </ref>
      <ref id="cit2">
        <label>2</label>
        <mixed-citation xml:lang="ru">Li J., Su H. Gradient Tracking: A Unified Approach to Smooth Distributed Optimization. arXiv. URL: https://arxiv.org/abs/2202.09804 [Accessed 25th December 2025].</mixed-citation>
      </ref>
      <ref id="cit3">
        <label>3</label>
        <mixed-citation xml:lang="ru">Yuan K., Ying B., Zhao X., et al. Exact Diffusion for Distributed Optimization and Learning – Part I: Algorithm Development. arXiv. URL: https://arxiv.org/abs/1702.05122 [Accessed 25th December 2025].</mixed-citation>
      </ref>
      <ref id="cit4">
        <label>4</label>
        <mixed-citation xml:lang="ru">Boyd S., Parikh N., Chu E., et al. Distributed optimization and statistical learning via the alternating direction method of multipliers. Foundations and Trends® in Machine Learning. 2011;3(1):1–122. https://doi.org/10.1561/2200000016</mixed-citation>
      </ref>
      <ref id="cit5">
        <label>5</label>
        <mixed-citation xml:lang="ru">Halsted T., Shorinwa O., Yu J., et al. A survey of distributed optimization methods for multi-robot systems. arXiv. URL: https://arxiv.org/abs/2103.12840 [Accessed 25th December 2025].</mixed-citation>
      </ref>
      <ref id="cit6">
        <label>6</label>
        <mixed-citation xml:lang="ru">McMahan B., Moore E., Ramage D., et al. Communication-efficient learning of deep networks from decentralized data. In: Proceedings of the 20th International Conference on Artificial Intelligence and Statistics (AISTATS 2017), 20–22 April 2017, Fort Lauderdale, FL, USA. PMLR; 2017. P. 1273–1282.</mixed-citation>
      </ref>
      <ref id="cit7">
        <label>7</label>
        <mixed-citation xml:lang="ru">Wang Y., Lin H., Lam J., et al. Differentially private consensus and distributed optimization in multi-agent systems: A review. Neurocomputing. 2024;597:127986. https://doi.org/10.1016/j.neucom.2024.127986</mixed-citation>
      </ref>
      <ref id="cit8">
        <label>8</label>
        <mixed-citation xml:lang="ru">Alistarh D., Grubic D., Li J., et al. QSGD: Communication-efficient SGD via gradient quantization and encoding. In: Advances in Neural Information Processing Systems 30: Annual Conference on Neural Information Processing Systems, 04–09 December 2017, Long Beach, CA, USA. 2017. P. 1709–1720.</mixed-citation>
      </ref>
      <ref id="cit9">
        <label>9</label>
        <mixed-citation xml:lang="ru">Koloskova A., Lin T., Stich S.U., et al. Decentralized deep learning with arbitrary communication compression. arXiv. URL: https://arxiv.org/abs/1907.09356 [Accessed 25th December 2025].</mixed-citation>
      </ref>
      <ref id="cit10">
        <label>10</label>
        <mixed-citation xml:lang="ru">Yurdem B., Kuzlu M., Gullu M.K., et al. Federated learning: Overview, strategies, applications, tools and future directions. Heliyon. 2024;10(19):e38137. https://doi.org/10.1016/j.heliyon.2024.e38137</mixed-citation>
      </ref>
      <ref id="cit11">
        <label>11</label>
        <mixed-citation xml:lang="ru">Wang Zh., Wang Ch., Wang J., et al. An accelerated exact distributed first-order algorithm for optimization over directed networks. Journal of the Franklin Institute. 2023;360(14):10706–10727. https://doi.org/10.1016/j.jfranklin.2023.08.015</mixed-citation>
      </ref>
      <ref id="cit12">
        <label>12</label>
        <mixed-citation xml:lang="ru">Xin R., Pu Sh., Nedić A. A general framework for decentralized optimization with first-order methods. Proceedings of the IEEE. 2020;108(11):1869–1889. https://doi.org/10.1109/JPROC.2020.3024266</mixed-citation>
      </ref>
      <ref id="cit13">
        <label>13</label>
        <mixed-citation xml:lang="ru">Lian X., Zhang C., Zhang H., et al. Can decentralized algorithms outperform centralized algorithms? A case study for decentralized parallel stochastic gradient descent. In: Advances in Neural Information Processing Systems 30: Annual Conference on Neural Information Processing Systems, 04–09 December 2017, Long Beach, CA, USA. 2017. P. 5330–5340.</mixed-citation>
      </ref>
      <ref id="cit14">
        <label>14</label>
        <mixed-citation xml:lang="ru">Nedic A., Ozdaglar A. Distributed subgradient methods for multi-agent optimization. IEEE Transactions on Automatic Control. 2009;54(1):48–61. https://doi.org/10.1109/TAC.2008.2009515</mixed-citation>
      </ref>
    </ref-list>
    <fn-group>
      <fn fn-type="conflict">
        <p>The authors declare that there are no conflicts of interest present.</p>
      </fn>
    </fn-group>
  </back>
</article>