第5回

【第5回】富士通次世代コンピューティングセミナー

• 「SCIP Optimization Suite の紹介」
  • 品野 勇治/Shinano Yuji (Zuse Institute Berlin)

    平成９年３月東京理科大博士課程修了．平成９年４月より東京理科大助手．東京農工大准教授を経て，平成２２年４月よりZuse Institute Berlin (ZIB) 研究員

  • Abstract
    • 本講演では，今回一緒に講演するZIBの研究者の仕事内容を理解するために必要となるSCIP Optimization Suiteを紹介するとともに，研究分野に関して日本語により紹介する．
• 「The software development workflow – shown at the example of SCIP」
  • Franziska Schlösser (Zuse Institute Berlin／FICO)

    Franziska Schloesser earned her Master’s degree in Mathematics from TU Berlin in 2017. Afterwards she became the DevOps manager for the SCIP Optimization Suite at Zuse Institute Berlin. As DevOps manager, she was responsible for maintaining both the development and testing infrastructure as well as the performance evaluation tools. Since 2023, Franziska works at FICO Xpress Optimization as a Lead Engineer, supporting their CI/CD infrastructure and Quality Assurance. Her expertise spans over code management, various continuous integration environments, machine maintenance, and development workflow management.

  • Abstract
    • Software development is not just writing code. Especially in a larger development team, it is important to maintain a clear organization. We will talk about source code management, testing and deployment strategies, as well as performance evaluation and quality control. Showing the above at a real example, we will give an insight about how these measures are implemented in the development workflow of the SCIP solver.
• 「SCIP Optimization Suite: Cutting Plane Selection and Embedded ML Models」
  • Mark Ruben Turner (Zuse Institute Berlin)

    Mark Turner earned his Master’s degree in Mathematics and Statistics from The University of Melbourne in 2017. Following his degree’s completion, Mark begun research for his doctorate at TU Berlin and working in the EnergyLab at Zuse Institute Berlin. Near the end of 2023, Mark defended his dissertation titled Cutting Plane Selection for Mixed-Integer Linear Programming. Mark now works as a post-doc at Zuse Institute Berlin continuing research for the software Solving Constrained Integer Programs (SCIP), while working in the EnergyLab.

  • Abstract
    • In this talk we will be introducing the SCIP optimization suite, a popular open-source tool for formulating and solving mixed-integer programs. Following this introduction, depending on the time available, we will discuss recent advances in cutting plane selection algorithms for mixed-integer programming. Alternatively, we will discuss a growing field, where trained ML models are embedded as surrogate models in larger optimization processes. We will then show how SCIP can be used to both automatically formulate the embedded ML model into a mixed-integer program and then optimize the larger optimization process.

• 日時
  • 2024年02月21日（水）10:00-12:00
• 場所（ハイブリッド開催）
  • オフライン：広島大学工学部 102講義室
    • 東広島キャンパスマップ
  • オンライン：Microsoft Teams会議
• 参加方法
  • こちらの参加登録フォームよりお申し込みください
  • 学内からの参加は基本的に登録不要です（オンライン参加でURLがわからない場合はご登録ください）

現地約20名＋オンライン約25名の約45名に参加していただきました。