• Search Research Projects
  • Search Researchers
  • How to Use
  1. Back to previous page

Development of a dynamic load balancing method based on prediction by cooperative use of simulation and machine learning

Research Project

Project/Area Number 20K21787
Research Category

Grant-in-Aid for Challenging Research (Exploratory)

Allocation TypeMulti-year Fund
Review Section Medium-sized Section 60:Information science, computer engineering, and related fields
Research InstitutionThe University of Tokyo

Principal Investigator

Shimokawabe Takashi  東京大学, 情報基盤センター, 准教授 (40636049)

Project Period (FY) 2020-07-30 – 2024-03-31
Project Status Completed (Fiscal Year 2023)
Budget Amount *help
¥6,500,000 (Direct Cost: ¥5,000,000、Indirect Cost: ¥1,500,000)
Fiscal Year 2022: ¥2,210,000 (Direct Cost: ¥1,700,000、Indirect Cost: ¥510,000)
Fiscal Year 2021: ¥2,210,000 (Direct Cost: ¥1,700,000、Indirect Cost: ¥510,000)
Fiscal Year 2020: ¥2,080,000 (Direct Cost: ¥1,600,000、Indirect Cost: ¥480,000)
Keywordsステンシル計算 / 高性能計算 / 機械学習 / 適合細分化格子法 / 高生産フレームワーク / 動的負荷分散
Outline of Research at the Start

流体計算などの格子に基づく計算では、高精度が必要な領域をより高精細な格子で計算できる適合細分化格子(AMR)法がマルチスケール問題解決の鍵となる技術として注目されている。本研究では、大規模GPUスパコンで従来と比較して極めて高性能なAMR計算を実現するため、機械学習によりシミュレーションの「未来」の結果を予測し、それに基づき計算負荷を動的に分散する手法を開発する。開発手法を課題代表者らが開発中のAMR法フレームワークへ導入し、様々な実アプリケーションへ適用する。本研究では、従来のシミュレーションだけでは不可能であった「未来」の予測を含めた時系列変化を考慮した負荷分散を実現することを目指す。

Outline of Final Research Achievements

Recently, adaptive mesh refinement (AMR), which is well suited for GPU computation, has been attracting attention because it can locally refine regions where high accuracy is required. This research aims to develop a method to predict "future" results of simulations by machine learning and to dynamically balance the load of the simulations based on these predictions. We have realized the prediction of fluid simulations by deep learning, the construction of optimal domain decomposition methods based on the amount of computation and communication, the improvement of the AMR method framework, and applied it to the lattice Boltzmann method. We have found machine learning to be helpful in predicting simulations.

Academic Significance and Societal Importance of the Research Achievements

格子計算はスパコンを利用する代表的なアプリケーションで、局所的に高精細な大規模計算を実現させる意義は大きい。 米国エネルギー省は、AMR法は所謂「エクサスケール」でのマルチスケール問題解決の鍵となる技術と位置付けている。本研究では、機械学習という全く異なるアプローチで数値計算結果を予測する。本研究の目標は予測に基づいた動的負荷分散の実現であるが、近似的ではあるが超高速な予測が可能である機械学習は計算科学分野の様々な要素技術で従来手法を凌駕する可能性を秘めており、本研究でその有用性を示した意義は大きい。

Report

(5 results)
  • 2023 Annual Research Report   Final Research Report ( PDF )
  • 2022 Research-status Report
  • 2021 Research-status Report
  • 2020 Research-status Report
  • Research Products

    (23 results)

All 2024 2023 2022 2021 2020

All Journal Article (4 results) (of which Peer Reviewed: 1 results,  Open Access: 1 results) Presentation (19 results) (of which Int'l Joint Research: 8 results,  Invited: 3 results)

  • [Journal Article] 複数GPUでの埋め込み境界-格子ボルツマン法の計算の最適化と性能モデルの構築2023

    • Author(s)
      畠山 昂, 下川辺 隆史
    • Journal Title

      研究報告ハイパフォーマンスコンピューティング(HPC)

      Volume: 2023-HPC-188 Pages: 1-9

    • Related Report
      2022 Research-status Report
  • [Journal Article] AMR-Net: Convolutional Neural Networks for Multi-resolution Steady Flow Prediction2021

    • Author(s)
      Asahi Yuuichi、Hatayama Sora、Shimokawabe Takashi、Onodera Naoyuki、Hasegawa Yuta、Idomura Yasuhiro
    • Journal Title

      The 2nd Workshop on Artificial Intelligence and Machine Learning for Scientific Applications, IEEE Cluster 2021

      Volume: - Pages: 686-691

    • DOI

      10.1109/cluster48925.2021.00102

    • Related Report
      2021 Research-status Report
    • Peer Reviewed / Open Access
  • [Journal Article] 深層学習と境界交換を用いた複数領域にまたがる定常流のシミュレーション結果の予測2020

    • Author(s)
      畑山そら, 下川辺隆史, 小野寺直幸
    • Journal Title

      研究報告ハイパフォーマンスコンピューティング(HPC)

      Volume: 2020-HPC-175 Pages: 1-7

    • Related Report
      2020 Research-status Report
  • [Journal Article] 深層学習による混相流の時間発展シミュレーション結果の予測手法の検討2020

    • Author(s)
      長谷川敦,下川辺隆史
    • Journal Title

      研究報告ハイパフォーマンスコンピューティング(HPC)

      Volume: 2020-HPC-177 Pages: 1-7

    • Related Report
      2020 Research-status Report
  • [Presentation] Accelerating Lattice Boltzmann method with C++ standard language parallel algorithm2024

    • Author(s)
      Ziheng Yuan and Takashi Shimokawabe
    • Organizer
      International Conference on High Performance Computing in Asia-Pacific Region (HPCAsia) 2024
    • Related Report
      2023 Annual Research Report
    • Int'l Joint Research
  • [Presentation] oneAPIを用いた様々なデバイス上でのステンシル計算の実装2023

    • Author(s)
      佐久間 大我、下川辺 隆史、大森 拓郎
    • Organizer
      第28回計算工学講演会
    • Related Report
      2023 Annual Research Report
  • [Presentation] Accelerating lattice Boltzmann method simulation with GPU computation using C++ standard language parallelism2023

    • Author(s)
      Ziheng Yuan, Takashi Shimokawabe
    • Organizer
      第28回計算工学講演会
    • Related Report
      2023 Annual Research Report
  • [Presentation] Accelerating lattice Boltzmann method with GPU and C++ standard parallelization2023

    • Author(s)
      Ziheng Yuan, Takashi Shimokawabe
    • Organizer
      10th International Congress on Industrial and Applied Mathematics
    • Related Report
      2023 Annual Research Report
    • Int'l Joint Research
  • [Presentation] 深層学習を用いたシミュレーション結果を予測する代理モデル開発の取り組み2023

    • Author(s)
      下川辺隆史
    • Organizer
      第7回HPCものづくり統合ワークショップ
    • Related Report
      2023 Annual Research Report
    • Invited
  • [Presentation] 複数GPUを用いる際の埋め込み境界-格子ボルツマン法の性能向上2022

    • Author(s)
      畠山 昂, 下川辺 隆史
    • Organizer
      第27回計算工学講演会
    • Related Report
      2022 Research-status Report
  • [Presentation] OpenMP Offloadingを用いたGPUでの格子ボルツマン法実行における性能評価2022

    • Author(s)
      大森 拓郎, 下川辺 隆史, 朝比 祐一
    • Organizer
      第27回計算工学講演会
    • Related Report
      2022 Research-status Report
  • [Presentation] Performance Optimization Of Lattice Boltzmann Method On A64FX2022

    • Author(s)
      Takuro Omori, Takashi Shimokawabe
    • Organizer
      15th World Congress on Computational Mechanics & 8th Asian Pacific Congress on Computational Mechanics
    • Related Report
      2022 Research-status Report
    • Int'l Joint Research
  • [Presentation] Performance improvement of immersed boundary-lattice Boltzmann method on multiple GPUs2022

    • Author(s)
      Akira Hatakeyama, Takashi Shimokawabe
    • Organizer
      15th World Congress on Computational Mechanics & 8th Asian Pacific Congress on Computational Mechanics
    • Related Report
      2022 Research-status Report
    • Int'l Joint Research
  • [Presentation] 深層学習による流体シミュレーション結果の予測2022

    • Author(s)
      下川辺 隆史
    • Organizer
      第35回計算力学講演会
    • Related Report
      2022 Research-status Report
    • Invited
  • [Presentation] Acoustic simulation using lattice Boltzmann method by multi-GPU parallel computing2022

    • Author(s)
      Shota Suzuki, Takashi Shimokawabe
    • Organizer
      International Conference on High Performance Computing in Asia-Pacific Region (HPCAsia) 2022
    • Related Report
      2021 Research-status Report
    • Int'l Joint Research
  • [Presentation] 格子ボルツマン法によるインピーダンス境界を用いた音響解析手法の構築2022

    • Author(s)
      鈴木 翔太, 下川辺 隆史
    • Organizer
      日本音響学会 2022年春季研究発表会
    • Related Report
      2021 Research-status Report
  • [Presentation] Multi-GPU computing of moving boundary flow using lattice Boltzmann method2022

    • Author(s)
      Akira Hatakeyama, Takashi Shimokawabe
    • Organizer
      International Conference on High Performance Computing in Asia-Pacific Region (HPCAsia) 2022
    • Related Report
      2021 Research-status Report
    • Int'l Joint Research
  • [Presentation] 深層学習による流体シミュレーション結果予測2022

    • Author(s)
      下川辺隆史
    • Organizer
      第41回計算数理工学フォーラム
    • Related Report
      2021 Research-status Report
    • Invited
  • [Presentation] 格子ボルツマン法に基づくGPUを用いた音響解析2021

    • Author(s)
      鈴木翔太, 下川辺隆史
    • Organizer
      第26回計算工学講演会
    • Related Report
      2021 Research-status Report
  • [Presentation] 埋め込み境界法を適用した格子ボルツマン法に基づく3次元音響解析2021

    • Author(s)
      鈴木 翔太, 下川辺 隆史
    • Organizer
      オープンCAEシンポジウム2021
    • Related Report
      2021 Research-status Report
  • [Presentation] 畳み込みニューラルネットワークと境界交換を用いた複数領域にまたがる定常流のシミュレーション結果の予測2020

    • Author(s)
      畑山そら, 下川辺隆史, 小野寺直幸
    • Organizer
      第25回計算工学講演会
    • Related Report
      2020 Research-status Report
  • [Presentation] Steady Flow Prediction across Multiple Regions using Deep Learning and Boundary Exchange2020

    • Author(s)
      Sora Hatayama, Takashi Shimokawabe and Naoyuki Onodera
    • Organizer
      3rd International Conference on Computational Engineering and Science for Safety and Environmental Problems
    • Related Report
      2020 Research-status Report
    • Int'l Joint Research
  • [Presentation] High-Resolution Simulations using an AMR Framework on GPU Supercomputers2020

    • Author(s)
      Takashi Shimokawabe and Naoyuki Onodera
    • Organizer
      3rd International Conference on Computational Engineering and Science for Safety and Environmental Problems
    • Related Report
      2020 Research-status Report
    • Int'l Joint Research

URL: 

Published: 2020-08-03   Modified: 2025-01-30  

Information User Guide FAQ News Terms of Use Attribution of KAKENHI

Powered by NII kakenhi