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

2021 Fiscal Year Final Research Report

Development of error resilience technology for complex moment-based parallel eigensolvers

Research Project

  • PDF
Project/Area Number 17K12690
Research Category

Grant-in-Aid for Young Scientists (B)

Allocation TypeMulti-year Fund
Research Field High performance computing
Research InstitutionUniversity of Tsukuba

Principal Investigator

Imakura Akira  筑波大学, システム情報系, 准教授 (60610045)

Project Period (FY) 2017-04-01 – 2022-03-31
Keywords数値解析 / 高性能計算 / 耐障害技術
Outline of Final Research Achievements

In this research project, we focused on the complex moment-based eigensolvers, which are massively parallel eigenvalue analysis methods originated in Japan and have been actively studied in recent years. We developed mathematical fault-tolerance techniques through theoretical error analysis of the method, aiming to realize fault-tolerance at the algorithm level. Theoretical and experimental evaluations have shown that the developed technology can obtain eigenpairs with high speed and accuracy even in the presence of failures. In addition, we have also developed technologies to improve usability for practical use. The results of this research project were presented at domestic and international conferences and submitted to international journals.

Free Research Field

数値計算、機械学習

Academic Significance and Societal Importance of the Research Achievements

近年のスーパーコンピュータの大規模・高性能化に伴い、システムの故障率が上昇しており、特に次世代スーパーコンピュータでの長時間シミュレーションの困難さが懸念されている。このため、システムレベルでの対策とともに、システム障害によらず正しい計算結果を与える「アルゴリズムレベルでの耐障害性」の実現が近年重要視されている。各種のシミュレーションで基盤となる固有値解析において、アルゴリズムレベルでの耐障害性を実現する本研究課題の成果は、次世代スーパーコンピュータ上での長時間シミュレーションの実現に寄与し、幅広い応用分野において、シミュレーションの大規模化・高精度化につながることが期待される。

URL: 

Published: 2023-01-30  

Information User Guide FAQ News Terms of Use Attribution of KAKENHI

Powered by NII kakenhi