092-802-4402

092-802-4405

Ph.D. Sc.

No

Mathematical Optimization, Graph Analysis, High Performance Computing, Data Science, Deep Learning

0000-0001-8549-641X

00years00months

The objective of our ongoing research project is to develop an advanced computing and optimization infrastructure for extremely large-scale graphs on post peta-scale supercomputers. Recent emergence of extremely large-scale graphs in various application fields, such as transportation, social network, cyber-security, and bioinformatics, etc., requires fast and scalable analysis. We propose a new framework of software stacks for extremely large-scale graph analysis systems, such as parallel graph analysis and optimization libraries on multiple CPUs and GPUs, hierarchal graph stores using non-volatile memory (NVM) devices, and graph processing and visualization systems.
We explain our challenge to Graph 500 and Green Graph 500 benchmarks that are designed to measure the performance of a computer system for applications that require irregular memory and network access patterns. The 1st Graph500 list was released in November 2010. The Graph500 benchmark measures the performance of any supercomputer performing a BFS (Breadth-First Search) in terms of traversed edges per second (TEPS). In 2014 and 2020, by utilizing Fugaku and K supercomputer, our project team was a winner of the eighth, 10th to 18th, 20th, and 23rd Graph500 and the 3rd to 6th Green Graph500 benchmarks, respectively.
We also present our parallel implementation for large-scale mathematical optimization problems. In the last decade, mathematical optimization programming (MOP) problems have been intensively studied in both their theoretical and practical aspect in a wide range of fields, such as combinatorial optimization, structural optimization, control theory, economics, quantum chemistry, sensor network location, data mining, and machine learning. The semidefinite programming (SDP) problem is a predominant problem in mathematical optimization. We demonstrate that our implementation is a high-performance general solver for SDPs in various application fields through numerical experiments at the TSUBAME 2.5 supercomputer in the Tokyo Institute of Technology, and we solved the largest SDP problem, thereby creating a new world record.　For these achievements, he received the Commendation for Science and Technology by the Minister of Education, Culture, Sports, Science and Technology (Research Category) in 2017.

Research Interests

• Massive Parallelization for Finding Shortest LatticeVectors Based on Ubiquity Generator Framework
  keyword : Lattice based cryptography, Shortest vectorproblem, Parallel computation, DeepBKZ, ENUM, UbiquityGenerator Framework
  2019.04.
• Industrial Application Development in Cyber Physical
  keyword : CPS, AI, BigData, HPC, Optimization
  2019.04.
• Infrastructure for Extremely Large-Scale Graphs on Post Peta-Scale Supercomputers
  keyword : Mathematical Optimization, Graph Analysis, High Performance Computing
  2011.10.
• Graph Analysis and High Performance Computing Techniques for Realizing Urban OS
  keyword : Urban OS
  2015.09.
• High Performance Computing for Mathematical Optimization Problems
  keyword : Optimization Problem
  1995.04.
• A Challenge to Graph 500 and Green Graph 500 benchmarks
  keyword : High Performance Computing
  2015.09.
• Development of High Performance Graph Search and Optimization Library for Graph Analysis
  keyword : Graph Analysis
  2015.09.

Current and Past Project

• We present our ongoing research project that is supported by the Japan Science and Technology Agency (JST), the Core Research of Evolutionary Science and Technology (CREST). The objective of this project is to develop an ad- vanced computing and optimization infrastructure for extremely large-scale graphs on post peta-scale supercomputers. We explain our challenge to Graph 500 and Green Graph 500 benchmarks that are designed to measure the performance of a computer system for applications that require irregular memory and network access patterns. Following its announcement in June 2010, the Graph500 list was released in November 2010. The list has been updated biannually ever since. The Graph500 benchmark mea- sures the performance of any supercomputer performing a BFS (Breadth-First Search) in terms of traversed edges per second (TEPS). We have implemented world ’s first GPU-based BFS on the TSUBAME 2.0 supercomputer at Tokyo Institute of Tech- nology in 2012. The Green Graph 500 list collects TEPS-per-watt metrics. In ISC14, our project team was a winner of the 8th Graph500 benchmark and 3rd Green Graph 500 benchmark. We also present our parallel implementation for large-scale SDP (SemiDefinite Programming) problem. We solved the largest SDP problem (which has over 2.33 million constraints), thereby creating a new world record. Our im- plementation also achieved 1.713 PFlops in double precision for large-scale Cholesky factorization using 2,720 CPUs and 4,080 GPUs on the TSUBAME 2.5 supercomputer.

Academic Activities Membership in Academic Society

• INFORMS
• THE JAPAN SOCIETY FOR INDUSTRIAL AND APPLIED MATHEMATICS
• INFORMATION PROCESSING SOCIETY OF JAPAN
• SIAM
• IEEE

Awards

• A joint research group consisting of Professor Katsuki Fujisawa of Kyushu University's Institute of Mass-for-Industry Research, RIKEN, Fixtures, Inc. and Fujitsu Limited has successfully improved the performance of the supercomputer "Fugaku" in the international "Graph500" performance ranking for large-scale graph analysis using the full specifications of the supercomputer "Fugaku". The research group, consisting of Fixed Stars, Inc. and Fujitsu Limited, succeeded in improving the performance of the supercomputer "Fugaku" in the "Graph500," an international performance ranking of supercomputers for large-scale graph analysis, and ranked No. 1 in the world for six consecutive terms following June 2022.
  This ranking was announced by the Graph500 Committee on November 15 (November 16, Japan time) in conjunction with SC22, an international conference on high-performance computing (HPC).
  The performance of large-scale graph analysis is an important indicator in the analysis of big data, which requires large-scale and complex data processing, and we are promoting application development (smart factories, MaaS (delivery optimization), etc.) with private companies using technology that has achieved the world's highest performance.
• A joint research group consisting of Professor Katsuki Fujisawa of Kyushu University's Institute of Mass-for-Industry Research, RIKEN, Fixtures, Inc. and Fujitsu Limited has successfully improved the performance of the supercomputer "Fugaku" in the international "Graph500" performance ranking for large-scale graph analysis using the full specifications of the supercomputer "Fugaku". The research group, consisting of FITSTARS and FUJITSU LIMITED, succeeded in improving the performance of the supercomputer "Fugaku" in the "Graph500," an international performance ranking of supercomputers for large-scale graph analysis, and ranked No. 1 in the world for five consecutive terms following November 2021.
  The ranking was announced by the Graph500 Committee on May 30 (Japan Standard Time) in conjunction with ISC21, an international conference on high-performance computing (HPC).
  The performance of large-scale graph analysis is an important indicator in the analysis of big data, which requires large-scale and complex data processing, and we are promoting application development (smart factories, MaaS (delivery optimization), etc.) with private companies using technology that has achieved the world's highest performance.
• 第23回 Graph500 ベンチマーク 世界1位 (SC21, セントルイス, アメリカ)
• The Eighth Graph 500 list and the Third Green Graph 500 list were announced at the 2014 International Supercomputing Conference (ISC'14) in Leipzig, Germany. We won the first positions in the world ranking of both the Graph500 Benchmark and the Green Graph500 Benchmark. These results show that our algorithm and implementation is suitable for big data computation in several computing environments, such as distributed-memory supercomputers, shared-memory supercomputers, Intel/AMD servers, and Android devices.
  http://www.graphcrest.jp/eng/news2014-06.html

Educational Activities

Professional and Outreach Activities