The mission of the Center for Computational Sciences (CCS) is the promotion of “Multidisciplinary Computational Science” through enhanced cooperation between, and the fusion of, computational and computer sciences. To that end, the CCS works toward the development of high-performance computing systems and networks, conducts sophisticated simulations in a variety of scientific research fields, and endeavors to expand the frontiers toward innovative information technology. The scientific research areas of our Center encompass particle physics, astrophysics, nuclear physics, nano-science, life science, environmental science, and information science. On the other hand, the research on computer systems covers high performance computing technology from the architecture to the algorithm and the informatics including big data analysis on large scale storage and database system, and even media science.
CCS Vision of “Multidisciplinary Computational Science”
Computational science is explored by large-scale simulations with high performance computing resources as well as the high-bandwidth networks. The pivot for computational science is the collaboration between wide fields of computational scientists and computer scientists having expertise of computer hardware, software, algorism, and application coding. This collaboration allows us to develop computers best suited for scientific exploration. We define such a “codesign” as “Multidisciplinary Computational Science”.
Based on the codesign concept, the CCS developed the massively parallel supercomputer named the Computational Physics by Parallel Array Computer System (CP-PACS) in 1996, which was ranked as the No. 1 system on the TOP500 List of November 1996. Large-scale computations carried out on CP-PACS have resulted in significant progress in particle physics and astrophysics. Since then, a large-scale cluster system named the Parallel Array Computer System for Computational Sciences (PACS-CS, 2006), and a special-purpose parallel system for astrophysics, the Cosmo-simulator FIRST (2007), have been developed. In 2011, another codesign project for the Highly Accelerated Parallel Advanced system for Computational Sciences (HA-PACS) was launched.
The CCS has been striving to carry out collaborative research with computer scientists in a field of fundamental science for more than 25 years. Currently, in addition to fundamental science (particle physics, astrophysics, and nuclear physics), applied fields are spread to materials science, life science, and global environment. As a result, a wide variety of collaborations among different fields have been initiated. This is a significant potentiality of “Multidisciplinary Computational Science”.
Collaboration and Alliance
As for nationwide alliances, the CCS has entered the T2K alliance, which was an alliance aimed at the acquisition of a T2K supercomputer systems. In 2013, we joined with the University of Tokyo in establishing the Joint Center for Advanced HPC (JCAHPC) as the first interuniversity supercomputer center in Japan. This alliance between two universities allowed us to construct an unprecedentedly large-scale computational facility, and installed a novel many-core architecture supercomputer, Oakforest-PACS, which was ranked as the fastest supercomputer in Japan on November 2016. Furthermore, a wide variety of multidisciplinary collaborations has been accelerated in Joint Institute for Computational Fundamental Science (JICFuS) and Organization for Collaborative Research on Computational Astrobiology (CAB), and new collaborative activities with computational informatics or medial science have been initiated. Also, under the alliance with RIKEN Center for Computational Science (R-CCS, former AICS), the technology for high-performance computing has been explored. Besides, since our university is located in Tsukuba Science City, which is home to many government research organizations, we already have firm connections to a number of major research institutions located there. These include the High Energy Accelerator Research Organization (KEK), the Advanced Institute for Science and Technology (AIST), the National Institute for Materials Science (NIMS), the Meteorological Research Institute (MRI), and the National Institute for Environmental Studies (NIES).
International collaborations and alliances in computational science research are important factors for promoting and accelerating multidisciplinary computational science. To that end, we have strengthened international collaborations through alliances with the University of Edinburgh in the UK and Lawrence Berkeley National Laboratory (LBNL) in the USA. With Korea Institute of Science and Technology Information (KISTI), we organize a winter school and a workshop for advanced computational science every year. Additionally, the CCS computational material scientists collaborate with Vanderbilt University and Washington University in the USA, under a bilateral program sponsored by the Japan Society for the Promotion of Science (JSPS). The International Lattice Data Grid (ILDG) is an international project aimed at the development of data grids for sharing lattice quantum chromodynamics (QCD) configurations worldwide. The construction of regional ILDG grids has been finalized in the US, UK, Germany, Australia, and Japan, where it is known as the Japan Lattice Data Grid (JLDG). The JLDG is supported by the “Cyber Science Infrastructure Project” carried out by the National Institute of Informatics as a part of grid infrastructure developments for academic research between the universities and research institutions.
Educational Activities for advanced computational science
The Computational Science Dual-Degree (double major) Program is an educational program which enables a graduate student in a doctoral program to simultaneously belong to a masters program of a different graduate school, and receive both a doctoral degree in science and a master’s degree in computer science, or vice versa, upon graduation. The program is intended to educate researchers who can advance new interdisciplinary computational science from global viewpoints. We are running the program since 2008 with curriculum and courses for advanced computational science and begin this program in 2008.
Campus-wide courses on computational sciences for graduate students are other educational activities. Currently, we have two courses:
- Computational Science Literacy
- High-performance Parallel Computing Technology for Computational Sciences (overlapped with the HPC Seminar)
The lectures are given by faculty members of CCS and is worth one unit of graduate credit.
The HPC Seminar presents knowledge, methods and techniques for programming modern high-performance computer systems, including recent microprocessors and their performance, and parallel programming. This is an inter-university activity open for researchers outside the Center, even researchers in companies not related to the university. The seminar is held during 2 or 3 days in the summer. It is also was broadcasted via the Internet.