Simulation-based engineering and science

The Yang group maintains their own 50 cpu cluster. As VP for research of Fudan University, Prof. Yang authorized the purchase of a 259-processor cluster for use across campus. The computer was acquired with funds from the Ministry of Education and is supported by university funds, and thus far researchers on campus are not charged for usage in order to encourage new users. Once the machine becomes heavily used (it was nearly 100% utilized as of December 2007), the university will charge researchers a subsidized fee that will help to pay for a new, larger computer. Prof. Yang noted that this free resource is raising the level of computation and SBE&S on campus. There is no data management plan in place at this time.

[1] http://english.people.com.cn/90001/6381319.html

[2] Li Lixu (2004). "China’s Higher Education Reform 1998-2003: A Summary". Asia Pacific Education Review V (1).

[3] http://en.wikipedia.org/wiki/Project_211

[4] Li Lixu (2004). "China’s Higher Education Reform 1998-2003: A Summary". Asia Pacific Education Review V (1).. Comrade Zemin quotation is on p.17.

[5] Pallavi Aiyar (2006). "China hunts abroad for academic talent". Asia Times Online Feb (18).

[6] "International Rankings and Chinese Higher Education Reform" (2006). World Education News and Review XIX (5).

[7] http://en.wikipedia.org/wiki/Project_985

http://www.edu.cn/20010101/21852.shtml

Fudan University web site (English): www.fudan.edu.cn/englishnew/about/intro.html

Fudan University Department of Macromolecular Science website: http://www.polymer.fudan.edu.cn/

Prof. Hongxia Guo

Prof. Qi Liao

Prof. Xiaozhen Yang

Prof. Charles C. Han (in absentia)

Prof. Wan Li-jun

The Joint Laboratory of Polymer Science and Materials was founded in 2003 by Dr. Charles C. Han (email: c.c.han@iccas.ac.cn) who serves to the present as its director, by unifying two laboratories for polymer research housed in the Institute of Chemistry, Chinese Academy of Sciences (ICCAS). ICCAS is a multidisciplinary institute dedicated to basic research in the chemical sciences, and to the development of innovative high-technology application and technology transfer to meet strategic national needs. Its current director is Dr. Wan Li-jun, who gave an overview of the Institute. Major research directions at ICCAS include macromolecular science, physical chemistry, organic chemistry, and analytical chemistry. Major research areas include polymer science and materials; chemical reaction dynamics and structural chemistry; organic solids; photochemistry and photo-functional materials; nanoscience and nanotechnology; colloid, interface sciences and chemical thermodynamics; molecular recognition and selective synthesis; analytical chemistry for life sciences; theoretical chemistry; and high-tech materials. ICCAS has 442 staff, 82 professors, 8 academicians, 117 associate professors, and 171 assistant professors. There is no theory division per se; instead theorists are embedded within various labs, which increases interaction. Currently 6% of the faculty and staff are theorists/simulators performing SBE&S related work.

The State Key Laboratory of Polymer Physics and Chemistry (SKLPPC), founded in 1956, and the CAS Key Laboratory of Engineering Plastics Laboratory (EPL), founded in 1991 and achieving Key Lab status in 2004, are two Key Laboratories that conduct research on polymers within ICCAS. The Joint Laboratory of Polymer Science and Materials (JLPSM) was created in 2003 to oversee both labs. Dr. Han serves as director and chief scientist of both SKLPPC and JLPSM. The vision of the Joint Laboratory is to unify the experimental, theoretical and computational spectrum of polymer science and technology activities ranging from atomistic scale, to mesoscale materials characterization to macroscale applied polymer rheology and processing. SBES plays an especially foundational role in the Joint Laboratory in that the Lab’s experimental program is prioritized to provide insights at the scale-to-scale transitions along the multiscale spectrum, a vision provided by Han. Between 1998-2004, the combined labs published 1038 papers, which includes 643 papers in SCI journals, and 66 papers in journals with impact factors over 3.0. The lab applied for 231 patents during this period, of which 121 were granted. The labs also received four China science prizes, and presented 73 invited lectures at international conferences, during this period.

Professor Charles Han is one of the world’s leading polymer scientists. He graduated in Chemical Engineering from the National Taiwan University in 1966. He received his MS in Physical Chemistry from the University of Houston (1969) and PhD in Physical Chemistry from the University of Wisconsin, Madison (1973). The following year he joined the National Institute of Standards & Technology and remained there for the next 29 years as Research Scientist, including Group Leader for the Polymer Blends Group (1985–1995), Group Leader for the Multiphase Materials Group (1999–2002) and NIST Fellow (1995–2002). During his NIST period, Charles Han was Visiting Professor at Kyoto University (Japan) and Changchun Institute of Applied Chemistry at the Chinese Academy of Sciences and Adjunct Professor at Pecking University (China) and the University of Connecticut (a post he still holds). In 2002 Charles Han became a Director and Chief Scientist at the Joint Lab of Polymer Science and Materials of the Institute of Chemistry at the Chinese Academy of Sciences in Beijing. His research interests are in the areas of structure and properties of polymers and polymer blends, scattering in polymers, statics and kinetics in polymer phase separation, and multi-phase phenomena. He is the recipient of several awards, including the Bronze (1980), Silver (1982) and Gold (1986) Medal of the US Department of Commerce, the Dillon Medal (1984) and the Polymer Physics Prize (1999) of the American Physical Society, and Humboldt Senior Research Award from the Alexander von Humboldt Foundation in Germany (1995), and he is a Fellow of the American Physical Society. He is co-author of more than 300 publications in archival journals and books.

Dr. Han was on travel, so the spectrum of activities in the laboratory were covered by Drs. Yan and Guo.

• 
  SBE&S research: Examples of SBE&S related research highlighted in the most recent brochure includes molecular dynamics simulations of polyelectrolytes in a poor solvent (Q. Liao, et al., Macromolecules, 2003, 36, 3386-3398 and free energy calculations on effects of confinement on the order-disorder transition in diblock coplymer melts. The panel heard short overviews on research examples in SBE&S:
  

• 
  Zhigang Shuia – organic electronic and photonic materials via ab initio computations. OLED/PLEDs, transport, organic photonics. Uses VASP.
  

-- O/PLEDs: Focus on aggregation-induced emission and prediction of efficiency. Related work published in JCP 126 114302 (2007) and Adv. Chem. Phys. 121 (2002).

-- DMRG approaches: EOM-CC methods for non-linear optics. Also “correction-vector” approach (cf. JPC A 111, 9291 (2007)). This code is used by a number of other groups, such as Bredas’ group.

• 
  Hongxian Guo – soft matter at multiple length scales, including molecular dynamics and coarse-grained models.
  
• 
  Qi Lao: Molecular dynamics-Lattice Boltzmann hybrid algorithm. Has three students.
  
• 
  Xiaozhen Yang: Microscopic and mesoscale simulations of polymers. Develops own codes. Has six students, one asst. professor, and uses part of a 64-node cluster.
  
• 
  Dadong Yan: Theory of polymer crystallization and self-assembly.
  
• 
  Charles Han: Overseeing major project on integrated theory/simulation/experiment for multiscale studies of condensed phase polymer materials. Funded with 8M RMB for four years, 20 investigators, with roughly 100K RMB per investigator. This is a highly unique and challenging program, inspired by the Doi project in Japan, but goes substantially beyond that project to closely integrate experiment both as a validation tool and to supply needed input and data for models.
  

• 
  Education of students in SBES: At CAS, there are fewer chances to take courses than at the university in general, and this is also true in simulation. One undergraduate course on introduction to simulation is typical. The graduate school offers one course on molecular simulation where students can learn molecular dynamics. Although many of the researchers’ backgrounds are in physics, this is not true of the students. The student training issues in SBE&S are very similar to elsewhere, with insufficient physics and math preparation.
  
• 
  Funding: Today, 60-70% of student funding comes from their “NSF”, with almost none coming from the academy. This is a different model than in the past. Increased funding is leading to 20-30% growth per year in students and faculty. 30% of funds are used for equipment.
  
• 
  Interactions with industry: Chinese industry historically has not done much R&D, and thus innovation has been very difficult. This is starting to change. The State encourages interactions between universities and industry, and the 863 Project supports research with industry specifically. However, simulation has not been popular in 863 projects. One reason is that the industrial timeline is very short term, with results desired within six months. Thus nearly all research in China is funded by the government. The State Key Lab of Polymer Science has funding from the Beijing branch of P&G for a collaboration on detergents for three years.
  
• 
  HPC resources: Shuai has a 300 cpu cluster and a four-year old 64-cpu cluster. ICCAS has a 128-cpu cluster, and the CAS has a 1024-cpu cluster.
  

ICCAS. N.d. Annual report of the Joint Laboratory. Beijing: ICCAS.

Site: Institute for Molecular Science (IMS)

38 Nishigo-Naka

Myodaiji, Okazaki 444-8585 Japan

http://www.ims.ac.jp

Date Visited: December 5, 2007
WTEC Attendees: P. Cummings (report author), G. Karniadakis, L. Petzold, T. Arsenlis, C. Cooper, D. Nelson
Hosts: Prof. Fumio Hirata, Director, Dept. Theoretical and Computational Molecular Science
Group Leader, Developing the Statistical Mechanics Theory in Chemistry and Biophysics

Email: hirata@ims.ac.jp

Prof. Susumu Okazaki, Director, Research Center for Computational Science
Group Leader, Molecular Dynamics Study of Classical Complex Systems and Quantum Systems in Condensed Phase

Email: okazaki@ims.ac.jp

Dr. Kenji Yonemitsu, Associate Professor, Department of Theoretical and Computational Molecular Science
Group Leader, Theory of Photoinduced Phase Transitions

Email: kxy@ims.ac.jp

Dr. Katsuyuki Nobusada, Associate Professor, Department of Theoretical and Computational Molecular Science
Group Leader, Theoretical Studies of Electron Dynamics

Email: nobusada@ims.ac.jp

Prof. Shigeru Nagase, Dept. of Theoretical and Computational Molecular Science Group Leader, Theoretical Study and Design of Functional Molecules: New Bonding, Structures, and Reactions

BACKGROUND

Japan’s Institute for Molecular Science (IMS) was founded in 1975. In 2004, it began a process of privatization as part of the newly established National Institutes of Natural Sciences (NINS), an interuniversity research institute corporation comprised of the National Astronomical Observatory of Japan, the National Institute for Fusion Science, the National Institute for Basic Biology (NIBB), the National Institute for Physiological Sciences (NIPS) and the IMS. The NIBB, NIPS, and IMS are all located in Okazaki, in Aichi Prefecture, and from 1981 to 2004 constituted the Okazaki National Research Institutes. The aim of IMS is to investigate fundamental properties of molecules and molecular assemblies through both experimental and theoretical methods. It has a staff of ~75 research personnel and ~35 technical personnel. IMS is also a part of the Graduate University for Advanced Studies (SOKENDAI); therefore, research groups consist of faculty, graduate students, and post-doctoral researchers. The IMS is divided into three sections: Research Departments (of which there are four), Research Facilities (of which there are seven) and a Technical Division. Research groups at the IMS belong to Research Departments or to Research Facilities. All of the IMS research groups, including research groups at the Research Center for Computational Science, support visiting researchers from national, public, and private universities in Japan.

One of the four IMS Research Departments is the Department of Theoretical and Computational Molecular Science (DTCMS), which is closely affiliated with the Research Center for Computational Science (RCCS). The RCCS appears in the IMS organizational chart as being part of a Research Department (along with the DTCMS) as well as being a Research Facility. The stated goal of the DTCMS and RCCS is “to develop theoretical and computational methodologies that include quantum mechanics, statistical mechanics, and molecular simulations in order to understand the structures and functions of molecules in gases and condensed phases, as well as in bio and nano systems” (IMS 2007, 3). Seven research groups, all conducting research in theoretical and computational molecular science are contained within the DTCMS and RCCS. The WTEC delegation met with five of the seven group leaders, who also included the directors of the DTCMS (Hirata) and RCCS (Okazaki).

RESEARCH ACTIVITIES

The research groups conducting theoretical and computational molecular science research within the DTCMS and RCCS are all well regarded in their respective areas. Information about their research is available at http://www.ims.ac.jp/eng/researchers/index.html#keisan. Some of the capabilities within the IMS are unique. For example, the three-dimensional reference interaction site model (3D-RISM) theory for molecular structure is unique to the IMS and represents expertise built up by Fumio Hirata over almost three decades. The 3D-RISM theory can be used to study proteins in solution and is particularly useful for predicting the distribution of water around a protein in solution.

Most of the research codes used with the IMS are in-house codes developed within the institution and are not open-source, although they are shared with collaborators. The RCCS is a substantial computational resource for IMS and external researchers. The website (http://ccinfo.ims.ac.jp) is in Japanese only; the most recent data for which usage is publicly available is February 2007, at which time the RCCS was used by 589 scientists in 144 project groups. The RCCS computer systems (March 2008), consisting of Fujitsu PrimeQuest, SGI Altix4700, and Hitachi SR16000, are used to support computational research in quantum chemistry, molecular simulation, chemical reaction dynamics, and solid state physics. These systems are linked to international networks through Super Science Information Network (super SINET).

Most of the WTEC delegation’s discussion at the IMS did not focus on scientific research but on the IMS role in the Next Generation Supercomputing Project (NGSP), discussed in detail in the RIKEN site report. The NGSP will produce a combined vector-scalar supercomputer that is designed to achieve 10 petaflops on the LINPACK benchmark by 2011/2012 at a hardware cost of ¥115 billion. Two major software projects are being funded to develop software to run efficiently (i.e., scale effectively) on the NGSP machine. One project, being pursued by RIKEN with $15 million/year in funding, will generate software in the life sciences. The other project, funded at $5 million/year at the IMS, will generate software for the nanoscale sciences. The IMS played a similar role in the National Research Grid Initiative (NAREGI), but it is now focusing on the NGSP project. The issue of software development to scale effectively on petaflop architectures, which will have unprecedented numbers of processor cores, is of great interest in the United States as well as in Japan.

Both the RIKEN and IMS teams developing software for the NGSP face the challenge that their 5-year software development funding began in 2006, while the actual NGSP machine will not be available until 2011, and many of the architectural details of the machine are still being debated. At this time, it appears that there is no simulator for the NGSP machine that would permit software strategies for parallelization to be evaluated before the machine is built. The IMS researchers also stated that they believed that computational scientists would play a limited role in the efficient parallelization on the NGSP hardware. They will collaborate with computational scientists from the University of Tokyo and Tsukuba University in optimizing the codes. This view of the limited role of computational specialists in modifying codes (and, indeed, rewriting them from scratch or completely changing the algorithmic approach) in order to be efficient on large parallel architectures is in contrast to the prevailing view in the U.S. high-performance computing (HPC) community and the U.S. funding model for many large computational projects. For petaflop-and-beyond architectures, this problem is being made even more acute with the proliferation of multicore processors.

The software development activities at IMS and RIKEN are being driven by several computational grand challenges, such as modeling/designing post-silicon electronics, understanding enzymatic catalysis, full-physics simulation of fuel cells, and ab initio protein structure prediction, as well as pragmatism about which codes are best ready to migrate to the NGSP-level architecture.

The IMS is a scientifically first-class institution where researchers have the opportunity to focus on research activities (e.g., no undergraduate teaching duties, as in U.S. national laboratories) while still enjoying an academic environment (e.g., having a graduate program with graduate students, and having tenure with university-style titles and hierarchy). The IMS has an unusual policy of not promoting within the institution—for example, one cannot be promoted from associate professor to full professor within IMS. Thus, an IMS associate professor with aspirations to be a full professor at IMS will have to spend some time at another institution before he/she could be hired as an IMS full professor. This is a policy that clearly is designed to create high standards and to avoid in-breeding, resulting in vitality and breadth in the research programs. However, the IMS is not without challenges. Having a graduate program with no undergraduate program creates the problem that the recruitment of graduate students is more difficult than for regular universities in Japan, who primarily recruit graduate students from the cream of the undergraduate pool of students at their own institutions.

Reforms in the government funding and structure of Japanese science have resulted in the IMS, and the other institutions of the NINS, undergoing a process of privatization since 2004. This was not discussed among the IMS researchers and the WTEC delegation; however, it is clear that these reforms are having significant impact. The IMS Director-General, Hiroki Nakamura, in his introductory remarks for the 2007 IMS brochure accessible from the IMS home page (http://www.ims.ac.jp), is unusually frank in saying that, for the future health of Japanese science, “…it would be necessary to reconsider the structure by analyzing the problems we have faced. The present situation of Japanese science policy is unfortunately quite serious. We, scientists, should strive to improve this situation, and at the same time we have to take this adversity as a spring to carry on basic researches of high quality.”

The IMS clearly has the opportunity to play a major role in the NGSP, with the funding and the mandate to develop the scalable software base to address nanoscience grand challenges. The schedule of the NGSP deployment and the timetable for funding software development will create a challenge for the IMS researchers.

IMS. 2007. IMS (Institute for Molecular Science) 2007 (Annual Report). Available online at http://www.ims.ac.jp/eng/publications/ims_2007/index.html.

Prof. Aihui Zhou, Vice-Director

Prof. Zhong-ci Shi, Academician

Prof. Qun Lin, Academician

Prof. Jun-zhi Cui, Academician

Prof. Linbo Zhang

Prof. Li Yuan

The Institute of Computational Mathematics and Scientific/Engineering Computing (ICMSEC) is one of the four Institutes that comprise the Academy of Mathematics and System Science (AMSS), of the Chinese Academy of Sciences (CAS). Established in 1995, the institute focuses on fundamental research on efficient and robust numerical methods to address problems that arise from scientific and engineering applications. The results of this research are incorporated in high-performance computer software and programs. The institute presently has 17 professors, 4 associate professors, and 7 assistant professors, about 80 graduate students, and 3 postdoctoral fellows. The professors include 2 academicians of the Chinese Academy of Sciences,⁵ and 1 of the Chinese Academy of Engineering.

An overview of research areas related to SBES at the institute was presented by the Director, Prof. Zhiming Chen, although these areas are only part of the ICMSEC research portfolio.

• 
  The first such area is multiscale simulations, which are conducted with the aim of aiding hydraulic engineering design (dam construction), and also of simulating the mechanical performance of composite materials and structures made from those materials.
  
• 
  A second area is concerned with developing improved environments for high-performance computing, as exemplified by a project on a 3D adaptive parallel finite element software platform, called parallel hierarchical grid (PHG). This project is targeted at the use of roughly 1000 processors and is built in C using MPI on top of open source packages.
  
• 
  A third area is the development of a new real-space approach to solving the Kohn-Sham density functional theory equations for molecular and materials problems using adaptive finite elements. This relatively recent development is now applicable to energy calculations on molecules of fullerene size.
  
• 
  A fourth area also uses adaptive finite elements to solve problems in computational electromagnetism, as illustrated by transformer design for power stations. There is linkage with industry for this project. Another example from this area relates to the microelectronics industry, where the objective is parasitic extraction.
  
• 
  The fifth area is computational fluid dynamics, where a present example involves solving the compressible Navier-Stokes equations with chemical reactions occurring.