Plenary Talk 1 October 27, 09:45 – 10:45, Symposium Hall
Hiroaki NAKAMURA, Prof.
Department of Helical Plasma Research, National Institute for Fusion Science
Department of Energy Engineering and Science, Nagoya University
Molecular simulation for soft and hard matters
Using molecular dynamics (MD) and its related simulation. i.e., dissipative particle dynamics (DPD) simulation, the motion of atoms and molecules were revealed by a lot of researchers. Picking up three simulation related to morphology of amphiphilic molecules in soft-matter physics and sputtering phenomena in the plasma-wall interaction study, we show that these MD simulations are useful tool enough to analyze the collective motion of these particle systems. Specifically, the first MD simulation result shows that short chain molecules form domain structure by the Lennard-Jones interaction and that the domain interacts with each other like rigid-body. The other MD simulation gives the information of the graphite erosion when hydrogen atoms are irradiated to graphite. In this simulation, the modified Brenner potential was used. This potential was made to treat carbon and hydrogen atoms in carbon-nano-tubes. We adopted this potential to analyze the plasma-wall interaction study. Finally, using DPD simulation, we obtained the structure formation of amphiphilic molecules. The DPD simulation is one of the coarse-grained molecular dynamics simulations and is often used for soft matter research. By this DPD simulation, the phase diagram of amphiphilic molecule was obtained.
Plenary Talk 2 October 28, 09:45 – 10:45, Symposium Hall
Satoshi KITAZAKI, Ph.D.
Director, Automotive Human Factors Research Center,
National Institute of Advanced Industrial Science and Technology (AIST)
A big challenge for safe automated driving – Potential problems of Human Factors
The automotive industry is facing a big challenge for development and commercialization of automated vehicles. The challenge may be the largest in the history of the industry. One of the major goals of the automated vehicles is dramatic reduction of traffic accidents. It is known that over 90% of accidents are caused by human errors of drivers. Therefore, it is expected that automated system will eliminated such human errors, resulting in reduction of accidents. Commercialization of driverless vehicles is still a long way to go. Car manufacturers are targeting partially automated vehicles in which the driver still has some driving task and the responsibility for safe driving. Unfortunately, partial replacement of the driving task by the system most likely causes new type of accidents induced by the system. The purpose of human factors research is to understand the human factors behind the system induced problems and to minimize the problems to achieve the goal, which is also a big challenge. In this presentation, I will introduce the on-going human factors research sponsored by SIP-adus, cross-ministerial Strategic Innovation Promotion program for innovation of Automated Driving for Universal Services in Japan.
Plenary Talk 3 October 28, 13:30 – 14:30, Symposium Hall
Hiroshi Nakashima, Prof.
Academic Center for Computing and Media Studies, Kyoto University
Regularity: A New Important Player in the Game of High-Performance Simulations in Manycore Era
The recent emergence of manycore processors with wide SIMD mechanisms
brings a new important player, namely regularity, into the game of high-performance simulations to let it join other key players, parallelism and locality. Since the SIMD mechanism requires a set of operands onto which a particular operation is performed for its efficient work, a simulation code must be mainly composed of loops to access data sets and to perform arithmetic operations on them in a regular fashion in order to exploit the computation power of the mechanism. However, this requirement does not always mean that the
simulation code must operate on very regular data structures, e.g., a simple structured Cartesian grid, but allows us to work on more flexible ones, such as unstructured grids and sets of objects, with some irregularity in general providing we can find small-scale regularity in them and represent it as a set of loops working on the majority regular part of the data set. This paper discusses the issues shown above with an example of our particle-in-cell (PIC) plasma simulation.