Loosely Coupled Approach on Nano-Science Simulations

Mutsumi AOYAGI

Professor, Computing and Communication Center, Kyushu University



<Abstract>

It is becoming increasingly important to analyze multi-scale and interdisciplinary problems in Nano- and Bio-computing, in which complex physical phenomena in different spatial and temporal scales are coupled. For instance, in drug design to analyze electronic structure of solute atoms in solvent, it is indispensable to exploit a coupled simulation in which electronic structure calculation of nano-scale molecule based on molecular orbital method is combined with solvent distribution calculation using statistical dynamics for macro region. The coupled simulation is expected to solve large-scale problems with high physical accuracy by combining simulation components associated with different kinds of physical model and numerical method.

We have been developing application specific middleware tools, named as mediator, to provide advanced technique of exchanging data for coupled simulations on distributed computational environments. Feature functionalities of the mediator consist of process management for exchanging data through the mediator, semantic transformation between different physical quantities, and synchronous file transfer on grids referred as storage-based communication (SBC). In order to minimize impact on original and legacy simulation codes, API functions of the mediator associated with the above functionalities are provided that allow to construct simulation components and realize complex data exchange between them.

To demonstrate the general applicability of the framework of loosely coupled simulations using the mediators, RISM-FMO coupled simulation is developed to calculate entire electronic structure of hydrate protein in infinite solvent for the first time. We also developed an accelerator tool for coupled simulations to greatly reduce the iteration times of the self-consistent calculation. The applicability of these method and tools for the peta-scale computing in the next generation will be discussed.

Whole electronic distribution of

Lysozyme surrounded with charge

distribution of oxygen atom of water.

Algorithm of RISM-FMO coupled simulation

<Biographical Notes>

Present

Professor, Computing and Communication Center, Kyushu University

Education

1979 - 1983 Undergraduate Course in the Faculty of Science,

Keio University, Obtained B.Sc.

1983 - 1988 Graduate Student in the Department of Science,

Nagoya University, Obtained Ph.D. (Thesis advisor; Dr. Kato)

Academic Career

1988 - 1990 Post-doctoral fellow at Argonne National Laboratory.

1990 - 1993 Researcher at National Institute of Advanced Industrial Science and Technology (AIST).

1993 - 2002 Associate Professor at the Institute for Molecular Science,

2002 - Professor at Kyushu University.

I have been working in the research field of computational molecular science such as parallel computations in quantum chemistry and molecular dynamics studies. Recently I have been involved in the national research grid project as a head of grid-enabling team of nano-science applications. Present research interest is to investigate the method of coupled simulation on peta-scale computing environment in the national leadership project.