All matter is composed of particles called atoms, and due to existence of inter-atomic interaction, atoms form molecules. In another word, our world is, in varying degrees, made of atoms and molecules, and such many-body problems display structural and unique phenomena in nature.
My research is in the area of carbon nanotubes, fullerenes, DNA, and my interest centers on modeling and simulation techniques in order to study the characteristics such as heat transfer and mechanical properties, electronic structure, structural stability and the like.
Hopefully, the research will result in discovery of novel structural properties using simulation techniques, and revolutionizing diverse applications, pioneering new materials, medicines and so on.

Physical and structural properties in a substance composed of tens of thousands of atoms strongly depend on its composition elements, its number of atoms and atomic structure. Nature is not only a simple extrapolation of the properties of a few particles. At each stage in scale, new laws, concepts and generalizations are necessary. At each level of complexity, entirely new properties appear, which are hardly predictable without computational simulation results. This is the most fascinating part of my research, and it makes this R&D more challenging.

Tight-binding method (TB) and density functional theory (DFT) are the most conventional approaches available in computational physics and computational chemistry that incorporate quantum theory, in order to investigate the properties of many-body systems, atoms and molecules. TB is an approximate method involving parameterization, in which parameters extracted by experimental results are used. The approximation is very practical and indispensable for treating multi-particle dynamics because of its being computationally inexpensive in comparison with DFT. Whereas, DFT can be an alternative for treating certain elements to which experimentally obtained parameters can not apply. DFT is also of importance for close investigation on electronic structures of multi-elements.

Accepting the argument that evolving computational techniques such as simulation, code optimizing and so on, are needed for further advancement in computational sciences & technologies, my research is passing from a rudimentary stage to a mature one, while computational technologies are undergoing a drastic change every moment.

As a part of collaboration with Dr. Tomanek (Michigan State University), we have taken an active role of developing and implementing CRTMD (Carbon-Recursion-Technique- Molecular-Dynamics) code based on a tight-binding theory with a recursion technique to solve an eigenvalue problem of Hamiltonian. Our further research has also been involved in the vector optimization and parallelization for massively-parallel systems such as the Earth Simulator (Fig.1 ). In consequence, the high performance was achieved by the code optimization to the point where it efficiently performed five hundred thousands of MD steps which simulated the physical behaviors of CNT consisting of forty thousands of atoms (Fig. 2 ). In the project, we observed mechanical properties and obtained a thermal conductivity of CNT. The research results in presenting potential use of the properties of CNT rope.

Recently, we moved to the optimization of PWscf (Plane-Wave Self-Consistence-Field, http://www.pwscf.org) which is a code for electronic structure calculations based on a density functional theory. The code optimization has been successfully provided on the Earth Simulator, and is an ongoing process on Hitachi SR11000. In such optimization processes, FFT, eigenvalue, eigenfunction equation, orthogonalization of eigenfunction are listed as key controlling factors. In a word, mathematical-oriented programming is of a great importance in high-speed parallel algorithms. Fig. 5 illustrates the simulated hydrogen adsorption state by applying transition metal on carbon nanotubes. The DOS analysis clarifies the absorption mechanism as shown in Fig. 6 . The computation time required by the simulation is comparatively small, with several hundred atoms and hundred MD steps being involved. Besides the above introduced specific research themes, we are working on the understanding of the properties and structures of DNA, CNT and so on.

Optimization for the code may be a rigorous approach that requires detailed consideration in every factor, and all the more, it makes this field of research attractively challenging.