Kenji Higuchi

Vectorization of the KENO IV code

The multigroup criticality safety code KENO-IV has been vectorized and tested on the FACOM VP-100 vector processor. At first, the vectorized KENO-IV on a scalar processor was slower than the original one by a factor of 1.4 because of the overhead introduced by vectorization. Making modifications of algorithms and techniques for vectorization, the vectorized version has become faster than the original one by a factor of 1.4 on the vector processor. For further speedup of the code, some improvements on compiler and hardware, especially on addition of Monte Carlo pipelines to the vector processor, are discussed.

Ab Initio Study of Hydrogen Hydrate Clathrates for Hydrogen Storage within the ITBL Environment

Recently, for the first time a hydrate clathrate was discovered with hydrogen. Aside from the great technological promise that is inherent in storing hydrogen at high density at modest pressures, there is great scientific interest as this would constitute the first hydrate clathrate with multiple guest molecules per cage. The multiple cage occupancy is controversial, and reproducibility of the experiments has been questioned. Therefore, in this study we try to illucidate the remarkable stability of the hydrogen hydrate clathrate, and determine the thermodynamically most favored cage occupancy using highly accurate ab initio computer simulations in a parameter survey. To carry out these extraordinary demanding computations a distributed ab initio code has been developed using the SuperSINET with the Information Technology Based Laboratory (ITBL) software as the top-layer.

Grid computing supporting system on ITBL project

Prototype of the middleware for Grid project promoted by national institutes in Japan has been developed. Key technologies that are indispensable for construction of virtual organization were already implemented onto the prototype of middleware and examined in practical computer/network system from a view point of availability. In addition several kinds of scientific applications are being executed on the prototype system. It seems that successful result in the implementation of those technologies such as security infrastructure, component programming and collaborative visualization in practical computer/network systems means significant progress in Science Grid in Japan.

A Visual Resource Integration Environment for Distributed Applications on the ITBL System

TME, the Task Mapping Editor, has been developed for handling distributed resources and supporting the design of distributed applications on the ITBL. On the TME window, a user can design a workflow diagram, just like a drawing tool, of the distributed applications. All resources are represented as icons on the TME naming space and the data-dependency is defined by a directed arrow linking the icons. Furthermore, it is equipped with an important mechanism allowing integration and sharing of user-defined applets among the users who belong to a specific community. TME provides users a higher-level view of schematizing the structure of applications on the Grid-like environment as well as on the ITBL system.

Difficulties in vector-parallel processing of Monte Carlo codes

Experiences with vectorization of production-level Monte Carlo codes such as KENO-IV, MCNP, VIM, and MORSE have shown that it is difficult to attain high speedup ratios on vector processors because of indirect addressing, nests of conditional branches, short vector length, cache misses, and operations for realization of robustness and generality. A previous work has already shown that the first, second, and third difficulties can be resolved by using special computer hardware for vector processing of Monte Carlo codes. Here, the fourth and fifth difficulties are discussed in detail using the results for a vectorized version of the MORSE code. As for the fourth difficulty, it is shown that the cache miss-hit ratio affects execution times of the vectorized Monte Carlo codes and the ratio strongly depends on the number of the particles simultaneously tracked. As for the fifth difficulty, it is shown that remarkable speedup ratios are obtained by removing operations that are not essential to the specific problem being solved. These experiences have shown that if a production-level Monte Carlo code system had a capability to selectively construct source coding that complements the input data, then the resulting code could achieve much higher performance.

Application of Hexagonal Element Scheme in Finite Element Method to Three-Dimensional Diffusion Problem of Fast Reactors

The finite element method is applied in Galerkin-type approximation to three-dimensional neutron diffusion equations of fast reactors. A hexagonal element scheme is adopted for treating the hexagonal lattice which is typical for fast reactors. The validity of the scheme is verified by applying the scheme as well as alternative schemes to the neutron diffusion calculation of a gas-cooled fast reactor of actual scale. The computed results are compared with corresponding values obtained using the currently applied triangular-element and also with conventional finite difference schemes. The hexagonal finite element scheme is found to yield a reasonable solution to the problem taken up here, with some merit in terms of saving in computing time, but the resulting multiplication factor differs by 1% and the flux by 9% compared with the triangular mesh finite difference scheme. The finite element method, even in triangular element scheme, would appear to incur error in inadmissible amount and which could not be e...