RIKEN Center for Computational Science

Kcomputer 京について

Research Highlights

The K computer is being used in a broad range of fields including drug discovery, earthquake/tsunami research, weather forecasting, space science, manufacturing and material development.
Here are some examples of scientific results using the K computer.

You can find the most recent list of research achievements in "Press Release" and "Topics".

You can search some of the published achievements of researches using the HPCI system including the K computer on the HPCI Publication Database.

*HPCI(High Performance Computing Infrastructure)

2014.11.13 update

K computer used for discovery of novel functions of super concentrated electrolyte, with possible use in next generation advanced batteries 

The University of Tokyo, Kyoto University, and National Institute for Materials Science (NIMS) collaborated to develop a novel electrolyte for lithium ion batteries (LIBs), and elucidated the mechanisms of the electrolyte using the K computer.
The novel electrolyte contains lithium ions at super high concentration. This super concentrated solution has two novel functions “remarkably fast reaction kinetics” and “high reduction stability”. This finding overturns the conventional concept that highly concentrated solution is inappropriate for battery electrolyte because of its slow reaction kinetics.
The novel electrolyte is expected to be useful for developing fast-charging (a third of existing charging time) and high-voltage (5 V-class) LIBs.

Journal: Y. Yamada, et al (2014) J. Am. Chem. Soc. 136, 5039-5046. doi: 10.1021/ja412807w

Related Links

The University of Tokyo (Press release in Japanese)

NIMS (Press release in Japanese)

2014.11.13 update

Elucidating mechanisms of dissociation of methane hydrate using the K computer

Researchers elucidated the mechanisms of dissociation of methane hydrate for generating methane gas at the molecular level. Methane hydrate is a sherbet-like substance composed of water and methane. Although it has attracted attention as an energy resource, little is known about its dissociation mechanisms.
In this study, the researchers found the following mechanisms. When the methane hydrate dissolves, methane is supersaturated in water. As the concentration approaches the stability limit, bubbles form and accelerate the dissociation. It may be possible to regulate the dissociation of hydrate by control of bubble formation.
These mechanisms are expected to be useful for developing methods to recover methane effectively.

T. Yagasaki, M. Matsumoto, Y. Andoh, S. Okazaki, and H. Tanaka, (2014) J. Phys. Chem. B. 118. 1900. doi: 10.1021/jp412692d.

Figure : Dissociation of methane hydrate

Related Links

Okayama University (Press release in Japanese)

2014.11.12 update

The new EigenExa software dramatically accelerates the performance of applications running on the K computer

The K computer was used to perform one of the world’s most complex computations, an eigenvalue computation with a 1,000,000 x 1,000,000 matrix, using the new software.
The computing time was reduced from about 1 week, the generally accepted time at present, to just 1 hour.
The acceleration could be applied in bioinformatics* and social science in addition to semiconductor device design, new materials development and drug discovery. The EigenExa is freely available.

*Bioinformatics : Computational analysis of big biological data such as genomic data

Related Links

RIKEN (Press release in Japanese)

Large-scale Parallel Numerical Computing Technology Research Team (Download)

2014.08.29 update

A high-resolution atmosphere-ocean coupled model substantially improves typhoon intensity forecasts verified by a large number of experiments using the K computer

It is important to improve typhoon forecasts for disaster prevention. However, the forecast skill of typhoon intensities (minimum sea level pressure and maximum wind speed) has not been improved over the past 20 years. A high-resolution atmosphere-ocean coupled model is believed to improve typhoon intensity forecasts because it can represent a typhoon inner core structure and ocean feedback. Nevertheless, a number of studies using such a model are not sufficient partly because the calculation requires a heavy computational cost. In this study, we developed a high-resolution coupled model and applied it to 281 forecasts thanks to K computer. This study confirms that a high-resolution coupled model substantially improves the typhoon intensity forecasts.

Related Links

JAMSTEC (Research results)

2014.05.24 update

The K computer made it possible to realistically simulate cumulonimbuses in a global atmosphere model with the world’s highest resolution higher than 1 km.   

The results of simulations with different resolutions showed that once the grid spacing was 2 km or less, the simulation became much closer to the actual process of cumulonimbus development.

This finding will contribute to breakthroughs in understanding the mechanism for the occurrence of typhoons and torrential rain.

Journal:Miyamoto et al (2013) , Geophys. Res. Lett.,  40, 4922–4926, doi:10.1002/grl.50944.

Related Links

RIKEN (News piece)

RIKEN (Riken Research)

2014.05.24 update

Largest neuronal network simulation achieved using K computer

RIKEN has succeeded in simulating the largest neuronal network by exploiting the full computational power of the supercomputer K in cooperation with Julich and OIST.

The simulated network consists of 1.73 billion nerve cells connected by 10.4 trillion synapses, exceeding the previous record in neurons by 6% and in synapses by 16%.

The research is expected to contribute to the development of supercomputer and software design, promoting advances in research toward whole brain simulation.

Related Links

RIKEN (Press release)

Okinawa Institute of Science and Technology (OIST) (Press release)

Forschungszentrum Juelich GmbH (Press release

2014.05.23 update

Elucidating reduction reactions mechanisms of lithium ion battery (LIB) electrolyte using the K computer. Leading the way in computational materials design for lithium ion battery with higher performance and better safety

LIB applications to electric vehicles and smart grids have recently attracted much attention for saving energy.  However, the electrolyte-electrode interfaces, which govern the LIB performance and reliability, have not been understood yet on the atomic scale due to the experimental difficulty.

A research group carried out quantum mechanical simulations of decomposition and binding reactions in the LIB electrolyte on the K computer, and succeeded in elucidating the reaction mechanisms, on the atomic scale.

Their findings will contribute to the materials design for LIBs with larger capacity and power, better safety and longer cycle life.

Related Links

National Institute for Materials Science (NIMS)/Japan Science and Technology Agency (Press release)

Fujifilm (Press release in Japanese)

2014.05.23 update

New simulation technology for permanent magnet design toward high performance magnets without using rare earth elements

Fujitsu developed a large-scale magnetization simulation technology using the K computer and performed the world’s largest scale simulation of a permanent magnet.

The new technology made it possible to analyze microscopic magnetic domain structures, which has been difficult with a conventional simulator.

The new technology is expected to accelerate research and development of  rare earth free magnets.

Figure: Simulation of the magnetization-reversal process using a polycrystalline model

Related Links

FUJITSU (Press release)

2014.05.23 update

Forecast experiments for the heavy rainfall event in northern Kyushu in July 2012 with the K computer

The Meteorological Research Institute conducted data assimilation and ensemble forecast experiments for the heavy rainfall event as the first full-scale trial where the K computer was used for a research on prevention of natural disasters.

The results indicate torrential rainfalls can potentially be captured12-24 h before the occurrence with high probability.

It is expected that this outcome would contribute to improve the prediction of local severe rainfalls in future.

Related Links

Meteorological Research Institute

Japan Agency for Marine-Earth Science and Technology (JAMSTEC)
SPIRE (Strategic Programs for Innovative Research) - Field 3 (Results of Research)

2014.05.23 update

Ensemble experiments using a numerical model that expresses convections were conducted to forecast the tornadoes that occurred on 6 May 2012.

The tornado distribution reproduced by the downscale experiments from the analysis of the ensemble forecasts indicates that the ensemble forecasts provided the generation probability of tornadoes and reduced the miss rate of tornado forecast.
These results will contribute to the probability forecasts specifying locations and timing of tornado outbreak.

Related Links

Meteorological Research Institute

Japan Agency for Marine-Earth Science and Technology (JAMSTEC)
SPIRE (Strategic Programs for Innovative Research) - Field 3  (Results of Research)