2016.05.30 update

K computer used to demonstrate possibility of predicting typhoon genesis two weeks in advance - First step toward realization of typhoon prediction -

Researchers demonstrated that it is possible to predict tropical cyclogenesis about two weeks in advance. The finding is based on simulations of eight tropical cyclones (11-18th typhoons) in August 2004, by applying NICAM (Nonhydrostatic ICosahedral Atmospheric Model), which is designed to represent detailed features of clouds around the globe.

Currently tropical cyclogenesis is predicted 1 to 5 days in advance in actual weather forecasts, and it is inaccurate despite the short forecast period. The new simulation examined convective activity in the Philippine Sea. Out of the 8 typhoons, 6 were well predicted. Especially, the 15-18th typhoons were successfully predicted two weeks before the genesis.

We expect the post K supercomputer, which will be completed in 2020, to enable a number of simulations with more highly accurate data than now.

Nakano M., Sawada M., Nasuno T., and Satoh M. (2015) Geophys. Res. Lett. doi:10.1002/2014GL062479.

Figure : Typhoon 18th of the year 2004 ("Typhoon 200418 SONGDA" by NASAOriginal uploader was Tdk at ja.wikipedia - http://eol.jsc.nasa.gov/scripts/sseop/photo.pl?mission=ISS009&roll=E&frame=21526Transfered from ja.wikipedia. Licensed under Public Domain via Wikimedia Commons )

Related Links

JAMSTEC (Press Release)

2016.05.23 update

K computer used to elucidate the mechanisms of iron-based superconductivity

In this study, researchers performed an analysis of iron-based superconductors utilizing the K computer, and succeeded in theoretically reproducing the superconductivity of an iron-based superconductor, elucidating its mechanism.

In many conventional superconductors, superconductivity appears at very low temperature (-269~-234℃). In contrast, iron-based superconductors and copper-oxide superconductors are known to behave as superconductors at substantially higher temperatures (more than -220℃), and have been widely believed to emerge from mechanism entirely different from the conventional ones.

This research could contribute to the exploration of materials showing superconductivity at higher temperatures.

T Misawa and M Imada. (2014) Nature Commun., 5:5738.　doi:10.1038/ncomms6738.

Figure : Superconductivity will create the new future
(・Tokyo Institute of Technology
・"JR-Maglev-MLX01-2" by Yosemite - Own work. Licensed under CC BY-SA 3.0 via Wikimedia Commons
・"Modern 3T MRI" by KasugaHuang. Licensed under CC BY-SA 3.0 via Wikimedia Commons)

Related Links

The University of Tokyo (Press Release)

2016.05.23 update

Large scale bubble simulation using K computer　- Elucidation of interactions between bubbles could contribute to various industrial applications -

When you uncork a bottle of champagne, bubbles immediately form, and larger bubbles grow, while the smaller ones shrink. This phenomenon is called “Ostwald ripening”.

Researchers simulated this phenomenon with 700 million particles using the K computer, and elucidated the first process of bubble formation at the microscopic level. They found that the time evolutions of number of bubbles were consistent with theoretical predictions.

The simulations will allow us to elucidate bubble formation and growth, interactions between bubbles at the molecular level, and could contribute to various industrial applications, such as design of turbines in power plants and propellers on ships, as well as the creation of metal alloys.

Watanabe H, Suzuki M, Inaoka H, and Ito N. (2014) J. Chem. Phys. 141:234703 doi: 10.1063/1.4903811.

Figure : Simulation of bubble formation (H. Watanabe, ISSP, the University of Tokyo and H. Inaoka, RIKEN AICS)

Related Links

AIP Publishing (Journal Highlights)
RIKEN (Website News)

2016.05.23 update

Realization of first-principles molecular dynamics of large molecules using K computer

Materials consist of huge numbers of atoms. The properties of materials are determined by the interactions of atoms and electrons, which are described by quantum mechanics. First-principles calculations can elucidate phenomena at the atomic or electronic level. However, it is now limited to only a limited number of atoms (usually a few hundred atoms), because it requires large-scale and complex calculations.

Research teams from NIMS and University College London developed a novel calculation technique to perform first-principles molecular dynamics on systems with more than 30 thousands atoms using the K computer and FX10 at the University of Tokyo.

They hope to clarify the behaviors of biological molecules and nanostructured materials, which include tens of thousands to millions of atoms. This research could contribute to drug design and the development of next-generation devices.

Arita M, Bowler D. R., and Miyazaki T. (2014) J. Chem. Theory Comput. 10. 5419–5425. doi: 10.1021/ct500847y.

Figure : DNA simulation in water solution (Collaborative research with Dr. Otsuka @ RIKEN QBiC)

Related Links

NIMS (NIMS Now International)
NIMS (Press release in Japanese)

2014.11.13 update

K computer used to elucidate several amplification mechanisms of magnetic fields during binary neutron star merger and black hole evolution

Researchers conducted magnetohydrodynamic simulations of binary neutron star mergers in numerical relativity with the highest resolution ever. They clarified several amplification mechanisms of magnetic fields during binary neutron star merger.
Until now, it was thought that magnetic fields amplify in an accretion torus around the black hole formed by the merger of binary neutron stars. In this study, researchers found the magnetic fields amplified before the formation of the black hole.　The researchers plan to conduct more systematic research.

K. Kiuchi, K. Kyutoku, Y. Sekiguchi, M. Shibata, T. Wada. (2014) PHYSICAL REVIEW D. 90. 041502. doi:10.1103/PhysRevD.90.041502.

Figure : Binary neutron star merger and black hole evolution

Related Links

Kyoto University (Press release in Japanese)

2014.11.13 update

K computer runs largest ever ensemble simulation of global weather

Researchers succeeded in running 10,240 parallel simulations of global weather, the largest number ever performed, using data assimilation to reduce the range of uncertainties. This required 1 million times the computations used in the conventional 100-member simulation. For the simulation, the "Local Ensemble Transform Kalman Filter" (LETKF), an already efficient system, was further improved by a factor of eight using the "EigenExa" high-performance eigenvalue solver software, making possible a three-week computation of data from the 10,240 ensembles for simulated global weather.
The team discovered that faraway observations, even going beyond 10,000 kilometers in distance, may have an immediate impact on the estimation. This research could potentially lead to an improvement of weather forecasts.

T. Miyoshi, K. Kondo, and T. Imamura. (2014) Geophysical Research Letters. 41. 5264–5271. doi:10.1002/2014GL060863.

Figure : Humidity correlation map on day 18 of ensemble data assimilation

Related Links

RIKEN (Press release)

2014.11.13 update

K computer grabs top spot in Graph500

The K computer grabbed top place in Graph500, a new benchmark that seeks to gauge the ability of supercomputers on data-intensive loads. The goal of Graph500 is to improve computing related to complex data problems in five key areas: cybersecurity, medical informatics, data enrichment, social networks, and symbolic networks.
In Graph500, the speed of a breadth-first graph search, measured by number of traversed edges per second (TEPS) is used, with "edges" indicating the connection between two data points called “nodes”. Using 65,536 compute nodes of the K computer, the researchers solved a search of an extremely large graph (1 trillion nodes and 16 trillion edges in 0.98 second). The K computer grabbed the top spot with a score of 17,977 giga TEPS.
This achievement demonstrates the flexibility of the K computer in application to a wide range of applications, including Big Data analysis.

This news was announced on June 23 (June 24 Japan time) at ISC14 held in Germany. Complete results of Graph 500 at Jun 2014: GRAPH 500

Related Links

RIKEN (Press release)

Tokyo Institute of Technology (Press release in Japanese)

2014.11.13 update

Month-long forecasts of Madden-Julian Oscillations (MJO) in the tropics realized on K computer

Researchers performed numerical experiments on Madden-Julian Oscillations (MJO), a huge atmospheric pulse that propagates eastward, affecting the entire globe, on the K computer, using a highly complex atmospheric model NICAM (Nonhydrostatic Icosahedral Atmospheric Model), which is designed to represent detailed features of clouds on the Earth. They showed that NICAM is effective for month-long forecasts of MJO.
This research could help us forecast atmospheric activity on a global-scale at an earlier stage, and improve accuracy in predicting seasonal forecasts and typhoons around Japan.
The datasets of these simulations are expected to contribute to elucidating the mechanisms behind MJO by complementing unobserved data.

Miyakawa, T. et al. (2014)　 Nature Commun., 5, 3769. doi:10.1038/ncomms4769

Figure : simulated precipitation anomalies, for different MJO phases

Related Links

JAMSTEC (Press release)

The University of Tokyo (Press release)

The University of Tokyo (Press release in Japanese)

2014.11.13 update

K computer simulation provides evidence that neutrino-heating is involved in supernovae

Using the K computer, researchers performed large-scale simulations demonstrating the possibility that supernovae are triggered by a neutrino-heating mechanism. For 50 years astronomers have puzzled over how supernovae are triggered. In the past, simulations were done on the assumption of a spherical star, which is unlike real supernovae. Thus, it was impossible to conclude whether the neutrino-heating mechanism is reasonable.
In this study, the researchers performed the calculation under a natural assumption close to the real situation. They found strong evidence to support the neutrino-heating hypothesis. The researchers will perform the simulation with even greater accuracy and with larger scale than now, to explore the details of supernovae.

Tomoya Takiwaki et al. (2014) The Astrophysical Journal. 786. 83. doi:10.1088/0004-637X/786/2/83.

Figure : Image of a supernova

Related Links

CfCA (Web release in Japanese)

Fukuoka University (Press release in Japanese)

Kyoto University (Press release in Japanese)

2014.11.13 update

Using the K computer to calculate the solar convection zone with the highest resolution ever

Using the K computer, researchers calculated solar thermal convection with the highest resolution ever. Nuclear fusion produces energy in the solar core. In the proximal layer (below about 0.7 solar radius), which is called the radiation zone, energy is transferred by radiation. The layer beyond 0.7 solar radius from the center is called the convection zone. There, energy is transferred outward by thermal convection.
Turbulent flows are observed to be dominant in the solar convection zone. To understand energy flows and magnetic field generation in the sun, it is important to model the turbulent flows using simulations. Doing so will contribute to elucidating the mechanism of sunspot formation and predicting fluctuations in solar activity.

Journal： H. Hotta, et al. (2014) The Astrophysical Journal. 786. 24. doi:10.1088/0004-637X/786/1/24.

Related Links

The University of Tokyo (Press release in Japanese)

Space & Planetary Science Group, The University of Tokyo (Movie courtesy of Dr. Hideyuki Hotta)