View full abstract

  In recent years, electronic states calculations of the whole proteins can be performed now. By the appearance of the supercomputer ‘K’, this method which requires the plenty of computational resource is shifting to a practical phase. In this situation, the visualizing method is indispensable to pull out direct and useful information from very large and complex all-electron wave function calculation results of proteins. In this study, the new visualizing method for analyzing correlation between protein structure and electronic structure was proposed. The validity of this method was investigated with two typical examples, and the significant features were found among these structures. The comprehensive analysis by using the supercomputer ‘K’ etc. will progress the production of such a new information database that will be useful for understanding of the electronic structure principle in proteins and the molecular design of new proteins.

View full abstract

  Silicon nanowires have attracted much attention as a next-generation semiconductor device. We performed electronic-structure calculations of tens-of-thousands-atom silicon nanowires, whose sizes are same as realistic devices, by our RSDFT code on K computer. We found that efficiency of the nanowire device is unaffected by surface roughness of nanowire. We also performed three-dimensional visualization by using the π-CAVE system to obtain detailed distribution of wave functions. Finally, we state that visualization technique is important field to find new physical phenomenon.

View full abstract

  In this paper, the developments of visualization tools for large-scale geoscience simulations (e.g. Earth quakes, Tsunami and Typhoon simulations) using such as K-computer are presented. We first describe the development of the middleware to deal with the expensive rendering cost for the huge simulation data on a PC-cluster system. Our method combined with ray-tracing calculations provides a quality of photorealistic image to scientific visualization with a reasonable rendering time. We also implement a co-processing mode that can mimic an in-situ visualization process which may be the most effective method to reduce the time and storage space required for the visualization process. Then we discuss the new approach of the in-situ visualization method with petascale data. We propose to use bullet-time technique in the in-suite visualizations process, which generates many images/movies with arbitrary selected visualization parameters for a snapshot/sequential data. In order to deal with the visualization results brought by this method, we develop application software to play animation movies which can change the files on screen interactively in a real time. Finally we introduce the development of the application software “EXTRAWING” which visualize simulation result on Google Earth, as a powerful presentation tool of the simulation result.

View full abstract

  Computational Fluid Dynamics (CFD) has achieved rapid development in aspects of software and hardware. Then, the CFD becomes an alternate tool of a wind-tunnel experiment in a development process of road vehicle aerodynamics. However, the development process is still similar to the past. Therefore, toward an innovation of the process, our research group is developing a new development and design system of road vehicle aerodynamics based on unsteady and three-dimensional turbulent field simulation.
   Here, we show two cases of unsteady aerodynamics simulations using a next-generation aerodynamics simulator of road vehicles, which is a basis of the new development system. One application is understand of physical mechanism that causes an aerodynamic damping acting on a pitching road vehicle. The other is a coupled simulation of aerodynamics and vehicle dynamics in a gusty crosswind situation. Additionally, concept, target, and key problem of the new development and design system running on petascale computers, such as “K”, are mentioned.

View full abstract

  In the field of astronomy, particle simulations are often performed. We once have made a tool for visualizing particle data and used it to make movies for educational purpose and public outreach. With recent developments on computing, larger and larger numbers of particles are used in simulations. Thus, we have now developing a new application “Zindaiji3” to visualize particle data with large number. Our purpose is to develop a tool which is fast and can handle as large number of particles as possible with a single PC. By fully revising the algorithms, “Zindaiji3” can now make movies from date with up to 10-100 million particles. We introduce new features of “Zindaiji3” and discuss how much particles can be handled in a single PC in the aspect of memory limit and rendering speed. In an ideal case, “Zindaij3” can handle up to billion particles. PC clusters would be necessary to handler larger data.

View full abstract