Publication Date: November 29, 2024
Access counts:0
Development of In Situ Visualization and Control Techniques for Large-Scale Nuclear Simulations
Fig. 1 Overview of in situ control
Simulation is utilized for design optimization and solving inverse problems through a loop. This loop involves visualization processing, observation of temporal and spatial information, and control of computational conditions.
Fig. 2 Visualization screen for in situ control
A spatial-domain view for displaying spatial information in computational data, a temporal-domain view for demonstrating temporal information, and a user interface for controlling computational conditions.
In the field of nuclear energy, extremely large-scale simulations are being conducted, and real-time computation has become possible thanks to advancements in supercomputing. To utilize this advancement for exploring design variables and inverse problem analysis, which reconstructs reality from measured values, it is essential to visualize computational data, deduce spatial and temporal behaviors, and control computational conditions using in situ control technology (Fig. 1).
Conventional methods require several hours for visualization processing as they convert the entire region of computational data into polygonal visualization elements. Additionally, it takes several days to modify computational conditions and resume calculations after interrupting computation on supercomputers.
In this study, we considerably accelerated the visualization process, reducing the duration to only few seconds using particle-based visualization. This method generates visualization particles only in regions that require visualization. Furthermore, we developed a technique that allows the interactive control of computational conditions without interrupting computation by enabling communication between the supercomputer and user’s PC via files stored on storage media. By integrating these technologies, we developed a graphical user interface that enables the visualization of spatial and temporal information and interactive changes under computational conditions (Fig. 2). This technology was applied to a pollutant dispersion simulation on the SGI8600 supercomputer at our institute, successfully performing inverse problem analysis to reproduce pollutant concentrations at observation points.
We were awarded the 35th Paper Award of the Visualization Society for this study.
This study was supported by JSPS KAKENHI Grant-in-Aid for Scientific Research (C) Grant Number JP20K11844, "Interactive In-Situ Visualization of Large-scale Distributed GPGPU Simulations."
If you have any comments or feedback about this page, please click the button to share your thoughts with us.