| 초록 |
EDISON, which stands for EDucation-research Integration through Simulation On the Net, is the well-known hub platform for multidisciplinary areas in computational science and engineering by supporting various high-performance computing (HPC) simulation programs via web-based services. The number of disciplines and users of EDISON have increased annually since its launch in 2011 and currently, there are eight academic disciplines and more than ten thousand users of the platform each year. EDISON scientific workbench is a user-friendly browser-based simulation running panel that contains the customized work areas of input, output, visualization, monitoring, and job control. This tool enables users to conduct their simulations efficiently and lowers the barriers to users with fewer computational technology skill. The EDISON portal also provides a workflow execution environment that can incorporate scientific apps developed by various simulation developers from different disciplines. Recently, computing-data integration techniques are developed to EDISON to promote self-sustaining research ecosystems. Through such efforts, current EDISON contains several specialized sites, such as EDISON-Materials, EDISON-MQCP (Modular Quantum Chemistry Package), EDISON-AI, et cetera. With the recent great interest and trends toward AI, now EDISON platform can be used for web-based HPC simulations as well as data analysis, AI modeling, and sharing. For example, EDISON-AI provides a GUI interface to develop machine learning models to build a programming code block automatically which novice users want to learn about. Users can run the scripts on the web and modify/add the generated code to improve their model whenever their dataset has been changed by their scientific insights. Users are also allowed to share datasets, collections, Jupyter notebooks among authorized community members. Research results that exploit machine learning technologies via EDISON simulation data will be explained in detail in the presentation. |
