This talk first provides an overview of our project, “Backcasting Digital Systems by Super-Dimensional State Engineering” [1,2], which is part of the Moonshot R&D Program Goal 10 (MS10): “Realization of a dynamic society in harmony with the global environment and free from resource constraints through diverse applications of fusion energy by 2050” [3]. This program is a Japanese flagship initiative for fusion energy. Our project was launched at the end of 2024 as an interdisciplinary effort bridging the fusion energy field with other domains, including AI and data-driven science, simulation science, applied mathematics, and high-performance computing (HPC). The project aims to develop digital systems for the design of fusion devices, particularly tokamak and helical types, as well as for related materials experiments. This talk then introduces our data analysis framework, ODAT-SE (Open Data-analysis Tool for Science and Engineering, pronounced "oh-daht ess-ee")[4]. ODAT-SE enables a variety of analysis methods, including parallelized Monte Carlo methods for Bayesian inference and parallelized Bayesian optimization. These methods are designed to run efficiently on both small-scale PCs and massively parallel supercomputers. This talk gives several preliminary results by ODAT-SE.

[1] https://www.jst.go.jp/moonshot/en/program/goal10/A3_hoshi.html
[2] https://ms10ds.nifs.ac.jp/ (Tentatively, Japanese only)
[3] https://www.jst.go.jp/moonshot/en/program/goal10/index.html
[4] https://www.pasums.issp.u-tokyo.ac.jp/odat-se/en/