Initialising ...
Initialising ...
Initialising ...
Initialising ...
Initialising ...
Initialising ...
Initialising ...
Ebihara, Kenichi; Yamaguchi, Masatake; Itakura, Mitsuhiro
no journal, ,
The results obtained so far in the hydrogen embrittlement research at the Center for Computational Science and e-Systems are presented in two topics: "Interaction between hydrogen and defects by first-principles calculations" and "Estimation of hydrogen distribution state by temperature rise desorption simulation". In addition, our past activities and ongoing first-principles calculations on austenitic steels are also presented.
Kobayashi, Keita
no journal, ,
Machine learning molecular dynamics (MLMD) simulation is the method that utilizes interatomic potentials created by learning the results of first-principles calculations through neural networks. The MLMD enables us to conduct large-scale molecular dynamics calculations with almost first-principles calculation accuracy. In this presentation, I will report applications of machine learning molecular dynamics to nuclear material science. Examples of high-precision calculations using MLMD in materials such as clay, cement, nuclear fuel, and glass, as well as insights obtained from these calculations, will be reported.
Onodera, Naoyuki
no journal, ,
The Center for Computational Science and Engineering Center of Japan Atomic Energy Agency (CCSE) is developing real-time wind simulation and data assimilation methods as a wind digital twin for nuclear disaster prevention. In this presentation, we will show urban wind simulations and data assimilation of wind tunnel experiments on a GPU supercomputer SGI8600.
Kawamura, Takuma
no journal, ,
In the nuclear field, the DX approach to research and development is being promoted. Large-scale simulations involving a large number of experts require the following functions: 1) control of calculation conditions in real time, 2) remote VR visualization, and 3) multi-point collaboration. CCSE has developed a particle-based visualization technology, In-Situ PBVR, which compresses large-scale data into small particle data at the same time as supercomputer calculations, and transfers the data to a user's PC for real-time visualization. The In-Situ PBVR is a particle-based visualization technology developed by CCSE. In this study, we extended In-Situ PBVR to solve the above three issues and constructed a visualization infrastructure for multi-point remote VR visualization.