From the 2015 fiscal year, the Materials Design and Characterization Laboratory (MDCL) at ISSP has started the project for advancement of software usability in materials science to enhance the usability of the supercomputer system of ISSP. We develop and enhance the usability of programs adopted in this project, release them as open source software, and support dissemination activities such as supporting hands-on lectures. In addition, by using the developed software, we theoretically study research subjects in a wide range of fields such as derivation and analysis of low-energy effective Hamiltonian of organic conductors and spin relaxation phenomena in quantum dot systems. In addition to these activities, we focus on the information processing and have been trying to apply this technique to materials science such as analyzing data obtained by the quantum Monte Carlo method by the sparse modeling method and searching new materials using the machine learning method.
- A paper related to abICS has been published in The Journal of Chemical Physics.
- A paper "Facilitating ab initio configurational sampling of multicomponent solids using an on-lattice neural network model and active learning" related to abICS has been published in The Journal of Chemical Physics. This paper discusses structural stability using aenet, a software that uses neural networks to calculate interatomic potentials. This functionality has been implemented in abICS ver. 2.0 and is already available. Report information："Facilitating ab initio configurational sampling of multicomponent solids...
Koretsune Group. Condensed Physics Theory Group, Department of Physics, Tohoku University