Presentation Information
[23p-1BN-11]Trial of non-empirical calculation of parameters for DPD simulation with the aid of machine learning #2
〇Hideo Doi1, Sota Matsuoka1, Koji Okuwaki1,2, Ryo Hatada1, Sojiro Minami1, Ryosuke Suhara1, Yuji Mochizuki1,3 (1.Rikkyo Univ., 2.JSOL Corp., 3.Univ. Tokyo)
Keywords:
Dissipative particle dynamics simulation,FMO,Machine learning
In response to the demand for highly functional materials in materials science and drug discovery, the need to predict and control mesoscale structures and behaviors has become increasingly important. However, dissipative particle dynamics (DPD) simulation requires parameters to represent molecular interactions, which poses a challenge in parameter development due to the time-consuming calculation of interaction energies between molecules. To solve this problem, we have explored the application of machine learning to replace the calculation of interaction energies with the ML predicted energy. The goal is to accelerate the parameter development process and reduce the computational cost.