Abstract
Abstract
The performance of liquid crystal displays (LCDs) is expected to be improved further with the emergence of their new applications. Numerical simulations such as the finite differential method (FDM) and the finite element method are useful in optimum design. However, they take a long time because dynamical systems in LCDs are nonlinear multiphysics composed of electromagnetism, fluid dynamics, and elastic mechanics. A machine learning method is one of the solutions to reduce computational cost. In this paper, we have extended the parallel reservoir computing framework and applied it to LCD simulation. We have discussed how to implement each natural feature of liquid crystal cells, namely, non-autonomy, multiphysics and long-range orientational order, into the framework of parallel reservoir computing. Sufficient higher accuracy was obtained with several display patterns and driving frequencies at computational speeds more than 100 times higher than FDM.
Funder
Tateisi Science and Technology Foundation
KAKENHI
Subject
General Physics and Astronomy,Physics and Astronomy (miscellaneous),General Engineering
Cited by
1 articles.
订阅此论文施引文献
订阅此论文施引文献,注册后可以免费订阅5篇论文的施引文献,订阅后可以查看论文全部施引文献