1 August 2017 in Ocean Engineering
22 November 2017 in International Journal for Numerical Methods in Engineering
22 February 2023 in Complex & Intelligent Systems
In the complex tasks environment, efficient state feature learning is a key factor to improve the performance of the agent’s policy. When encountering a similar new environment, reinforcement learning agents usually need to learn from scratch. However, humans naturally have a common sense of the environment and are able to use prior knowledge to extract environmental state features. Although the prior knowledge may not be fully applicable to the new environment, it is able to speed up the learning process of the state feature. Taking this inspiration, we propose an artificial potential field-based reinforcement learning (APF-RL) method. The method consists of an artificial potential field state feature abstractor (APF-SA) and an artificial potential field intrinsic reward model (APF-IR). The APF-SA can introduce human knowledge to accelerate the learning process of the state feature. The APF-IR can generate an intrinsic reward to reduce the invalid exploration and guide the learning of the agent’s policy. We conduct experiments on PySC2 with different mini-games. The experimental results show that the APF-RL method achieves improvement in the learning efficiency compared to the benchmarks.
3 August 2017 in Advances in Mechanical Engineering
The hybrid atomistic–continuum coupling method based on domain decomposition serves as an important tool for the microfluidic simulation. However, major modifications to existing codes are often required to enable such simulations, which pose significant difficulties. In this article, in order to provide an efficient and easy-to-use software framework for field users, we propose a hybrid atomistic–continuum parallel coupling framework, named HACPar, based on open-source software platforms. We abstract the software architecture of the hybrid atomistic–continuum coupling framework based on geometric decomposition for the first time, demonstrate the detailed implementation of the framework, and present deep research on the coupling-oriented parallel issues which may improve the flexibility and efficiency of other multiscale parallel applications. The benchmark cases verify the correctness and efficiency of our HACPar framework. The benchmark results show that the scalability of the hybrid simulations is reached up to 1536 cores.
1 September 2019 in Journal of Applied Fluid Mechanics