Search

This paper proposes a novel neural network framework, denoted as spectral integrated neural networks (SINNs), for resolving three-dimensional forward and inverse dynamic problems. In the SINNs, the spectral integration method is applied to perform temporal discretization, and then a fully connected neural network is adopted to solve resulting partial differential equations (PDEs) in the spatial domain. Specifically, spatial coordinates are employed as inputs in the network architecture, and the output layer is configured with multiple outputs, each dedicated to approximating solutions at different time instances characterized by Gaussian points used in the spectral method. By leveraging the automatic differentiation technique and spectral integration scheme, the SINNs minimize the loss function, constructed based on the governing PDEs and boundary conditions, to obtain solutions for dynamic problems. Additionally, we utilize polynomial basis functions to expand the unknown function, aiming to enhance the performance of SINNs in addressing inverse problems. The conceived framework is tested on six forward and inverse dynamic problems, involving nonlinear PDEs. Numerical results demonstrate the superior performance of SINNs over the popularly used physics-informed neural networks in terms of convergence speed, computational accuracy and efficiency. It is also noteworthy that the SINNs exhibit the capability to deliver accurate and stable solutions for long-time dynamic problems.

Similar to the carbon nanotube fabricated from graphene sheet, a black phosphorus nanotube (BPNT) also can theoretically be produced by curling the rectangular single-layer black phosphorus (SLBP). In present study, the effect of thermal vibration of atoms on the failure of a BPNT is investigated using molecular dynamics simulations. Two types of double-shell BPNTs, which are obtained by curling the rectangular SLBP along its armchair/pucker direction and zigzag direction (in-plane normal) respectively, are involved in simulation. At finite temperature, a bond on the outer shell of tube is under tension due to both of curvature of tube and serious thermal vibration of atoms. As the length of a bond with such elongation approaches its critical value, i.e., 0.279 nm, or the smallest distance between two nonbonding phosphorus atoms is over 0.389nm caused by great variation of bond angle, the tube fails quickly. The critical stable states of either an armchair or a zigzag BPNT at finite temperature are calculated and compared. To achieve a stable BPNT with high robustness, the curvature of the tube should be reduced or the tube should work at a lower temperature. Only when the BPNT has structural stability, it has a potential application as a nanowire in a future nano electro-mechanical system (NEMS).