Pointing jitter correction in hollow-core fiber nonlinear systems based on LSTM

Hollow-core fiber has been widely used for laser spectrum broadening and pulse duration compression, and its performance is highly sensitive to pointing instability caused by environmental interference and system nonlinearity. Here, we propose a feedback control system based on a three-layer long short-term memory network for real-time pointing jitter correction, which reduces the beam pointing jitter from 4.04 and 4.64 µrad to 0.869 and 0.903 µrad in the horizontal and vertical directions, respectively. Our work demonstrates the application of machine learning in pointing stabilization of nonlinear systems and paves the way for highly stable few-cycle laser pulse generation.