In this paper, we present a fast mode decomposition method for few-mode fibers, utilizing a lightweight neural network called MobileNetV3-Light. This method can quickly and accurately predict the amplitude and phase information of different modes, enabling us to fully characterize the optical field without the need for expensive experimental equipment. We train the MobileNetV3-Light using simulated near-field optical field maps, and evaluate its performance using both simulated and reconstructed near-field optical field maps. To validate the effectiveness of this method, we conduct mode decomposition experiments on a few-mode fiber supporting six linear polarization (LP) modes (LP01, LP11e, LP11o, LP21e, LP21o, LP02). The results demonstrate a remarkable average correlation of 0.9995 between our simulated and reconstructed near-field light-field maps. And the mode decomposition speed is about 6 ms per frame, indicating its powerful real-time processing capability. In addition, the proposed network model is compact, with a size of only 6.5 MB, making it well suited for deployment on portable mobile devices.

In this paper, a simple theoretical model combining Monte Carlo simulation with the enthalpy method is provided to simulate the damage resistance of B₄C/Si-sub mirror under X-ray free-electron laser irradiation. Two different damage mechanisms are found, dependent on the photon energy. The optimum B₄C film thickness is determined by studying the dependence of the damage resistance on the film thickness. Based on the optimized film thickness, the damage thresholds are simulated at photon energy of 0.4–25 keV and a grazing incidence angle of 2 mrad. It is recommended that the energy range around the Si K-edge should be avoided for safety reasons.

Fiber lasers are a paradigm of dissipative systems, which distinguish themselves from a Hamilton system where energy is conservative. Consequently, pulses generated in a fiber laser are always accompanied by the continuous wave (CW). Under certain hypothesis, pulses generated in a fiber laser can be considered as a soliton, a product of a Hamilton system. Therefore, all the descriptions of solitons of a fiber laser are approximate. Coexistence of solitons and the CW from a fiber laser prevents unveiling of real nonlinear dynamics in fiber lasers, such as soliton interactions. Pulse behavior in a fiber laser can be represented by the state of single pulse, the state of period doubling of single pulse, the states of two pulses either tightly bound or loosely distributed, the states of three pulses, and various combinations of the above-mentioned states. Recently, soliton distillation was proposed and numerically demonstrated based on the nonlinear Fourier transform (NFT) [J. Lightwave Technol.39, 2542 (2021)JLTEDG0733-872410.1109/JLT.2021.3051036]. Solitons can be separated from the coherent CW background. Therefore, it is feasible to isolate solitons from CW background in a fiber laser. Here, we applied the NFT to various pulses generated in a fiber laser, including single pulse, single pulse in period doubling, different double pulses, and multiple pulses. Furthermore, with the approach of soliton distillation, the corresponding pure solitons of those pulses are reconstructed. Simulation results suggest that the NFT can be used to identify soliton dynamics excluding CW influence in a fiber laser, which paves a new way for uncovering real soliton interaction in nonlinear systems.