Rogue waves were first discovered in the ocean and refer to huge waves that suddenly occur in a calm ocean. Such extremely powerful waves are extremely destructive, which is further compounded by their difficulty in predicting. In addition to the ocean, it is also found in various complex systems, including financial models, Bose-Einstein condensates, etc., but its unpredictability and huge amplitude make it difficult to generate and control in the laboratory.
Fiber optic systems can record large amounts of data in a relatively short period of time, providing an ideal platform for observing the dynamic behavior of giant wave phenomena. In recent years, many optical systems have been combined with machine learning techniques to study steady-state or recurring events, such as solitons or breathing solitons. However, using machine learning to control giant waves remains a challenge because their occurrence is rare and unpredictable.
Recently, the research team of Professor Zeng Heping used a new machine learning method to generate the strongest optical giant wave to date based on mode-locked fiber lasers. This optical system can generate giant waves that are 33 times more powerful than the effective waves. Significant wave height is a parameter that defines giant waves. Waves that are twice stronger than it are called rogue waves.
Genetic algorithm is a machine learning model based on the evolutionary principle of natural selection and has been widely used to solve search and optimization problems in nonlinear optics and other complex systems. In this work, real-time spectral measurement data are used as inputs to a genetic algorithm, which is used to optimize the parameters within the laser cavity to produce giant waves with controllable intensity. This intelligent control system can generate ordinary giant waves and super giant waves. The work found that super giant waves in optical systems are more like giant waves in the ocean, because like giant waves in the ocean, they come and go without a trace.
By analyzing massive optical giant wave data obtained through machine learning, this work also revealed a new mechanism for generating giant waves - the asymmetry of the initial pulse spectrum. This asymmetry will be continuously amplified by the laser gain, and combined with the Kerr effect, will eventually cause one edge of the spectrum to evolve into a high-energy giant wave, and eventually lead to pulse collapse. This dynamic process well explains the formation and annihilation mechanism of giant waves. The team further confirmed this dynamic process through numerical simulations of laser dynamics based on the nonlinear Schrödinger equation.
The research results were published in Laser & Photonics Reviews (2023): 2200470 with the State Key Laboratory of Precision Spectroscopy of East China Normal University as the first unit. The "OPTICS & PHOTONICS NEWS" column of the Optical Society of America also reported on this work.
Figure 1: Schematic diagram of intelligently controlled optical rogue waves.
Source:【科研动态】机器学习调控光学巨波(qq.com)
Paper link:Control of spectral extreme events in ultrafast fiber lasers by a genetic algorithm
