量子级联激光器（Quantum Cascade Laser，QCL）是依赖电子在量子阱子带间跃迁辐射出光子而发生激射的单极型半导体激光器。大量的理论与实验研究已经证明轻微的外部扰动（如光反馈、光注入）或内部足够强的非线性模式耦合能够引起半导体激光器的非线性输出。QCL作为新型半导体器件，具有腔内强度高、子带间光学非线性强以及电子弛豫时间快等特点，激发了学者研究其非线性动力学的兴趣。本文详细综述了QCL的非线性动态特性研究进展情况，探究了QCL非线性动力学性质的产生机理，总结了QCL非线性特性的应用场景。

Diverse spatial mode bases can be exploited in mode-division multiplexing (MDM) to sustain the capacity growth in fiber-optic communications, such as linearly polarized (LP) modes, vector modes, LP orbital angular momentum (LP-OAM) modes, and circularly polarized OAM (CP-OAM) modes. Nevertheless, which kind of mode bases is more appropriate to be utilized in fiber still remains unclear. Here, we aim to find the superior mode basis in MDM fiber-optic communications via a system-level comparison in air-core fiber (ACF). We first investigate the walk-off effect of four spatial mode bases over 1-km ACF, where LP and LP-OAM modes show intrinsic mode walk-off, while it is negligible for vector and CP-OAM modes. We then study the mode coupling effect of degenerate vector and CP-OAM modes over 1-km ACF under fiber perturbations, where degenerate even and odd vector modes suffer severe mode cross talk, while negligible for high-order degenerate CP-OAM modes based on the laws of angular momentum conservation. Moreover, we comprehensively evaluate the system-level performance for data-carrying single-channel and two-channel MDM transmission with different spatial mode bases under various kinds of fiber perturbations (bending, twisting, pressing, winding, and out-of-plane moving). The obtained results indicate that the CP-OAM mode basis shows superiority compared to other mode bases in MDM fiber-optic communications without using multiple-input multiple-output digital signal processing. Our findings may pave the way for robust short-reach MDM optical interconnects for data centers and high-performance computing.

针对双模扰动下的烧蚀瑞利-泰勒不稳定性增长问题，采用高精度的数值计算方法，研究了不同预热程度下模耦合产生的多个高次谐波幅值的发展和演化问题。研究表明，三种预热烧蚀条件下，当扰动基模满足长波与短波耦合方式时，谐波中的长波模态占主导，而短波模发展明显受到抑制；当满足短波与短波耦合时，耦合结果带来了许多新的增长较快的长波模态，此时短波模增长呈现小幅震荡形式。比较两种耦合方式可以发现，长波结构在烧蚀瑞利-泰勒不稳定性弱非线性阶段都占主导地位，尤其是短波与短波耦合中气泡与尖钉表现出不同于两个基模的长波模结构。进一步分析预热效应对模耦合增长的影响，发现预热程度越强就越能削弱耦合谐波的增长，这说明预热对烧蚀瑞利-泰勒不稳定性具有致稳作用，这对惯性约束聚变工程中控制烧蚀瑞利-泰勒不稳定性发展具有重要意义。

针对涡旋光受大气湍流而产生的相位畸变问题，设计了卷积神经网络检测涡旋光束轨道角动量态。将相位畸变的拉盖尔高斯涡旋光光强图像作为样本数据输入，网络利用输入的数据集进行自主学习，经过多次迭代能够精确检测出光束高阶轨道角动量信息。仿真结果表明：在大气湍流强度不确定的情况下，模型对轨道角动量态范围为1~40、1~100、1~160的涡旋光束检测准确率分别为94%、90%、86%；在不同传输距离、不同径向指数、不同基模束腰半径以及不同波长下轨道角动量态范围为1~100的涡旋光束的检测准确率均达到78%以上，且具有很好的鲁棒性。该方法检测精度高、范围广、移植性强，可为涡旋光束轨道角动量态的检测与识别提供一种新思路。