Optical frequency conversion based on the second-order nonlinearity (χ(2)) only occurs in anisotropic media (or at interfaces) and thus is intrinsically polarization-dependent. But for practical applications, polarization-insensitive or independent operation is highly sought after. Here, by leveraging polarization coupling and second-order nonlinearity, we experimentally demonstrate a paradigm of TE/TM polarization-independent frequency upconversion, i.e., sum frequency generation, in the periodically poled lithium niobate-on-insulator ridge waveguide. The cascading of quasi-phase-matched polarization coupling and nonlinear frequency conversion is exploited. With a proper transverse electric field, TE and TM mode fundamental waves can be frequency-upconverted with an equal efficiency in the frequency converter. The proposed method may find ready application in all-optical wavelength conversion and upconversion detection technologies.

Microcavities constructed from materials with a second-order nonlinear coefficient have enabled efficient second-harmonic (SH) generation at a low power level. However, it is still technically challenging to realize double resonance with large nonlinear modal overlap in a microcavity. Here, we propose a design for a robust, tunable, and easy coupling double-resonance SH generation based on the combination of a newly developed fiber-based Fabry–Perot microcavity and a sandwich structure, whose numerical SH conversion efficiency is up to 3000% W^-1. This proposal provides a feasible way to construct ultra-efficient nonlinear devices for generation of classical and quantum light sources.

VGGNet能提供高精度的火星图像分类，但需消耗大量内存资源。鉴于器载计算机内存资源有限，为解决这一矛盾，本文提出了基于迭代剪枝VGGNet的火星图像分类方法。首先，采用迁移学习训练网络的连通性，以便评估神经元的重要性；其次，通过迭代剪枝方法修剪不重要的神经元，以便将全连接层的参数量和内存占用量减少；最后，采用K-means++聚类实现权重参数的量化，利用霍夫曼编码压缩迭代剪枝与量化后的VGGNet权重参数，达到减少存储量和浮点数运算量的作用。此外，通过5种数据增强方法进行数据扩充，目的是解决类别不平衡的问题。实验结果表明，压缩后的VGGNet模型的所占内存、Flops和准确率分别为62.63 Mb、150.6 MFlops和96.15%。与ShuffleNet、MobileNet和EfficientNet等轻量级图像分类算法相比，所提模型具有更好的性能。

Epitaxial quantum dots formed by III–V compound semiconductors are excellent sources of non-classical photons, creating single photons and entangled multi-photon states on demand. Their semiconductor nature allows for a straightforward combination with mature integrated photonic technologies, leading to novel functional devices at the single-photon level. Integrating a quantum dot into a carefully engineered photonic cavity enables control of the radiative decay rate using the Purcell effect and the realization of photon–photon nonlinear gates. In this review, we introduce the basis of epitaxial quantum dots and discuss their applications as non-classical light sources. We highlight two interfaces—one between flying photons and the quantum-dot dipole, and the other between the photons and the spin. We summarize the recent development of integrated photonics and reconfigurable devices that have been combined with quantum dots or are suitable for hybrid integration. Finally, we provide an outlook of employing quantum-dot platforms for practical applications in large-scale quantum computation and the quantum Internet.

超高速激光熔覆是新兴的表面涂层技术，通过粉末和激光的最佳耦合实现熔覆效率的大幅提升，相比传统激光熔覆涂层可以获得更优质的表面质量，且对基材损伤更小。通过分析超高速激光熔覆的原理及技术优势，并与传统激光熔覆技术特点进行对比，总结了激光功率、扫描速率、送粉速度以及搭接率对熔覆层成形的影响，详细介绍了超高速激光熔覆涂层的硬度、耐磨、耐蚀等关键性能的研究现状，并且列举了国内外对超高速激光熔覆技术在工业应用现状。最后，基于目前的研究进展，指出目前对于超高速激光熔覆涂层和基体界面结合状态及涂层构件力学方面处于研究空白，并对该技术的发展提出展望。以期为超高速激光熔覆技术的广泛应用提供理论支持。