In the past few years, many groups have focused on the research and development of GaN-based ultraviolet laser diodes (UV LDs). Great progresses have been achieved even though many challenges exist. In this article, we analyze the challenges of developing GaN-based ultraviolet laser diodes, and the approaches to improve the performance of ultraviolet laser diode are reviewed. With these techniques, room temperature (RT) pulsed oscillation of AlGaN UVA (ultraviolet A) LD has been realized, with a lasing wavelength of 357.9 nm. Combining with the suppression of thermal effect, the high output power of 3.8 W UV LD with a lasing wavelength of 386.5 nm was also fabricated.

Electromagnetic topological chiral edge states mimicking the quantum Hall effect have attracted a great deal of attention due to their unique features of free backscattering and immunity against sharp bends and defects. However, the matching techniques between classical waveguides and the topological one-way waveguide deserve more attention for real-world applications. In this paper, a highly efficient conversion structure between a classical rectangular waveguide and a topological one-way waveguide is proposed and demonstrated at the microwave frequency, which efficiently converts classical guided waves to topological one-way edge states. A tapered transition is designed to match both the momentum and impedance of the classical guided waves and the topological one-way edge states. With the conversion structure, the waves generated by a point excitation source can be coupled to the topological one-way waveguide with very high coupling efficiency, which can ensure high transmission of the whole system (i.e., from the source and the receiver). Simulation and measurement results demonstrate the proposed method. This investigation is beneficial to the applications of topological one-way waveguides and opens up a new avenue for advanced topological and classical integrated functional devices and systems.

Valley Hall topological photonic crystals, inspired by topological insulators in condensed matter physics, have provided a promising solution to control the flow of light. Recently, the dynamic manipulation property of topological photonic crystals has been widely studied. Here, we propose a novel solution for programmable valley photonic crystals, called field programmable topological edge array (FPTEA), based on the field reorientation property of nematic liquid crystals and robust valley-protected edge modes. FPTEA is composed of an array of graphene-like lattices with C3 symmetry, in which the birefringence of liquid crystal is larger than 0.5105. Due to the dielectric anisotropy of liquid crystals being sensitive to external fields such as light, heat, electric, and magnetic fields, each lattice is tunable, and the topological propagation routes and even the lattice parameters can be dynamically changed while changing the distribution of external fields. We numerically demonstrate three methods of composing an FPTEA device to design arbitrary passive optical devices by electric driving, thermal inducing, or UV writing. These results show the great application potential of liquid crystals in topological photonic crystals, and enrich the design of programmable integrated topological devices with broad working bandwidth ranging from microwave to visible light.

Abstract

In this work, we reported the room-temperature continuous-wave operation of 6.0 W GaN-based blue laser diode (LD), and its stimulated emission wavelength is around 442 nm. The GaN-based high power blue LD is grown on a c-plane GaN substrate by metal organic chemical vapor deposition (MOCVD), and the width and length of the ridge waveguide structure are 30 and 1200 μm, respectively. The threshold current is about 400 mA, and corresponding threshold current density is 1.1 kA/cm².

The high performance of the state-of-the-art deep neural networks (DNNs) is acquired at the cost of huge consumption of computing resources. Quantization of networks is recently recognized as a promising solution to solve the problem and significantly reduce the resource usage. However, the previous quantization works have mostly focused on the DNN inference, and there were very few works to address on the challenges of DNN training. In this paper, we leverage dynamic fixed-point (DFP) quantization algorithm and stochastic rounding (SR) strategy to develop a fully quantized 8-bit neural networks targeting low bitwidth training. The experiments show that, in comparison to the full-precision networks, the accuracy drop of our quantized convolutional neural networks (CNNs) can be less than 2%, even when applied to deep models evaluated on ImageNet dataset. Additionally, our 8-bit GNMT translation network can achieve almost identical BLEU to full-precision network. We further implement a prototype on FPGA and the synthesis shows that the low bitwidth training scheme can reduce the resource usage significantly.