Entropy-Regularized Wasserstein Distributionally Robust Optimization Uncertainty in data poses a central challenge in operations research. Distributionally robust optimization (DRO) offers a principled framework for addressing this challenge by producing solutions resilient to distributional variations. Among various DRO approaches, the Wasserstein DRO has received significant attention though its computational efficiency relies on stringent assumptions, and its worst case distributions are typically discrete. In “Sinkhorn Distributionally Robust Optimization,” Wang, Gao, and Xie leverage the Sinkhorn distance—an entropy-regularized variant of the Wasserstein distance—to more realistically model uncertainty, enhancing computational efficiency. The authors establish a strong duality reformulation and propose a first order stochastic mirror descent algorithm with provable complexity guarantees for general loss functions. Unlike Wasserstein DRO, Sinkhorn DRO yields continuous worst case distributions, offering a more flexible representation of practical uncertainties. Extensive experiments in the newsvendor problem, portfolio optimization, and adversarial classification demonstrate its superior performance in both out-of-sample performance and efficiency.
We have designed three mode order converters using the sub-wavelength grating on the silicon-on-insulator substrate. The proposed mode order converters can separately realize the mode order conversion from the fundamental transverse electric mode to the first-order transverse electric mode (TE0-to-TE1), the second-order transverse electric mode (TE0-to-TE2) and the third-order transverse electric mode (TE0-to-TE3) with compact device sizes and good performances. The simulation results show that the mode order conversion efficiencies of TE0-to-TE1, TE0-to-TE2 and TE0-to-TE3 are larger than 94.4%, 95.7% and 83.7% in the wavelength ranging from 1.5 µm to 1.6 µm, the corresponding device length are 8.72 µm, 4.98 µm and 14.54 µm. In addition, the mode order converter can be fabricated with only once etching which will be advantage to the device fabrication.
An InGaAs/AlGaAs quantum-well laser structure was grown on a silicon substrate by adopting a three-step grown thin (1.8 μm) and simple buffer layer. The whole structure was grown by metalorganic chemical vapor deposition. The material quality was characterized by transmission electron microscopy, photoluminescence spectra, and electrochemical capacitance-voltage profiler. It shows that the threading dislocation density and interface roughness are effectively reduced, and the active region's optical properties grown on a silicon substrate are comparable to that grown on a GaAs substrate. Broad stripe lasers have also been fabricated. The cavity length and stripe width are 1 mm and 15 μm, respectively. An extremely low-threshold current density of 313 A/cm 2 has been achieved under pulsed condition at room temperature. Meanwhile, the laser can operate under continuous wave condition at the temperature of 240 K. The above results make us more confident for realizing better performance of Si-based III-V semiconductor lasers by further improvement of the material growth method.
We propose and numerically analyze a broadband and high extinction ratio mode converter to achieve the mode conversion between the fundamental transverse electric mode (TE0) and the first-order transverse magnetic mode (TM1). The mode converter is achieved by using the tapered hybrid plasmonic waveguide which the metal is directly covered on the Si waveguide without a thin low-index layer. After optimizing the structure of the tapered hybrid plasmonic waveguide, a wide operation bandwidth of 100 nm is obtained with the extinction ratio of the TM1 mode and TE0 mode larger than 20 dB in the output Si waveguide for the forward propagation. And for the back propagation, the operation bandwidth over 65 nm is obtained with the extinction ratio larger than 20 dB in the output Si waveguide. The whole footprint of the mode converter is only 0.8×11 μm2. Fabrication tolerance analysis is also demonstrated
Room-temperature continuous-wave operation of InGaAs/AlGaAs quantum well lasers directly grown on on-axis silicon (001) has been demonstrated. A 420 nm thick GaAs epilayer completely free of antiphase domains was initially grown on the silicon substrate in a metal-organic chemical vapor deposition system and the other epilayers, including four sets of five-period strained-layer superlattices and the laser-structural layers, were successively grown in a molecular beam epitaxy system. The lasers were prepared as broad-stripe Fabry–Pérot ones with a stripe width of 21.5 μm and a cavity length of 1 mm. Typically, the threshold current and the corresponding threshold current density are 186.4 mA and 867 A/cm2, respectively. The lasing wavelength is around 980 nm, and the slope efficiency is 0.097 W/A with a single-facet output power of 22.5 mW at an injection current of 400 mA. This advancement makes the silicon-based monolithic optoelectronic integration relevant to quantum well lasers more promising with an enhanced feasibility.
We report electrically pumped continuous-wave (CW) InAs/GaAs quantum dot lasers directly grown on planar exact silicon (001) with asymmetric waveguide structures. Surface hydrogen-annealing for the GaAs/ Si (001) templates and low-temperature growth for GaInP upper cladding layers were combined in the growth of the laser structure to achieve a high slope efficiency. For the broad-stripe edge-emitting lasers with 2-mm cavity length and 20-µm stripe width made from the above laser structure, a threshold current density of 203.5 A/cm2 and a single-facet slope efficiency of 0.158 W/A are achieved at ∼1.31 µm band under CW conditions. The extrapolated mean-time-to-failure reaches up to 21000 hours at room temperature, which is deduced from the data measured from C-mount packaged devices. Importantly, these results can provide a practical strategy to realize 1.3 µm wavelength band distributed feedback lasers directly on planar exact Si (001) templates with thin buffer layers.