torchquad is a Python module for n-dimensional numerical integration optimized for graphics processing units (GPUs). Various deterministic and stochastic integration methods, such as Newton-Cotes formulas and Monte Carlo integration methods like VEGAS Enhanced (Lepage, 2020), are available for computationally efficient integration for arbitrary dimensionality n d . As it is implemented using PyTorch (Paszke et al., 2019), one of the most popular machine learning frameworks, torchquad provides fully automatic differentiation throughout the integration, which is essential for many machine learning applications.

Flip-chip ultraviolet light-emitting diodes based on self-assembled GaN/AlGaN nanocolumns have been fabricated, exploiting single-layer graphene not only as a growth substrate but also as a transparent conducting electrode. High crystalline quality of the nanocolumns is confirmed by detailed electron microscopy characterization, also showing the intrinsic GaN quantum disk in the active region of the nanocolumns. These features are further confirmed in the optical emission, where the absence of defect-related yellow emission and the presence of blue-shifted (from the usual 365 nm band gap emission of bulk wurtzite GaN) emission at ∼350 nm, ascribed to quantum confinement and strain effects, are observed. Despite a noticeable graphene damage after the nanocolumn growth that causes high sheet resistance of graphene and high turn-on voltage, the proof of concept of single-layer graphene used as the transparent conducting substrate for a nanocolumn device is demonstrated. This study offers an alternative platform for the fabrication of next-generation nano-optoelectronic and electronic devices.

In the recent decade, computational tools have become central in material design, allowing rapid development cycles at reduced costs. Machine learning tools are especially on the rise in photonics. However, the inversion of the Maxwell equations needed for the design is particularly challenging from an optimization standpoint, requiring sophisticated software. We present an innovative, open-source software tool called Neural Inverse Design of Nanostructures (NIDN) that allows designing complex, stacked material nanostructures using a physics-based deep learning approach. Instead of a derivative-free or data-driven optimization or learning method, we perform a gradient-based neural network training where we directly optimize the material and its structure based on its spectral characteristics. NIDN supports two different solvers, rigorous coupled-wave analysis and a finite-difference time-domain method. The utility and validity of NIDN are demonstrated on several synthetic examples as well as the design of a 1550 nm filter and anti-reflection coating. Results match experimental baselines, other simulation tools, and the desired spectral characteristics. Given its full modularity in regard to network architectures and Maxwell solvers as well as open-source, permissive availability, NIDN will be able to support computational material design processes in a broad range of applications.

In the recent decade, computational tools have become central in material design, allowing rapid development cycles at reduced costs. Machine learning tools are especially on the rise in photonics. However, the inversion of the Maxwell equations needed for the design is particularly challenging from an optimization standpoint, requiring sophisticated software. We present an innovative, open-source software tool called Neural Inverse Design of Nanostructures (NIDN) that allows designing complex, stacked material nanostructures using a physics-based deep learning approach. Instead of a derivative-free or data-driven optimization or learning method, we perform a gradient-based neural network training where we directly optimize the material and its structure based on its spectral characteristics. NIDN supports two different solvers, rigorous coupled-wave analysis and a finite-difference time-domain method. The utility and validity of NIDN are demonstrated on several synthetic examples as well as the design of a 1550 nm filter and anti-reflection coating. Results match experimental baselines, other simulation tools, and the desired spectral characteristics. Given its full modularity in regard to network architectures and Maxwell solvers as well as open-source, permissive availability, NIDN will be able to support computational material design processes in a broad range of applications.