1.3 kV Vertical GaN-Based Trench MOSFETs on 4-Inch Free Standing GaN Wafer

In this work, a vertical gallium nitride (GaN)-based trench MOSFET on 4-inch free-standing GaN substrate is presented with threshold voltage of 3.15 V, specific on-resistance of 1.93 mΩ·cm2, breakdown voltage of 1306 V, and figure of merit of 0.88 GW/cm2. High-quality and stable MOS interface is obtained through two-step process, including simple acid cleaning and a following (NH4)2S passivation. Based on the calibration with experiment, the simulation results of physical model are consistent well with the experiment data in transfer, output, and breakdown characteristic curves, which demonstrate the validity of the simulation data obtained by Silvaco technology computer aided design (Silvaco TCAD). The mechanisms of on-state and breakdown are thoroughly studied using Silvaco TCAD physical model. The device parameters, including n−-GaN drift layer, p-GaN channel layer and gate dielectric layer, are systematically designed for optimization. This comprehensive analysis and optimization on the vertical GaN-based trench MOSFETs provide significant guide for vertical GaN-based high power applications.

Compressing Aerodynamic Hazard Data

Data compression of three-dimensional computational fluid dynamics (CFD) simulation data is crucial to allow effective data-streaming for drone navigation and control. This problem is computationally challenging due to the complexity of the geometrical features present in the CFD data, and cannot be tackled by standard compression techniques such as sphere-tree. In this report, we present two different methods based on octree and cuboid primitives to compress velocity isosurfaces and volumetric data in three dimensions. Our volume compression method achieves a 1400 compression rate of raw simulation data and allows parallel computing.

Deep Learning Strategies for ProtoDUNE Raw Data Denoising

In this work, we investigate different machine learning-based strategies for denoising raw simulation data from the ProtoDUNE experiment. The ProtoDUNE detector is hosted by CERN and it aims to test and calibrate the technologies for DUNE, a forthcoming experiment in neutrino physics. The reconstruction workchain consists of converting digital detector signals into physical high-level quantities. We address the first step in reconstruction, namely raw data denoising, leveraging deep learning algorithms. We design two architectures based on graph neural networks, aiming to enhance the receptive field of basic convolutional neural networks. We benchmark this approach against traditional algorithms implemented by the DUNE collaboration. We test the capabilities of graph neural network hardware accelerator setups to speed up training and inference processes.

Unraveling membrane properties at the organelle-level with LipidDyn

Cellular membranes are formed from many different lipids in various amounts and proportions depending on the subcellular localization. The lipid composition of membranes is sensitive to changes in the cellular environment, and their alterations are linked to several diseases, including cancer. Lipids not only form lipid-lipid interactions but also interact with other biomolecules, including proteins, profoundly impacting each other. Molecular dynamics (MD) simulations are a powerful tool to study the properties of cellular membranes and membrane-protein interactions on different timescales and at varying levels of resolution. Over the last few years, software and hardware for biomolecular simulations have been optimized to routinely run long simulations of large and complex biological systems. On the other hand, high-throughput techniques based on lipidomics provide accurate estimates of the composition of cellular membranes at the level of subcellular compartments. The community needs computational tools for lipidomics and simulation data effectively interacting to better understand how changes in lipid compositions impact membrane function and structure. Lipidomic data can be analyzed to design biologically relevant models of membranes for MD simulations. Similar applications easily result in a massive amount of simulation data where the bottleneck becomes the analysis of the data to understand how membrane properties and membrane-protein interactions are changing in the different conditions. In this context, we developed LipidDyn, an in silico pipeline to streamline the analyses of MD simulations of membranes of different compositions. Once the simulations are collected, LipidDyn provides average properties and time series for several membrane properties such as area per lipid, thickness, diffusion motions, the density of lipid bilayers, and lipid enrichment/depletion. The calculations exploit parallelization and the pipelines include graphical outputs in a publication-ready form. We applied LipidDyn to different case studies to illustrate its potential, including membranes from cellular compartments and transmembrane protein domains. LipidDyn is implemented in Python and relies on open-source libraries. LipidDyn is available free of charge under the GNU General Public License from https://github.com/ELELAB/LipidDyn.

WaveRange: wavelet-based data compression for three-dimensional numerical simulations on regular grids

A wavelet-based method for compression of three-dimensional simulation data is presented and its software framework is described. It uses wavelet decomposition and subsequent range coding with quantization suitable for floating-point data. The effectiveness of this method is demonstrated by applying it to example numerical tests, ranging from idealized configurations to realistic global-scale simulations. The novelty of this study is in its focus on assessing the impact of compression on post-processing and restart of numerical simulations. Graphical abstract

Impact of a science and technology strategy on the learning of physics

The use of smartphones and some applications for educational purposes are valuable tools in the laboratory since they are motivating for students and the teacher can take advantage of this advantage for the teaching of physics. The experience is based on the anthropological theory of didactics and the teaching approach in science, technology, engineering, and mathematics. It is proposed to investigate a trigger question in physics. To respond, an application is used that uses the smartphone’s sensors to record the simulation data. The experience is described, and results of its implementation are presented. Methodologically, a qualitative descriptive approach was used in a group of tenth grade students taking the physics course. Finally, it is concluded that the students felt motivated since they felt they participated in the construction of their own learning, supported using technologies that facilitate the integration of knowledge in physics.

Actor-Critic–Like Stochastic Adaptive Search for Continuous Simulation Optimization

Many systems arising in applications from engineering design, manufacturing, and healthcare require the use of simulation optimization (SO) techniques to improve their performance. In “Actor-Critic–Like Stochastic Adaptive Search for Continuous Simulation Optimization,” Q. Zhang and J. Hu propose a randomized approach that integrates ideas from actor-critic reinforcement learning within a class of adaptive search algorithms for solving SO problems. The approach fully retains the previous simulation data and incorporates them into an approximation architecture to exploit knowledge of the objective function in searching for improved solutions. The authors provide a finite-time analysis for the method when only a single simulation observation is collected at each iteration. The method works well on a diverse set of benchmark problems and has the potential to yield good performance for complex problems using expensive simulation experiments for performance evaluation.