•We applied an unsupervised deep learning algorithm to inflow generation of turbulent channel flows.•We combined GAN and RNN to generate 2D time-varying turbulent flows.•The trained network could ...generate the flows at various Reynolds numbers, outside of trained one.•We could achieve high statistical accuracy compared to DNS.
A realistic inflow boundary condition is essential for successful simulation of the developing turbulent boundary layer or channel flows. In the present work, we applied generative adversarial networks (GANs), a representative of unsupervised learning, to generate an inlet boundary condition of turbulent channel flow. Upon learning the two-dimensional spatial structure of turbulence using data obtained from direct numerical simulation (DNS) of turbulent channel flow, the GAN could generate instantaneous flow fields that are statistically similar to those of DNS. After learning data at only three Reynolds numbers, the GAN could produce fields at various Reynolds numbers within a certain range without additional simulation. Eventually, through a combination of the GAN and a recurrent neural network (RNN), we developed a novel model (RNN-GAN) that could generate time-varying fully developed flow for a long time. The spatiotemporal correlations of the generated flow are in good agreement with those of the DNS. This proves the usefulness of unsupervised learning in the generation of synthetic turbulence fields.
Recent attempts to use deep learning for super-resolution reconstruction of turbulent flows have used supervised learning, which requires paired data for training. This limitation hinders more ...practical applications of super-resolution reconstruction. Therefore, we present an unsupervised learning model that adopts a cycle-consistent generative adversarial network (CycleGAN) that can be trained with unpaired turbulence data for super-resolution reconstruction. Our model is validated using three examples: (i) recovering the original flow field from filtered data using direct numerical simulation (DNS) of homogeneous isotropic turbulence; (ii) reconstructing full-resolution fields using partially measured data from the DNS of turbulent channel flows; and (iii) generating a DNS-resolution flow field from large-eddy simulation (LES) data for turbulent channel flows. In examples (i) and (ii), for which paired data are available for supervised learning, our unsupervised model demonstrates qualitatively and quantitatively similar performance as that of the best supervised learning model. More importantly, in example (iii), where supervised learning is impossible, our model successfully reconstructs the high-resolution flow field of statistical DNS quality from the LES data. Furthermore, we find that the present model has almost universal applicability to all values of Reynolds numbers within the tested range. This demonstrates that unsupervised learning of turbulence data is indeed possible, opening a new door for the wide application of super-resolution reconstruction of turbulent fields.
We propose an Eulerian approach to investigate the motion of particles in turbulence under the assumption that the motion of particles remains smooth in space and time until a collision between ...particles occurs. When the first collision happens, particle velocity loses Formula: see text continuity, resulting in a finite-time blowup. The corresponding singularities in particle velocity gradient, particle number density, and particle vorticity for various Stokes numbers and gravity factors are numerically investigated for the first time in a simple two-dimensional Taylor-Green vortex flow, two-dimensional decaying turbulence, and three-dimensional isotropic turbulence. In addition to the critical Stokes number above which a collision begins to occur, the flow condition leading to collision is revealed; particles tend to collide in very thin shear layer constructed by two parallel same-signed vortical structures when Stokes number is above the critical one.
High-temperature ferromagnetic two-dimensional (2D) materials with flat surfaces have been a long-sought goal due to their potential in spintronics applications. Through comprehensive ...first-principles calculations, we show that the recently synthesized MoN2 monolayer is such a material; it is ferromagnetic with a Curie temperature of nearly 420 K, which is higher than that of any flat 2D magnetic materials studied to date. This novel property, made possible by the electron-deficient nitrogen ions, render transition-metal dinitrides monolayers with unique electronic properties which can be switched from the ferromagnetic metals in MoN2, ZrN2, and TcN2 to half-metallic ones in YN2. Transition-metal dinitrides monolayers may, therefore, serve as good candidates for spintronics devices.
Feedback field-effect transistor (FBFET), an alternative switching device, has received attention due to its ideal steep switching feature. By utilizing the positive feedback phenomenon, the total ...amount of electrons and holes contributing to drain current is sharply surged. Although the device has conspicuous subthreshold slope (SS) properties, advanced research for structure and performance of it is lacking. In this paper, single-gated and spacer-less silicon-on-insulator (SOI) FBFET with extremely steep switching (~1 mV/decade) characteristic is studied in various aspects; SS attribute, performance variation of scaled FBFET, the impact of structural variation, the gate margin for the device layout, and the hysteresis window. The prospect of SOI FBFET as a future candidate for CMOS logic application is investigated in detail.
Three-dimensional physical interactions within chromosomes dynamically regulate gene expression in a tissue-specific manner. However, the 3D organization of chromosomes during human brain development ...and its role in regulating gene networks dysregulated in neurodevelopmental disorders, such as autism or schizophrenia, are unknown. Here we generate high-resolution 3D maps of chromatin contacts during human corticogenesis, permitting large-scale annotation of previously uncharacterized regulatory relationships relevant to the evolution of human cognition and disease. Our analyses identify hundreds of genes that physically interact with enhancers gained on the human lineage, many of which are under purifying selection and associated with human cognitive function. We integrate chromatin contacts with non-coding variants identified in schizophrenia genome-wide association studies (GWAS), highlighting multiple candidate schizophrenia risk genes and pathways, including transcription factors involved in neurogenesis, and cholinergic signalling molecules, several of which are supported by independent expression quantitative trait loci and gene expression analyses. Genome editing in human neural progenitors suggests that one of these distal schizophrenia GWAS loci regulates FOXG1 expression, supporting its potential role as a schizophrenia risk gene. This work provides a framework for understanding the effect of non-coding regulatory elements on human brain development and the evolution of cognition, and highlights novel mechanisms underlying neuropsychiatric disorders.
Abstract
Magnetic anisotropy energy (MAE) is one of the most important properties in two-dimensional magnetism since the magnetization in two dimension is vulnerable to the spin rotational ...fluctuations. Using density functional theory calculation, we show that perpendicular electric field dramatically enhances the in-plane and out-of-plane magnetic anisotropies in Fe
3
GeTe
2
and Fe
4
GeTe
2
monolayers, respectively, allowing the change of easy axis in both systems. The changes of the MAE under the electric field are understood as the result of charge redistribution inside the layer, which is available due to the three-dimensional (3D) network of Fe atoms in the monolayers. As a result, we suggest that due to the unique structure of Fe
n
GeTe
2
compounds composed by peculiar 3D networks of metal atoms, the MAE can be dramatically changed by the external perpendicular electric field.
Two-way interaction-mediated modification of isotropic turbulence and bubble dispersion is investigated using direct numerical simulations. The asymmetric coupling force on vortical structures ...generates horizontal force gradients, transiently enhancing the flow vorticity, while the cumulative buoyant transfer induced by bubbles in the downflow regions decreases it by attenuating the horizontal gradients of the downward flow velocity. These vorticity fluctuations affect the non-uniform distortion in the flow dissipation spectrum with a large-scale energy reduction, decreasing the flow dissipation. In addition, the buoyancy force acting on the bubbles affects the bubble distribution. Smaller inhomogeneities in the bubble distribution with turbulence attenuation, owing to the two-way coupling effects, also contribute to the bubble dispersion.
We propose a CFD algorithm for real-time prediction of urban flow and dispersion based on large-eddy simulation (LES). To efficiently handle complex urban building geometry, we implement a modified ...immersed boundary method (IBM), which can be applied to bluff building boundary on a staggered grid. For an introduction of proper inflow condition, we apply a synthetic-eddy method to the periodic domain, which is necessary for direct solving of Poisson equation for pressure. All these implementations are conducted on GPU system for a significant reduction of calculation time for real-time prediction. For validation of our algorithm, we test our model in the prediction of flow and dispersion in urban area in Seoul against wind-tunnel experiment result and other simulation using fire dynamics simulator (FDS). Our model simulation results of flow and dispersion show good agreement with them. Simulation time is reasonably short to warrant real-time prediction of flow and dispersion.