Spatiotemporal modeling of water inrush spreading in mine roadway networks

Water inrush accidents seriously threaten underground mining production, so the accurate prediction of the spreading process of water inrush is essential for the formulation of water-inrush-control plans and rescue schemes. This paper proposes a spatiotemporal model based on pipe-flow theory to simulate the spreading process of water inrush in mine roadway networks. The energy-loss term is added to this model to improve the simulation accuracy in bifurcated roadways, and pumps and water-blocking equipment are considered in controlling the spreading process of water inrush. Through experimental case studies, the simulation results and the function of the energy-loss term are verified. A sensitivity analysis is then carried out to assess the impact of the model parameters. The results show that the model outputs are most sensitive to the roadway length, cross-section width, and energy-loss coefficient. The model exhibited maximal sensitivity to the geometric parameters compared with the hydraulic parameters. Furthermore, the spreading process of a real water inrush in a coal mine in North China is simulated, and the water-inrush-control measures are evaluated. The overall results indicate that the proposed spatiotemporal model accurately predicts the spreading process of water inrush and is thus applicable to large-scale mine roadway networks.

Detail-preserving depth estimation from a single image based on modified fully convolutional residual network and gradient network

Predicting a convincing depth map from a monocular single image is a daunting task in the field of computer vision. In this paper, we propose a novel detail-preserving depth estimation (DPDE) algorithm based on a modified fully convolutional residual network and gradient network. Specifically, we first introduce a new deep network that combines the fully convolutional residual network (FCRN) and a U-shaped architecture to generate the global depth map. Meanwhile, an efficient feature similarity-based loss term is introduced for training this network better. Then, we devise a gradient network to generate the local details of the scene based on gradient information. Finally, an optimization-based fusion scheme is proposed to integrate the depth and depth gradients to generate a reliable depth map with better details. Three benchmark RGBD datasets are evaluated from the perspective of qualitative and quantitative, the experimental results show that the designed depth prediction algorithm is superior to several classic depth prediction approaches and can reconstruct plausible depth maps.

Non-linear Archetypal Analysis of Single-cell RNA-seq Data by Deep Autoencoders

Advances in single-cell RNA sequencing (scRNA-seq) have led to successes in discovering novel cell types and understanding cellular heterogeneity among complex cell populations through cluster analysis. However, cluster analysis is not able to reveal continuous spectrum of states and underlying gene expression programs (GEPs) shared across cell types. We introduce scAAnet, an autoencoder for single-cell non-linear archetypal analysis, to identify GEPs and infer the relative activity of each GEP across cells. We use a count distribution-based loss term to account for the sparsity and overdispersion of the raw count data and add an archetypal constraint to the loss function of scAAnet. We first show that scAAnet outperforms existing methods for archetypal analysis across different metrics through simulations. We then demonstrate the ability of scAAnet to extract biologically meaningful GEPs using publicly available scRNA-seq datasets including a pancreatic islet dataset, a lung idiopathic pulmonary fibrosis dataset and a prefrontal cortex dataset.

Soliton Waves in Lossy Nonlinear Transmission Lines at Microwave Frequencies: Analytical, Numerical and Experimental Studies

In this paper, we performed analytical, numerical and experimental studies on the generation of soliton waves in discrete nonlinear transmission lines (NLTL) with varactors, as well as the analysis of the losses impact on the propagation of these waves. Using the reductive perturbation method, we derived a nonlinear Schrödinger (NLS) equation with a loss term and determined an analytical expression that completely describes the bright soliton profile. Our theoretical analysis predicts the carrier wave frequency threshold above which a formation of bright solitons can be observed. We also performed numerical simulations to confirm our analytical results and we analyzed the space–time evolution of the soliton waves. A good agreement between analytical and numerical findings was obtained. An experimental prototype of the lossy NLTL, built at the discrete level, was used to validate our proposed model. The experimental shape of the envelope solitons is well fitted by the theoretical waveforms, which take into account the amplitude damping due to the losses in commercially available varactors and inductors used in a prototype. Experimentally observed changes in soliton amplitude and half–maximum width during the propagation along lossy NLTL are in good accordance with the proposed model defined by NLS equation with loss term.

Learning Domain Invariant Representation via Self-Rugularization

Unsupervised domain adaptation often gives impressive solutions to handle domain shift of data. Most of current approaches assume that unlabeled target data to train is abundant. This assumption is not always true in practices. To tackle this issue, we propose a general solution to solve the domain gap minimization problem without any target data. Our method consists of two regularization steps. The first step is a pixel regularization by arbitrary style transfer. Recently, some methods bring style transfer algorithms to domain adaptation and domain generalization process. They use style transfer algorithms to remove texture bias in source domain data. We also use style transfer algorithms for removing texture bias, but our method depends on neither domain adaptation nor domain generalization paradigm. The second regularization step is a feature regularization by feature alignment. Adding a feature alignment loss term to the model loss, the model learns domain invariant representation more efficiently. We evaluate our regularization methods from several experiments both on small dataset and large dataset. From the experiments, we show that our model can learn domain invariant representation as much as unsupervised domain adaptation methods.

Particle-phase processing of α-pinene NO₃ secondary organic aerosol in the dark

The NO3 radical represents a significant night-time oxidant present in or downstream of polluted environments. There are studies that investigated the formation of secondary organic aerosol (SOA) from NO3 radicals focusing on yields, general composition, and hydrolysis of organonitrates. However, there is limited knowledge about how the composition of NO3-derived SOA evolves as a result of particle phase reactions. Here, SOA was formed from the reaction of α-pinene with NO3 radicals generated from N2O5, and the resulting SOA aged in the absence of external stimuli. The initial composition of NO3-derived α-pinene SOA was slightly dependent upon the concentration of N2O5 injected (excess of NO3 or excess of α-pinene), but was largely dominated by dimer dinitrates (C20H32N2O8-13). Oxidation reactions (e.g. C20H32N2O8 C20H32N2O9 C20H32N2O10 etc...) accounted for 60–70 % of the particle phase reactions observed. Fragmentation reactions and dimer degradation pathways made up the remainder of the particle-phase processes occurring. The exact oxidant is not known, though suggestions are offered (e.g. N2O5, organic peroxides, or peroxy-nitrates). Hydrolysis of −ONO2 functional groups was not an important loss term during dark aging under the relative humidity conditions of our experiments (58–62 %), and changes in the bulk organonitrate composition were likely driven by evaporation of highly nitrogenated molecules. Overall, 25–30 % of the particle-phase composition changes as a function of particle-phase reactions during dark aging representing an important atmospheric aging pathway.

Deep Hierarchical Embedding for Simultaneous Modeling of GPCR Proteins in a Unified Metric Space

GPCR proteins belong to diverse families of proteins that are defined at multiple hierarchical levels. Inspecting relationships between GPCR proteins on the hierarchical structure is important, since characteristics of the protein can be inferred from proteins in similar hierarchical information. However, modeling of GPCR families has been performed separately at each level. Relationships between GPCR proteins are ignored in these approaches as they process the information in the proteins with several disconnected models. In this study, we propose a deep learning model to simultaneously learn representations of GPCR family hierarchy from the protein sequences with a unified single model. Novel loss term based on metric learning is introduced to incorporate hierarchical relations between proteins. We tested our approach using a public GPCR sequence dataset. Metric distances in the deep feature space corresponded to the hierarchical family relation between GPCR proteins. Furthermore, we demonstrated that further downstream tasks, like phylogenetic reconstruction and motif discovery, are feasible in the constructed embedding space. These results show that hierarchical relations between sequences were successfully captured in both of technical and biological aspects.

Hydrodynamic Cross-Scale Archaeology at a Roman River Harbour

Trans-disciplinary research methods and data from archaeology, geology, hydrology, and hydraulic engineering are successfully merged to reevaluate hydrodynamic effects of Roman hydraulic structures at a Rhine river harbour. The archaeological site Colonia Ulpia Traiana, is characterized by its exceptional preservation, providing ample research data on its river harbour. Constructed by the Romans, the berthing area is lined by a wooden quay-wall. Setting this harbour apart is its up-stream tip, which is fitted with a unique hydraulic structure with unknown purpose. Structure related hydrodynamic impacts on the historic Rhine regime are examined by introducing a novel cross-scale multi model approach, consisting of three steps: (i) Scaled physical experiments are performed to investigate the roughness influence of the wooden quay on a local scale. (ii) A numerical representation of the physical experiments is done in Delft3D, validating a linear loss term to accurately capture the roughness influence on the velocity distribution. (iii) A mid-scale Rhine river model of the area is generated that approximates the historic river bathymetry through morphological evolution. The quay-wall is implemented in parametric form and induces a substantial velocity reduction throughout the harbour. The unique structure exhibits hydromechanic properties mimicking present day current-deflection walls, potentially rendering it their primal prototype.

Suppression of COVID-19 infection by isolation time control based on the SIR model and an analogy from nuclear fusion research

The coronavirus disease 2019 (COVID-19) has been damaging our daily life after declaration of pandemic. Therefore, we have started studying on the characteristics of Susceptible-Infectious-Recovered (SIR) model to know about the truth of infectious disease and our future.After detailed studies on the characteristics of the SIR model for the various parameter dependencies with respect to such as the outing restriction (lockdown) ratio and vaccination rate, we have finally noticed that the second term (isolation term) in the differential equation of the number of the infected is quite similar to the “helium ash particle loss term” in deuterium-tritium (D-T) nuclear fusion. Based on this analogy, we have found that isolation of the infected is not actively controlled in the SIR model. Then we introduce the isolation time control parameter q and have studied its effect on this pandemic. Required isolation time to terminate the COVID-19 can be estimated by this proposed method.To show this isolation control effect, we choose Tokyo for the model calculation because of high population density. We determine the reproduction number and the isolation ratio in the initial uncontrolled phase, and then the future number of the infected is estimated under various conditions. If the confirmed case can be isolated in 3∼8 days by widely performed testing, this pandemic could be suppressed without awaiting vaccination. If the mild outing restriction and vaccination are taken together, the isolation control time can be longer. We consider this isolation time control might be the only solution to overcome the pandemic when vaccine is not available.