An ANNs-Based Method for Automated Labelling of Schematic Metro Maps

Schematic maps are popular for representing transport networks. In the last two decades, some researchers have been working toward automated generation of network layouts (i.e., the network geometry of schematic maps), while automated labelling of schematic maps is not well considered. The descriptive-statistics-based labelling method, which models the labelling space by defining various station-based line relations in advance, has been specially developed for schematic maps. However, if a certain station-based line relation is not predefined in the database, this method may not be able to infer suitable labelling positions under this relation. It is noted that artificial neural networks (ANNs) have the ability to infer unseen relations. In this study, we aim to develop an ANNs-based method for the labelling of schematic metro maps. Samples are first extracted from representative schematic metro maps, and then they are employed to train and test ANNs models. Five types of attributes (e.g., station-based line relations) are used as inputs, and two types of attributes (i.e., directions and positions of labels) are used as outputs. Experiments show that this ANNs-based method can generate effective and satisfactory labelling results in the testing cases. Such a method has potential to be extended for the labelling of other transport networks.

Structural mechanism for bi-directional actin crosslinking by T-plastin

To fulfill the cytoskeleton's diverse functions in cell mechanics and motility, actin networks with specialized architectures are built by crosslinking proteins, which bridge filaments to control micron-scale network geometry through nanoscale binding interactions via poorly defined structural mechanisms. Here, we introduce a machine-learning enabled cryo-EM pipeline for visualizing active crosslinkers, which we use to analyze human T-plastin, a member of the evolutionarily ancient plastin/fimbrin family of tandem calponin-homology domain (CHD) proteins. We define a sequential bundling mechanism which enables T-plastin to bridge filaments in both parallel and anti-parallel orientations. Our structural, biochemical, and cell biological data highlight inter-CHD linkers as key structural elements underlying flexible but stable crosslinking which are likely to be disrupted by mutations causing hereditary bone diseases. Beyond revealing how plastins are evolutionary optimized to crosslink dense actin networks with mixed polarity, our cryo-EM workflow will broadly enable analysis of the structural mechanisms underlying cytoskeletal network construction.

The Shortest Path to Network Geometry

Real networks comprise from hundreds to millions of interacting elements and permeate all contexts, from technology to biology to society. All of them display non-trivial connectivity patterns, including the small-world phenomenon, making nodes to be separated by a small number of intermediate links. As a consequence, networks present an apparent lack of metric structure and are difficult to map. Yet, many networks have a hidden geometry that enables meaningful maps in the two-dimensional hyperbolic plane. The discovery of such hidden geometry and the understanding of its role have become fundamental questions in network science giving rise to the field of network geometry. This Element reviews fundamental models and methods for the geometric description of real networks with a focus on applications of real network maps, including decentralized routing protocols, geometric community detection, and the self-similar multiscale unfolding of networks by geometric renormalization.

The Robustness of Interdependent Directed Networks With Intra-layer Angular Correlations

The robustness of interdependent networks is a frontier topic in current network science. A line of studies has so far been investigated in the perspective of correlated structures on robustness, such as degree correlations and geometric correlations in interdependent networks, in-out degree correlations in interdependent directed networks, and so on. Advances in network geometry point that hyperbolic properties are also hidden in directed structures, but few studies link those features to the dynamical process in interdependent directed networks. In this paper, we discuss the impact of intra-layer angular correlations on robustness from the perspective of embedding interdependent directed networks into hyperbolic space. We find that the robustness declines as increasing intra-layer angular correlations under targeted attacks. Interdependent directed networks without intra-layer angular correlations are always robust than those with intra-layer angular correlations. Moreover, empirical networks also support our findings: the significant intra-layer angular correlations are hidden in real interdependent directed networks and contribute to the prediction of robustness. Our work sheds light that the impact of intra-layer angular correlations should be attention, although in-out degree correlations play a positive role in robustness. In particular, it provides an early warning indicator by which the system decoded the intrinsic rules for designing efficient and robust interacting directed networks.

Propagating uplift controls on high-elevation, low-relief landscape formation in the southeast Tibetan Plateau

High-elevation, low-relief surfaces are widespread in many mountain belts. However, the origin of these surfaces has long been debated. In particular, the southeast Tibetan Plateau has extensive low-relief surfaces perched above deep valleys and in the headwaters of three of the world’s largest rivers (Salween, Mekong, and Yangtze Rivers). Various geologic data and geodynamic models show that many mountain belts grow first to a certain height and then laterally in an outward propagation sequence. By translating this information into a kinematic propagating uplift function in a landscape evolution model, we propose that the high-elevation, low-relief surfaces in the southeast Tibetan Plateau are simply a consequence of mountain growth and do not require a special process to form. The propagating uplift forms an elongated river network geometry with broad high-elevation, low-relief headwaters and interfluves that persist for tens of millions of years, consistent with the observed geochronology. We suggest that the low-relief interfluves can be long-lived because they lack the drainage networks necessary to keep pace with the rapid incision of the large main-stem rivers. The propagating uplift also produces spatial and temporal exhumation patterns and river profile morphologies that match observations. Our modeling therefore reconciles geomorphic observations with geodynamic models of uplift of the southeast Tibetan Plateau, and it provides a simple mechanism to explain the low-relief surfaces observed in several mountain belts on Earth.

Clustering, Connectivity and Flow in Naturally Fractured Reservoir Analogs

A previous study by the authors on synthetic fractal-fracture networks showed that lacunarity, a parameter that quantifies scale-dependent clustering in patterns, can be used as a proxy for connectivity and also, is an indicator of fluid flow in such model networks. In this research, we apply the concepts thus developed to the study of fractured reservoir analogs and seek solutions to more practical problems faced by modelers in the oil and gas industry. A set of seven nested fracture networks from the Devonian Sandstone of Hornelen Basin, Norway that have the same fractal-dimension but are mapped at different scales and resolutions is considered. We compare these seven natural fracture maps in terms of their lacunarity and connectivity values to test whether the former is a reasonable indicator of the latter. Additionally, these maps are also flow simulated by implementing a fracture continuum model and using a streamline simulator, TRACE3D. The values of lacunarity, connectivity and fluid recovery thus obtained are pairwise correlated with one another to look for possible relationships. The results indicate that while fracture maps that have the same fractal dimension show almost similar connectivity values, there exist subtle differences such that both the connectivity and clustering values change systematically with the scale at which the fracture networks are mapped. It is further noted that there appears to be a very good correlation between clustering, connectivity, and fluid recovery values for these fracture networks that belong to the same fractal system. The overall results indicate that while the fractal dimension is an important parameter for characterizing a specific type of fracture network geometry, it is the lacunarity or scale-dependent clustering attribute that controls connectivity in fracture maps and hence the flow properties. This research may prove helpful in quickly evaluating connectivity of fracture networks based on the lacunarity parameter. This parameter can therefore, be used for calibrating Discrete Fracture Network (DFN) models with respect to connectivity of reservoir analogs and can possibly replace the fractal dimension which is more commonly used in software that model DFNs. Additionally, while lacunarity has been mostly used for understanding network geometry in terms of clustering, we, for the first time, show how this may be directly used for understanding the potential flow behavior of fracture networks.

Unfolding the multiscale structure of networks with dynamical Ollivier-Ricci curvature

Describing networks geometrically through low-dimensional latent metric spaces has helped design efficient learning algorithms, unveil network symmetries and study dynamical network processes. However, latent space embeddings are limited to specific classes of networks because incompatible metric spaces generally result in information loss. Here, we study arbitrary networks geometrically by defining a dynamic edge curvature measuring the similarity between pairs of dynamical network processes seeded at nearby nodes. We show that the evolution of the curvature distribution exhibits gaps at characteristic timescales indicating bottleneck-edges that limit information spreading. Importantly, curvature gaps are robust to large fluctuations in node degrees, encoding communities until the phase transition of detectability, where spectral and node-clustering methods fail. Using this insight, we derive geometric modularity to find multiscale communities based on deviations from constant network curvature in generative and real-world networks, significantly outperforming most previous methods. Our work suggests using network geometry for studying and controlling the structure of and information spreading on networks.

A simplified framework for prediction of sensor network coverage in real-time structural health monitoring of plate-like structures

In this article, the coverage area prediction of piezoelectric sensor network for detecting a specific type of under-surface crack in plate-like structures is addressed. In particular, this article proposes a simplified framework to estimate the coverage of any given sensor network arrangement when a critical defect is known. Based on numerical results from finite element methods (FEM), a simplified framework to estimate coverage area of any given network arrangement is developed. Using such a simplified framework, one can avoid time-consuming procedure of evaluating numerous FEM models in estimating sensor network coverage. Back-scatter fields of partial cracks are estimated using a proposed function, whose parameters are estimated from the results of a limited number of FEM simulations. The proposed function efficiently predicts the back-scattered field of any combination of transmitters and receivers for a given crack geometry. Superposition is used to estimate the coverage area of an arbitrary piezoelectric (e.g., PZT) sensor network. It is shown that the coverage area of a sensor network depends on both sensor network geometry and defect properties (e.g., crack inclination) and it is not necessarily a linear function of the number of sensors. Furthermore, it is shown that the network arrangement has an important effect on the geometry of the coverage area. Experimental results of a network of 14 PZTs in two clusters confirm the accuracy of the method.