Search for: All records

Abstract Forecasting all components in complex systems is an open and challenging task, possibly due to high dimensionality and undesirable predictors. We bridge this gap by proposing a data-driven and model-free framework, namely, feature-and-reconstructed manifold mapping (FRMM), which is a combination of feature embedding and delay embedding. For a high-dimensional dynamical system, FRMM finds its topologically equivalent manifolds with low dimensions from feature embedding and delay embedding and then sets the low-dimensional feature manifold as a generalized predictor to achieve predictions of all components. The substantial potential of FRMM is shown for both representative models and real-world data involving Indian monsoon, electroencephalogram (EEG) signals, foreign exchange market, and traffic speed in Los Angeles Country. FRMM overcomes the curse of dimensionality and finds a generalized predictor, and thus has potential for applications in many other real-world systems. 

Abstract Machine learning influences numerous aspects of modern society, empowers new technologies, from Alphago to ChatGPT, and increasingly materializes in consumer products such as smartphones and self-driving cars. Despite the vital role and broad applications of artificial neural networks, we lack systematic approaches, such as network science, to understand their underlying mechanism. The difficulty is rooted in many possible model configurations, each with different hyper-parameters and weighted architectures determined by noisy data. We bridge the gap by developing a mathematical framework that maps the neural network’s performance to the network characters of the line graph governed by the edge dynamics of stochastic gradient descent differential equations. This framework enables us to derive a neural capacitance metric to universally capture a model’s generalization capability on a downstream task and predict model performance using only early training results. The numerical results on 17 pre-trained ImageNet models across five benchmark datasets and one NAS benchmark indicate that our neural capacitance metric is a powerful indicator for model selection based only on early training results and is more efficient than state-of-the-art methods. 

Abstract It is essential to study the robustness and centrality of interdependent networks for building reliable interdependent systems. Here, we consider a nonlinear load-capacity cascading failure model on interdependent networks, where the initial load distribution is not random, as usually assumed, but determined by the influence of each node in the interdependent network. The node influence is measured by an automated entropy-weighted multi-attribute algorithm that takes into account both different centrality measures of nodes and the interdependence of node pairs, then averaging for not only the node itself but also its nearest neighbors and next-nearest neighbors. The resilience of interdependent networks under such a more practical and accurate setting is thoroughly investigated for various network parameters, as well as how nodes from different layers are coupled and the corresponding coupling strength. The results thereby can help better monitoring interdependent systems. 

Abstract Quantum networks describe communication networks that are based on quantum entanglement. A concurrence percolation theory has been recently developed to determine the required entanglement to enable communication between two distant stations in an arbitrary quantum network. Unfortunately, concurrence percolation has been calculated only for very small networks or large networks without loops. Here, we develop a set of mathematical tools for approximating the concurrence percolation threshold for unprecedented large-scale quantum networks by estimating the path-length distribution, under the assumption that all paths between a given pair of nodes have no overlap. We show that our approximate method agrees closely with analytical results from concurrence percolation theory. The numerical results we present include 2D square lattices of 200²nodes and complex networks of up to 10⁴nodes. The entanglement percolation threshold of a quantum network is a crucial parameter for constructing a real-world communication network based on entanglement, and our method offers a significant speed-up for the intensive computations involved. 

Abstract Compound failures occur when urban flooding coincides with traffic congestion, and their impact on network connectivity is poorly understood. Firstly, either three-dimensional road networks or the traffic on the roads has been considered, but not both. Secondly, we lack network science frameworks to consider compound failures in infrastructure networks. Here we present a network-theory-based framework that bridges this gap by considering compound structural, functional, and topological failures. We analyze high-resolution traffic data using network percolation theory to study the response of the transportation network in Harris County, Texas, US to Hurricane Harvey in 2017. We find that 2.2% of flood-induced compound failure may lead to a reduction in the size of the largest cluster where network connectivity exists, the giant component, 17.7%. We conclude that indirect effects, such as changes in traffic patterns, must be accounted for when assessing the impacts of flooding on transportation network connectivity and functioning. 

Abstract Many systems may switch to an undesired state due to internal failures or external perturbations, of which critical transitions toward degraded ecosystem states are prominent examples. Resilience restoration focuses on the ability of spatially-extended systems and the required time to recover to their desired states under stochastic environmental conditions. The difficulty is rooted in the lack of mathematical tools to analyze systems with high dimensionality, nonlinearity, and stochastic effects. Here we show that nucleation theory can be employed to advance resilience restoration in spatially-embedded ecological systems. We find that systems may exhibit single-cluster or multi-cluster phases depending on their sizes and noise strengths. We also discover a scaling law governing the restoration time for arbitrary system sizes and noise strengths in two-dimensional systems. This approach is not limited to ecosystems and has applications in various dynamical systems, from biology to infrastructural systems. 

Abstract Despite a number of successful approaches in predicting the spatiotemporal patterns of the novel coronavirus (COVID-19) pandemic and quantifying the effectiveness of non-pharmaceutical interventions starting from data about the initial outbreak location, we lack an intrinsic understanding as outbreak locations shift and evolve. Here, we fill this gap by developing a country distance approach to capture the pandemic’s propagation backbone tree from a complex airline network with multiple and evolving outbreak locations. We apply this approach, which is analogous to the effective resistance in series and parallel circuits, to examine countries’ closeness regarding disease spreading and evaluate the effectiveness of travel restrictions on delaying infections. In particular, we find that 63.2% of travel restrictions implemented as of 1 June 2020 are ineffective. The remaining percentage postponed the disease arrival time by 18.56 days per geographical area and resulted in a total reduction of 13,186,045 infected cases. Our approach enables us to design optimized and coordinated travel restrictions to extend the delay in arrival time and further reduce more infected cases while preserving air travel.