Synthesizing Brain-network-inspired Interconnections for Large-scale Network-on-chips

Brain network is a large-scale complex network with scale-free, small-world, and modularity properties, which largely supports this high-efficiency massive system. In this article, we propose to synthesize brain-network-inspired interconnections for large-scale network-on-chips. First, we propose a method to generate brain-network-inspired topologies with limited scale-free and power-law small-world properties, which have a low total link length and extremely low average hop count approximately proportional to the logarithm of the network size. In addition, given the large-scale applications, considering the modularity of the brain-network-inspired topologies, we present an application mapping method, including task mapping and deterministic deadlock-free routing, to minimize the power consumption and hop count. Finally, a cycle-accurate simulator BookSim2 is used to validate the architecture performance with different synthetic traffic patterns and large-scale test cases, including real-world communication networks for the graph processing application. Experiments show that, compared with other topologies and methods, the brain-network-inspired network-on-chips (NoCs) generated by the proposed method present significantly lower average hop count and lower average latency. Especially in graph processing applications with a power-law and tightly coupled inter-core communication, the brain-network-inspired NoC has up to 70% lower average hop count and 75% lower average latency than mesh-based NoCs.

Modeling Complex Networks Based on Deep Reinforcement Learning

The network topology of complex networks evolves dynamically with time. How to model the internal mechanism driving the dynamic change of network structure is the key problem in the field of complex networks. The models represented by WS, NW, BA usually assume that the evolution of network structure is driven by nodes’ passive behaviors based on some restrictive rules. However, in fact, network nodes are intelligent individuals, which actively update their relations based on experience and environment. To overcome this limitation, we attempt to construct a network model based on deep reinforcement learning, named as NMDRL. In the new model, each node in complex networks is regarded as an intelligent agent, which reacts with the agents around it for refreshing its relationships at every moment. Extensive experiments show that our model not only can generate networks owing the properties of scale-free and small-world, but also reveal how community structures emerge and evolve. The proposed NMDRL model is helpful to study propagation, game, and cooperation behaviors in networks.

Kernel Regression Coefficients for Practical Significance

Quantitative researchers often use Student’s t-test (and its p-values) to claim that a particular regressor is important (statistically significantly) for explaining the variation in a response variable. A study is subject to the p-hacking problem when its author relies too much on formal statistical significance while ignoring the size of what is at stake. We suggest reporting estimates using nonlinear kernel regressions and the standardization of all variables to avoid p-hacking. We are filling an essential gap in the literature because p-hacking-related papers do not even mention kernel regressions or standardization. Although our methods have general applicability in all sciences, our illustrations refer to risk management for a cross-section of firms and financial management in macroeconomic time series. We estimate nonlinear, nonparametric kernel regressions for both examples to illustrate the computation of scale-free generalized partial correlation coefficients (GPCCs). We suggest supplementing the usual p-values by “practical significance” revealed by scale-free GPCCs. We show that GPCCs also yield new pseudo regression coefficients to measure each regressor’s relative (nonlinear) contribution in a kernel regression.

Complex Contact Network of Patients at the Beginning of an Epidemic Outbreak: An Analysis Based on 1218 COVID-19 Cases in China

The spread of viruses essentially occurs through the interaction and contact between people, which is closely related to the network of interpersonal relationships. Based on the epidemiological investigations of 1218 COVID-19 cases in eight areas of China, we use text analysis, social network analysis and visualization methods to construct a dynamic contact network of the epidemic. We analyze the corresponding demographic characteristics, network indicators, and structural characteristics of this network. We found that more than 65% of cases are likely to be infected by a strong relationship, and nearly 40% of cases have family members infected at the same time. The overall connectivity of the contact network is low, but there are still some clustered infections. In terms of the degree distribution, most cases’ degrees are concentrated between 0 and 2, which is relatively low, and only a few ones have a higher degree value. The degree distribution also conforms to the power law distribution, indicating the network is a scale-free network. There are 17 cases with a degree greater than 10, and these cluster infections are usually caused by local transmission. The first implication of this research is we find that the COVID-19 spread is closely related to social structures by applying computational sociological methods for infectious disease studies; the second implication is to confirm that text analysis can quickly visualize the spread trajectory at the beginning of an epidemic.

Three-dimensional convective dissolution of carbon dioxide into brine

The sequestration of carbon dioxide (CO2) through storage into deep saline aquifers represents an indispensable support technology to achieve the zero-carbon target necessary to mitigate the impact of CO2 on climate change. The effectiveness of the sequestration process, partly driven by the convective dissolution of CO2 in brine, is nowadays well characterized for two-dimensional geometries, low permeabilities, and small pressures of injection of CO2. However, reliable predictions of process-efficiency are missing because of the lack of full understanding of the three-dimensional (3D) spatio-temporal behaviour of CO2-rich convective fingers in brine over a large range of injection pressures. Here, we show that the convective dissolution is determined by the instability of the boundary layer formed at the interface between the two phases and is totally independent of the overall vertical size. Experiments were conducted over a broad range of injection pressures, close to process-relevant conditions. The results show the formation of complex 3D structures, including interconnecting stream tubes at the CO2-liquid interface, which could not be detected in previous 2D Hele-Shaw studies, and fingerings. A scale-free theoretical modelling of the convective process allows us to remap our laboratory results to length-scales of relevance for geological reservoirs. The experiments and the model show that the times needed for the onset of convection and the convective flux are independent of the system size.

Analysis of epidemic propagation using mean field theory on signed graphs

Over the past few decades, the study of epidemic propagation has caught widespread attention from many areas. The field of graphs contains a wide body of research, yet only a few studies explore epidemic propagation’s dynamics in “signed” networks. Motivated by this problem, in this paper we propose a new epidemic propagation model for signed networks, denoted as S-SIS. To explain our analysis, we utilized the mean field theory to demonstrate the theoretical results. When we compare epidemic propagation through negative links to those only having positive links, we find that a higher proportion of infected nodes actually spreads at a relatively small infection rate. It is also found that when the infection rate is higher than a certain value, the overall spreading in a signed network begins showing signs of suppression. Finally, in order to verify our findings, we apply the S-SIS model on Erdös–Rényi random network and scale-free network, and the simulation results is well consist with the theoretical analysis.

The Effect of Information-Driven Resource Allocation on the Propagation of Epidemic with Incubation Period

In the pandemic of COVID-19, there are exposed individuals who are infected but lack distinct clinical symptoms. In addition, the diffusion of related information drives aware individuals to spontaneously seek resources for protection. The special spreading characteristic and coevolution of different processes may induce unexpected spreading phenomena. Thus we construct a three-layered network framework to explore how information-driven resource allocation affects SEIS (Susceptible-Exposed-Infected-Susceptible) epidemic spreading. The analyses utilizing microscopic Markov chain approach reveal that the epidemic threshold depends on the topology structure of epidemic network, and the processes of information diffusion and resource allocation. Conducting extensive Monte Carlo simulations, we find some crucial phenomena in the coevolution of information diffusion, resource allocation and epidemic spreading. Firstly, when E-state (exposed state, without symptoms) individuals are infectious, long incubation period results in more E-state individuals than I-state (infected state, with obvious symptoms) individuals. Besides, when E-state individuals have strong or weak infectious capacity, increasing incubation period have an opposite effect on epidemic propagation. Secondly, the short incubation period induces the first-order phase transition. But enhancing the efficacy of resources would convert the phase transition to a second-order type. Finally, comparing the coevolution in networks with different topologies, we find setting the epidemic layer as scale-free network can inhibit the spreading of the epidemic.