This paper studies the regional dependence of China’s stock during the stock market crash before, during and after 2015 by calculating the mutual information of regional stock indexes among 31 regions, and analyses its dynamic evolution by using the method of rolling window. It is found that during the stock market crash, the dependence of stock market has increased sharply and it reaches the maximum value. It is also found that geographical clustering is not obvious among regions in China, and for Shandong, Jiangsu and Zhejiang, there are the most extensive links with other regions. For Guangdong, Shanghai and Beijing, there are strong links among these three regions, but weak links with other regions.

In order to explore the systemic risk of China's commercial banks, this paper considers the impact of positive and negative market news on the bank's network structure and the complexity of systemic risk in financial institutions, based on the perspective of information spillovers. Firstly, it selected the daily rate of return of listed 14 commercial banks in China. Then, the data is divided into three stages according to major financial events “money shortage” and “stock disaster”. Secondly, it used the complex network method to construct the shock network and the volatility overflow network based on the BEKK-GARCH model, and it explored the wave spillover effect and linkage effect of the bank's wave network by analyzing the network's indicators. Finally, it selected the volatility overflow network as an example, and used the target network and the random immune strategy to do a robustness test . The research result shows that: 1) at different stages, bank volatility spillover networks have different network structures, and the impact of risks can make banks more closely; 2)when the risk spillover network in a high-risk zone system, the network agglomeration coefficient is increasing, and the average path of the network is shortening significantly, this feature indicates that banks in the spillover network will be closely linked to resist risks; 3)the impact of target immunity on network stability is much greater than that of random immunity.

This paper considers the inconsistency between the upgoing and downgoing bus lines. Taking the bus network in Guilin as an example, use the network analysis results to provide a theoretical basis for the optimization of the bus lines. Firstly, the characteristics of the degree distribution, average path length, etc. of the bus transfer network in the urban area of Guilin are studied. The results show that the cumulative probability of the network degree value is in the form of a logarithmic function. The Guilin station has the highest degree, betweenness and compactness, which indicate that the Guilin station is the core station. Secondly, both random and deliberate attack methods are used to destroy the network. The changes of average shortest path and connectivity under random attack are slighter than the deliberate attack, which means that the network is more robust against random attack. Finally, the PageRank algorithm is used to rank the importance of network nodes, and key stations in the bus transfer network are mined.

In order to ensure the safety of aviation system operations and improve the performance of aviation network system under failure, an aviation multi-layer network model was established, which used resilience fragility and resilience recovery to define resilience, using structural entropy, network structure resilience recovery and network traffic resilience recovery to measure the resilience of aviation multi-layer network under failure conditions. Taking North China region of civil aviation as an example, an empirical analysis is carried out to study the degree of recovery of network structure and the change of network flight flow loss under different recovery ratios of nodes. The results show that the aviation multi-layer network can better describe the coupling relationship between aviation systems, the network resilience is mainly affected by the airport network and the control sector network; the airport is the largest resilience recovery node; the greater the recovery rate of the network node, the better the performance of the network structure, the less the flight flow loss, and the better the network resilience.

To further study the details of the topological characteristics of Chinese airline network, this work explored the new topological features of China Aviation Multi-layer Network (CAMN) based on complex network theory. We defined each airline as a layer in network and established a multi-layer network by simulating the evolution of network characteristic parameters in the layer-by-layer process. The results indicate that overlapping degree value of CAMN obeys a power-law distribution and five airports with highest overlapping value are uniformly distributed among airlines. Meanwhile, CAMN presents the characteristics of scale-free networks in aggregation, but characteristics of the small world network can only be obviously observed in aggregation of multi-layer networks. Further, the cooperation of large-scale airlines makes network more efficient. The phenomenon of small world in Chinese aviation aggregation network is mainly caused by the corresponding layer of large and medium-sized airlines. The transportation efficiency of large and medium-sized airline networks is higher than that of low-cost airlines, but the homogeneity is poorer.

The evolutionary congestion game with attackers and feasible state constrained set is investigated, using the semi-tensor product method of matrix, game dynamic system is transformed into an algebraic form and studied its robustness. Firstly, evolutionary congestion game with attackers and controllers is transformed into an algebraic form. Secondly, the open-loop control and state feedback control are transformed, and the Nash equilibrium is robust for any initial profiles in the restricted set of feasible states. Finally, an example is presented to illustrate that the dynamic system of evolutionary congestion game with attackers can achieve robust reachable equilibrium under open-loop control and state feedback control.

This paper uses the AR (1)-GJR (1, 1)-SKT model to describe the marginal distribution of 15 stock index returns. A hybrid R-Vine Copula model is constructed to analyze the risk contagion effect of international stock markets under the four crisis events, which include the subprime mortgage crisis, the European debt crisis, the abnormal fluctuations of the Chinese stock market in 2015 and the Sino-US trade friction in 2018. The empirical results show that the international stock markets maintain symmetrical top-to-bottom dependence characteristics in the long term. The risk contagion will cause the Kendall rank correlation coefficient and tail correlation coefficient among stock markets to rise suddenly. The subprime crisis has a strong contagious effect and a long duration, and the European debt crisis is relatively mild. In 2015, the abnormal fluctuations in the Chinese stock market had a strong contagious effect on international stock markets, but the duration was short. In 2018, the Sino-US trade friction held a weaker contagious effect on the international stock markets. China's Shanghai and Shenzhen stock markets are more integrated with Hong Kong stock market in China.

In order to investigate how the agricultural product price influences the other product prices, we establish an input-output analysis model to conduct a policy simulation of agricultural product price by using the data from the National Bureau of Statistics. The results show that raising the price of agricultural product has a certain impact on minority industrial sectors which have strong relevancy with agriculture, and it has little impact on majority industrial sectors. Compared with the previous years, the differences between every two years are very small. The government should allow the price of the agricultural product to fluctuate in an acceptable range for the most people.

In order to explore the internal mechanism and law of the growth of high-tech industry, from the perspective of industry- university-research, the system dynamics model of the growth of high-tech industry is constructed and simulated, and the key factors affecting the growth of high-tech industry and their influence degree are analyzed. The results show that the R&D investment, talent policy support and technology intermediary service ability of enterprises have a significant positive impact on the sales revenue of new products, and the talent policy support and R&D investment of enterprises have a significant positive impact on the number of patents. In a comprehensive view, R&D investment and technology talent investment are the main factors affecting the growth of high-tech industry.

In order to provide a decision-making reference for physician over-treatment problems, the study established an evolutionary game model based on the game relationship between physician and patient. The result showed that there is no evolutionary stable strategy under the patient's static search knowledge strategy and there is a unique evolutionary stable strategy when the patient dynamically searches for knowledge. As the main conclusions, improving the effectiveness of knowledge in the unit, and increasing the reputational loss coefficient for physician's over-treatment behavior are effective manners to reduce the cost of knowledge search for a unit. The behavior of the physician group which has a lower ability and adopts a smaller degree of over-treatment should be paid more attentions. The focus of supervision should be on diseases with less room for growth of medical expenses.