From bacteria to vertebrate, collective motion in living groups in space has gradually become a research hot point. This paper reviews recent studies on collective motion according to the logic from phenomenology to dynamic mechanics. For the phenomenology study, we focus on “order parameter”, “collective symmetry breaking”, “group size distribution” and “spatial correlation” most of which are based on quantitative observations and experiments. Some of these rules are universal in the sense that they can be observed in various species. The “attraction and repulsion”, “alignment rule”, “interaction range”, “heterogeneity” and “linear superposition of interactions”, especially the inferred rules based on the traced tracks of various individual movements, are closely investigated in dynamics part of this paper. This review pays much attention to distinguishing observations and the experiment results from assumptions. Under this logic frame, not only the plentiful and substantial results in collective motion are reviewed, but also an explicit and clear picture is depicted on how many miles we have covered on pursuing the universal laws in collective motion and on what the most important problems are ahead of us.

This paper chooses network traffic as the vulnerability metric for global undersea cable infrastructure and proposes a vulnerability analysis method based on bi-level optimization theory which can identify the most vulnerable parts of the system under specific damage budget and take quantitative analysis for the rerouted traffic because of the damaged parts. Finally, the vulnerability analysis method is verified with two undersea cable systems in a certain hypothesis region, and characteristics of the dynamic vulnerability and corresponding protection strategy are discussed.

Based on what the previous studies did before, the researchers carry out human experiments to analyze the process of evacuation and the optimization of obstacles settings, thus providing theoretical support to improve the efficiency of evacuation in a panic situation. The experimental videos provides us peculiar phenomena during the evacuation, for instance, three individuals crowding together can block the exit and the evacuation process seems to be periodically fluctuated. By comparing experiments differentiated by the obstacle settings, the paper concludes that a reasonable placement of obstacles can improve the efficiency of evacuation, and that setting two obstacles is better than one. Furthermore, crowding is partly alleviated by the placement of obstacles, so that the evacuation process could become more stable, more orderly, thereby the efficiency of the evacuation is better.

For the research of the impacts of travellers’ adaptive behavior on traffic flow assignment under bounded rationality, the multi-agent model of route choice with interaction among travellers is established using cumulative prospect theory and cellular automaton, in which travellers are grouped into two types: risk lovers and risk averse. Travellers’ heterogeneous dynamic reference points and evolution rules are designed based on travel time reliability and the idea of cellular genetic algorithm, so travellers can dynamically adjust their budget of travel time according to environment. The new model is more in tune with travellers’ actual behavior. Then by combining multi-passengers’ evolution rule with method of successive average, the new traffic flow assignment is solved. The study found that (1) the new model inherited the characteristics of the traditional traffic flow assignment model; (2) proportions of travellers with different risk attitude and travellers’ information receiving degree are critical factors which affecting traffic flow assignment.

Under the development of the Internet finance, supernetwork theory was included to study the increasingly complex financial network. A framework was developed for modeling, analysis and solutions to multitiered Internet financial network problems with investors, Internet financial intermediaries, traditional intermediaries and investees. The competitive and cooperative relationships between various decision-makers were explicated, as well as their behavior and optimality conditions. Functions of credit risk and operational risk were included, considering the increasing credit risk and operational risk in the environment of Internet. Using the variational inequality theory, the existence and uniqueness of the equilibrium solution is studied.The governing equilibrium conditions of the Internet financial network is also given so that the trend and saturation condition of Internet finance can be studied later. Finally, a simulation example is given to verify the validity of the supernetwork model.

The paper proposes a weighted network which regards the potential quality failure factor as a node and the relationships between the potential quality failure factors as edges. We study the topology structure of the weighted network. The results show that the network is scale-free. According to vertex strength, cluster coefficient, betweenness and eigenvector, we find the important quality control points. We combine complex network theory with the qualitative quality tools such as fishbone diagram and FMEA. We can effectively find the key quality failure factors and make up the inadequacy of FMEA so that the key point of quality control is easily caught and the quality prevention and improvement is implemented. We provide a new perspective to study quality prevention and improvement, and improve the efficiency of zero defect quality management system and the effect of total quality management.

This paper proposes a new dynamic social network model to study the competition between two languages (language A and B), this model is based on agent modeling method and social circles theory, and this method also introduces a language structure with two F-length bit-strings vocabulary into the agent which is a node of the network. The evolutions of the average number of words of each language and mastery of the language for agents in the network were analyzed by changing the language states, the structure of social network and the probability of language learning and vertical transmission. The proposed model facilitates the study of the mechanisms of macro behavior which emerges from local interactions and language competition between agents and the evolution of internal language structure of speakers. The simulation result indicates that the model introduced here does account for two key factors, they are the chance and frequency of language usage which influence language competition and do not consider well in other models.

Considering many real networks have the community structure property, in this paper, by introducing modularity coefficient and under its control, we build a community network model based on random network, which is used to simulate the real social networks. Then we investigate the epidemic spreading behaviors by mean field theory and get the mathematical expression of epidemic threshold, we also verify it by Monte Carlo simulations. It is found that the existing of community structure can change the network degree distribution, namely, the stronger community structure networks have wider degree distribution. And the stronger the community structure is, the smaller the virus spread critical value will be. Moreover, when the infection rate far away from the epidemic threshold, the transmission sizes of networks with different community structure intensity almost the same, that is, the change of modularity coefficient barely affects the epidemic prevalence.

Based on the rumor spreading characteristics of classic infection model (SIR) on real online social networks, this paper divides the OSN users into 6 different kinds of people, including the ignorants, the knowns, the believers, the spreaders, the temporary stiflers and the permanent stiflers. Taking into account the reality that people enhance their believing level when they contact with the same rumor constantly, this paper combines the positive effects of social reinforcement with the theory of complex networks to construct an improved rumor propagation model of online social network in the scale-free network environment considering tunable clustering. Numerical simulation results show that the rumor spreading range and ability will increase in the rumor first-believing probability, the degree of initial spreader nodes, the density of isolated node and the positive effects of social reinforcement; but on the contrary, it will be restrained by the increase of cluster coefficient. The improved rumor propagation model we proposed well fits the rumor spreading characteristics on real online social network and some theory references will be applied to manage and control Internet rumors.

A rumor spreading model for the web forum based on the heterogeneity of web forum user behaviors and SEIR model is proposed in this paper. First, according to the mean-field equations of the model on inhomogeneous networks, the critical threshold of the spreading of rumor is deduced, the result of theoretical analysis shows that the increase in trust mechanism between users can effectively inhibit the rumor spread in the network with limited velocity of propagation; Then the simulation and numerical analysis of the model itself and the influences of trust mechanism to the model is given, which verify the validity of the model and the introduction of trust mechanism can effectively reduce the rumor influence, the velocity of rumor spreading and the rumor size; Finally, combined with the previous conclusions and the high-influence limited trust relationships between web forum users, a high-influence immunization strategy is given. The experimental results show that the strategy able to reach better effect than traditional immunization strategy.

The evolutionary process of social networks at initial period is very important, especially for the quantitative understanding of the formation and the evolution of interpersonal relationships. In this paper, combining with the “Ice Bucket Challenge”, we collect the data of this event from the launch to the sixth day in our country. The nodes stand for the challengers and the edges are the relations of called people in the social networks. By analyzing the rules of the structural characteristics, including the network size, the clustering coefficient, density, network efficiency and connectivity sub-graphs, we find that the clustering coefficient increased from zero to 0.0167 at the beginning and then decreases; the densityof the network declines from 0.1209 over time; the network efficiency reduces by 81.4% at first and then slowly increases; the connected sub-graphs rapidly increases five times and then decreases; the network efficiency and the number of sub-graphs are negatively correlative. Taking into account the specificity of the network,we compare with evolution of other social networks.Thiswork will be helpful for understanding the law of the formation and development of the early social networks.

PSO algorithm is poor in search accuracy and prone to fall into the local extremum when solving complex multimodal function optimization problem. So, we propose an improved quantum behaved particle swarm optimization algorithm. This paper studies the fundamentals and basic procedure of that algorithm, An orthogonal test for parameter selection is designed to select a set of reasonable control parameters. We use a suite of 28 test functions from CEC’13 as test set. NM-QPSO is compared with both of traditional PSO and QPSO by using the Wilcoxon Signed Ranks Test respectively. Tests show that the NM-QPSO algorithm has better performance than the traditional PSO and QPSO algorithms in statistical sense, and it has obvious advantages in the high-dimensional function optimization.

This paper is concerned with distributed containment control of multi-agent system with multiple stationary leaders under fixed directed network topologies. Under the assumption that communication time-delays in all channels are equal, the Laplace transform is used to study the first-order and second-order multi-agent delayed system with continuous-time. Some sufficient conditions are obtained to ensure the containment of the multi-agent system by stability analysis for transfer function to get conditions and applying final value theorem. Finally,computer simulations show the effectiveness of the conclusion.