Multi-agent Societies Research Articles

A wealth distribution model with a non-Maxwellian collision kernel

Abstract A non-Maxwellian collision kernel is employed to study the evolution of wealth distribution in a multi-agent society. The collision kernel divides agents into two different groups under certain conditions. Applying the kinetic theory of rarefied gases, we construct a two-group kinetic model for the evolution of wealth distribution. Under the continuous trading limit, the Fokker–Planck equation is derived and its steady-state solution is obtained. For the non-Maxwellian collision kernel, we find a suitable redistribution operator to match the taxation. Our results illustrate that taxation and redistribution have the property to change the Pareto index.

A multi-agent approach to the impact of epidemic spreading on commercial activities

We introduce a class of one-dimensional linear kinetic equations of Boltzmann and Fokker–Planck type, describing the dynamics of the amount of consumer spending transactions of a multi-agent society, that, in the presence of social distancing laws consequent to an epidemic spreading, can reduce the volume of a consistent part of economic commercial activities requiring the presence. At the Boltzmann level, the microscopic variation of the number of transactions around a desired commercial target, is built up by joining a random risk with a deterministic change. Both the risk and target value parameters are here dependent of the number of social contacts of the multi-agent system, which can vary in time in reason of the governments decisions to control the epidemic spreading by introducing restrictions on the social activities. In the asymptotics of grazing interactions, the solution density of the Boltzmann-type kinetic equation is shown to converge towards the solution of a Fokker–Planck-type equation with variable coefficients of diffusion and drift, characterized by the details of the elementary interaction. The economic description is then coupled with the evolution equations of a new SIR-type compartmental epidemic system suitable to describe both the classical epidemic spreading and the social contacts evolution in dependence of the multi-agent social heterogeneity. The (local in time) steady states of the statistical distribution of the amount of economic transactions predicted by the Fokker–Planck equation are shown to be inverse Gamma densities, with time-dependent polynomial tails at infinity which characterize the consequences of the control policies on the time evolution of the Pareto index.

A multi-agent description of the influence of higher education on social stratification

We introduce and discuss a system of one-dimensional kinetic equations describing the influence of higher education in the social stratification of a multi-agent society. The system is obtained by coupling a model for knowledge formation with a kinetic description of the social climbing in which the parameters characterizing the elementary interactions leading to the formation of a social elite are assumed to depend on the degree of knowledge/education of the agents. In addition, we discuss the case in which the education level of an individual is function of the position occupied in the social ranking. With this last assumption, we obtain a fully coupled model in which knowledge and social status influence each other. In the last part, we provide several numerical experiments highlighting the role of education in reducing social inequalities and in promoting social mobility.

Effects of Vaccination Efficacy on Wealth Distribution in Kinetic Epidemic Models.

The spread of the COVID-19 pandemic has highlighted the close link between economics and health in the context of emergency management. A widespread vaccination campaign is considered the main tool to contain the economic consequences. This paper will focus, at the level of wealth distribution modeling, on the economic improvements induced by the vaccination campaign in terms of its effectiveness rate. The economic trend during the pandemic is evaluated, resorting to a mathematical model joining a classical compartmental model including vaccinated individuals with a kinetic model of wealth distribution based on binary wealth exchanges. The interplay between wealth exchanges and the progress of the infectious disease is realized by assuming, on the one hand, that individuals in different compartments act differently in the economic process and, on the other hand, that the epidemic affects risk in economic transactions. Using the mathematical tools of kinetic theory, it is possible to identify the equilibrium states of the system and the formation of inequalities due to the pandemic in the wealth distribution of the population. Numerical experiments highlight the importance of the vaccination campaign and its positive effects in reducing economic inequalities in the multi-agent society.

Social climbing and Amoroso distribution

We introduce a class of one-dimensional linear kinetic equations of Boltzmann and Fokker–Planck type, describing the dynamics of individuals of a multi-agent society questing for high status in the social hierarchy. At the Boltzmann level, the microscopic variation of the status of agents around a universal desired target, is built up introducing as main criterion for the change of status a suitable value function in the spirit of the prospect theory of Kahneman and Twersky. In the asymptotics of grazing interactions, the solution density of the Boltzmann-type kinetic equation is shown to converge towards the solution of a Fokker–Planck type equation with variable coefficients of diffusion and drift, characterized by the mathematical properties of the value function. The steady states of the statistical distribution of the social status predicted by the Fokker–Planck equations belong to the class of Amoroso distributions with Pareto tails, which correspond to the emergence of a social elite. The details of the microscopic kinetic interaction allow to clarify the meaning of the various parameters characterizing the resulting equilibrium. Numerical results then show that the steady state of the underlying kinetic equation is close to Amoroso distribution even in an intermediate regime in which interactions are not grazing.

Non-Maxwellian kinetic equations modeling the dynamics of wealth distribution

We introduce a class of new one-dimensional linear Fokker–Planck-type equations describing the dynamics of the distribution of wealth in a multi-agent society. The equations are obtained, via a standard limiting procedure, by introducing an economically relevant variant to the kinetic model introduced in 2005 by Cordier, Pareschi and Toscani according to previous studies by Bouchaud and Mézard. The steady state of wealth predicted by these new Fokker–Planck equations remains unchanged with respect to the steady state of the original Fokker–Planck equation. However, unlike the original equation, it is proven by a new logarithmic Sobolev inequality with weight and classical entropy methods that the solution converges exponentially fast to equilibrium.

Highlighting the benefits of Industry 4.0 for production: an agent-based simulation approach

Abstract: Advanced Manufacturing or Industry 4.0 concepts bring new advances and challenges to current industrial processes. Such concepts are not always well understood and their results in terms of production performance may not be clear. This work proposes a comparison between a traditional manufacturing process and an advanced manufacturing process, both modelled by a multiagent society. In the traditional manufacturing simulation, the agents follow the defined times of each process, including the maintenance times. In the advanced manufacturing simulation, the decision about when to stop a piece of equipment for maintenance is defined by the agent according to data received from sensors and the definitions of the process. The results indicate a significant improvement in equipment usage and consequently higher production in the same time interval. The process simulation clearly indicates that the application of advanced manufacturing concepts in industry is relevant in order to increase the efficiency of production processes. Among the main concepts introduced in advanced manufacturing models are the Internet of Things (IoT), Cyber-Physical Systems (CPSs), and Artificial Intelligence (AI). The models generated are computationally simulated using an agent-based simulation method from the software AnyLogic. The results obtained should contribute to encouraging small and medium sized enterprises to adopt the concepts of Industry 4.0 in their businesses.

Hydrodynamic Models of Preference Formation in Multi-agent Societies

In this paper, we discuss the passage to hydrodynamic equations for kinetic models of opinion formation. The considered kinetic models feature an opinion density depending on an additional microscopic variable, identified with the personal preference. This variable describes an opinion-driven polarisation process, leading finally to a choice among some possible options, as it happens e.g. in referendums or elections. Like in the kinetic theory of rarefied gases, the derivation of hydrodynamic equations is essentially based on the computation of the local equilibrium distribution of the opinions from the underlying kinetic model. Several numerical examples validate the resulting model, shedding light on the crucial role played by the distinction between opinion and preference formation on the choice processes in multi-agent societies.

Decision Support for Negotiations among Microgrids Using a Multiagent Architecture

This paper presents a decision support model for negotiation portfolio optimization considering the participation of players in local markets (at the microgrid level) and in external markets, namely in regional markets, wholesale negotiations and negotiations of bilateral agreements. A local internal market model for microgrids is defined, and the connection between interconnected microgrids is based on nodal pricing to enable negotiations between nearby microgrids. The market environment considering the local market setting and the interaction between integrated microgrids is modeled using a multi-agent approach. Several multi-agent systems are used to model the electricity market environment, the interaction between small players at a microgrid scale, and to accommodate the decision support features. The integration of the proposed models in this multi-agent society and interaction between these distinct specific multi-agent systems enables modeling the system as a whole and thus testing and validating the impact of the method in the outcomes of the involved players. Results show that considering the several negotiation opportunities as complementary and making use of the most appropriate markets depending on the expected prices at each moment allows players to achieve more profitable results.

Wealth distribution in presence of debts: A Fokker–Planck description

We consider here a Fokker--Planck equation with variable coefficient of diffusion which appears in the modeling of the wealth distribution in a multi-agent society. At difference with previous studies, to describe a society in which agents can have debts, we allow the wealth variable to be negative. It is shown that, even starting with debts, if the initial mean wealth is assumed positive, the solution of the Fokker--Planck equation is such that debts are absorbed in time, and a unique equilibrium density located in the positive part of the real axis will be reached.

On a Fokker-Planck equation for wealth distribution

We study here a Fokker-Planck equation with variable coefficient of diffusion and boundary conditions which appears in the study of the wealth distribution in a multi-agent society [ 2 , 10 , 22 ]. In particular, we analyze the large-time behavior of the solution, by showing that convergence to the steady state can be obtained in various norms at different rates.

Information processing by networks of quantum decision makers

We suggest a model of a multi-agent society of decision makers taking decisions being based on two criteria, one is the utility of the prospects and the other is the attractiveness of the considered prospects. The model is the generalization of quantum decision theory, developed earlier for single decision makers realizing one-step decisions, in two principal aspects. First, several decision makers are considered simultaneously, who interact with each other through information exchange. Second, a multistep procedure is treated, when the agents exchange information many times. Several decision makers exchanging information and forming their judgment, using quantum rules, form a kind of a quantum information network, where collective decisions develop in time as a result of information exchange. In addition to characterizing collective decisions that arise in human societies, such networks can describe dynamical processes occurring in artificial quantum intelligence composed of several parts or in a cluster of quantum computers. The practical usage of the theory is illustrated on the dynamic disjunction effect for which three quantitative predictions are made: (i) the probabilistic behavior of decision makers at the initial stage of the process is described; (ii) the decrease of the difference between the initial prospect probabilities and the related utility factors is proved; (iii) the existence of a common consensus after multiple exchange of information is predicted. The predicted numerical values are in very good agreement with empirical data.

Trust-Based Decision Making in a Self-Adaptive Agent Organization

Interaction between agents is one of the key factors in multiagent societies. Using interaction, agents communicate with each other and cooperatively execute complex tasks that are beyond the capability of a single agent. Cooperatively executing tasks may endanger the success of an agent if it attempts to cooperate with peers that are not proficient or reliable. Therefore, agents need to have an evaluation mechanism to select peers for cooperation. Trust is one of the measures commonly used to evaluate the effectiveness of agents in cooperative societies. Since all interactions are subject to uncertainty, the risk behavior of agents as a contextual factor needs to be taken into account in decision making. In this research, we propose the concept of adaptive risk and agent strategy along with an algorithm that helps agents make decisions in an self-adaptive society utilizing an agent’s own experience and recommendation-based trust. Trust-based decision making increases the profit of the system along with lower task failure in comparison to a no-trust model in which agents do not utilize evaluation mechanisms for choosing their cooperation peers.

A Multi-Agent Emotional Society Whose Melodies Represent its Emergent Social Hierarchy and Are Generated by Agent Communications

In this article a multi-agent system is presented which generates melody pitch sequences with a hierarchical structure. The agents have no explicit melodic intelligence and generate the pitches as a result of artificial emotional influence and communication between agents, and the melody's hierarchical structure is a result of the emerging agent social structure. The system is not a mapping from multi-agent interaction onto musical features, but actually utilizes music for the agents to communicate artificial emotions. Each agent in the society learns its own growing tune during the interaction process. Experiments are presented demonstrating that diverse and non-trivial melodies can be generated, as well as a hierarchical musical structure.

A Proposal for Intelligent Agents with Adjusting Self-Reputation Capability for Preventing Fraud in Multi-Agent Societies *

Intelligent agents are increasingly being used for tasks such as e-commerce auto-bidding. Cooperation and defection challenges commonly found in multi-agent systems make the design and construction of a reliable reputation mechanism an important research and engineering goal. Here we describe our proposal for an intelligent agent model that addresses learning issues in terms of self-reputation. The agents in our proposed model are capable of evaluating their behaviors based on a mix of public and private interest considerations, and of testing various solutions aimed at mitigating self-discrepancy and meeting social standards. Our results indicate promise for integrating a self-reputation mechanism into a self-learning framework in a manner that encourages the early emergence of social cooperation among multiple agents and specific cooperation with individual agents.

Atmosphere Understanding for Humans Robots Interaction Based on SVR and Fuzzy Set

A method for understanding the atmosphere is proposed for humans-robots interactions in a multi-agent society, where the individual assessment of the atmosphere is estimated using a Support Vector Regression (SVR) method that represents the emotions of all agents and the atmosphere of the entire society is represented as a fuzzy set in a Fuzzy Atmosfield. This method provides the necessary information that allows each agent (human/robot) in the society to understand the differences between the objective characteristics of the atmosphere and the agent’s individual assessment of the subjective atmosphere and to make appropriate behavioral decisions thereafter. In the experiments, 13 scenarios are tested by four humans. The characteristics of the atmosphere are calculated by applying the proposed method to the emotion data from the four humans. The results are compared with the subjective atmosphere information from the four humans and it is found that the average accuracy reaches 90%. This proposal is planned in order to realize customized services for the humans-robots interactions in a “Multi-Agent Fuzzy Atmosfield,” which is the subject of the authors’ group’s ongoing research project.

An agent-based simulation for restricting exploitation in electronic societies through social mechanisms

One of the problems in artificial agent societies is the problem of non-cooperation, where individuals have motivations for not cooperating with others. An example of non-cooperation is the issue of freeriding, where some agents do not contribute to the welfare of the society but do consume valuable resources. New mechanisms for group self-organisation and management in multi-agent societies are presented and examined in a multi-agent societies where nodes of a P2P system are modelled as interacting agents belonging to different groups. The context of interaction between agents is the sharing of digital goods in electronic societies. We have simulated a decentralised P2P system which self-organises itself to avoid cooperative sharers being exploited by uncooperative free riders. Specifically, we illustrate how cooperative sharers and uncooperative free riders can be placed in different groups of an electronic society in a decentralised manner. Inspired by human society, we use social mechanisms such as tags, gossip and ostracism. Our aim here is to restrict exploitation or in other words restrict uncooperative behaviour by separating groups based on performance since it reduces the likelihood of bad agents exploiting the good agents in the better groups. The developed system shows promising results by encouraging sharers to move to better groups and also by restricting free riders without any centralised control, which makes these mechanisms appropriate for distributed policy governance. Our work offers new insights into policy mechanisms for regulation of distributed societies.

Social norms for self-policing multi-agent systems and virtual societies

Social norms are one of the mechanisms for decentralized societies to achieve coordi- nation amongst individuals. Such norms are conflict resolution strategies that develop from the population interactions instead of a centralized entity dictating agent protocol. One of the most important characteristics of social norms is that they are imposed by the members of the society, and they are responsible for the fulfillment and defense of these norms. By allowing agents to manage (impose, abide by and defend) social norms, societies achieve a higher degree of freedom by lacking the necessity of author- ities supervising all the interactions amongst agents. In this thesis we approach social norms as a malleable concept, understanding norms as dynamic and dependent on environmental situations and agents' goals. By integrat- ing agents with the necessary mechanisms to handle this concept of norm, we have obtained an agent architecture able to self-police its behavior according to the social and environmental circumstances in which is located. First of all, we have grounded the difference between conventions and essential norms from a game-theoretical perspective. This difference is essential as they favor coordination in games with different characteristics. With respect to conventions, we have analyzed the search space of the emergence of conventions when approached with social learning. The exploration took us to dis- cover the existence of Self-Reinforcing Structures that delay the emergence of global conventions. In order to dissolve and accelerate the emergence of conventions, we have designed socially inspired mechanisms (rewiring and observation) available to agents to use them by accessing local information. The usage of these social instruments rep- resent a robust solution to the problem of convention emergence, specially in complex networks (like the scale-free). On the other hand, for essential norms, we have focused on the "Emergence of Cooperation" problem, as it contains the characteristics of any essential norm scenario. In these type of games, there is a conflict between the self-interest of the individual and the group's interest, fixing the social norm a cooperative strategy. In this thesis we study different decentralized mechanisms by which cooperation emerges and it is maintained. An initial set of experiments on Distributed Punishment lead us to discover that certain types of punishment have a stronger effect on the decision making than a pure benefit-costs calculation. Based on this result, we hypothesize that punishment (utility detriment) has a lower effect on the cooperation rates of the population with respect to sanction (utility detriment and normative elicitation). This hypothesis has been inte- grated into the developed agent architecture (EMIL-I-A). We validate the hypothesis by performing experiments with human subjects, and observing that behaves accordingly to human subjects in similar scenarios (that represent the emergence of cooperation). We have exploited this architecture proving its efficiency in different in-silico scenar- ios, varying a number of important parameters which are unfeasible to reproduce in laboratory experiments with human subjects (because of economic and time resources). Finally, we have implemented an Internalization module, which allows agents to reduce their computation costs by linking compliance with their own goals. Inter- nalization is the process by which an agent abides by the norms without taking into consideration the punishments/sanctions associated to defection. We have shown with agent based simulation how the Internalization has been implemented into the EMIL- I-A architecture, obtaining an efficient performance of our agents without sacrificing adaptation skills.

Towards a Calendar Agent Society with Intelligent Agents in ASP-Updates

We present the analysis and some preliminary specifications to describe a multi-agent society to represent dynamic appointments under in Answer-Sets Programming, by means of knowledge-base updates. This is a case study to represent a group of intelligent BDI agents with a common goal of scheduling a meeting, and we use an example to show how to distribute and change their specifications under unforeseen circumstances, as well as a simple protocol to realise a consensual meeting. We claim that the proposed framework is appropriate to have the benefits from a strong foundation like Answer-Sets Programming, simplicity for its declarative logic programming, as well as practicality for existent implemented solvers, which can be used to implement a more-complete and useful system of agent societies.

Evaluation of Intelligent Agent Frameworks for Human Learning

<p align="left"> </p><p><strong><span style="font-size: xx-small;">Pedagogical Agents are intelligent agents supporting learning in virtual learning environments, VLE. The use of the multi-agent society model inhabited with intelligent virtual agents has shown to provide several benefits to learning. This paper reviews intelligent agents for learning and shows their educational value while demonstrating the new learning possibilities supported by them. Towards the objective of efficiently utilizing the agents in a distributed learning platform, the paper provides an evaluation of intelligent agent development frameworks. This evaluation will provide valuable information to those employing and integrating intelligent agents for different types of VLE with a view towards creating new learning scenarios </span></strong></p>