Dynamic peer-to-peer competition

The dynamic behavior of a multiagent system in which the agent size s_i is variable it is studied along a Lotka-Volterra approach. The agent size has hereby the meaning of the fraction of a given market that an agent is able to capture (market share). A Lotka-Volterra system of equations for prey-predator problems is considered, the competition factor being related to the difference in size between the agents in a one-on-one competition. This mechanism introduces a natural self-organized dynamic competition among agents. In the competition factor, a parameter σ is introduced for scaling the intensity of agent size similarity, which varies in each iteration cycle. The fixed points of this system are analytically found and their stability analyzed for small systems (with n=5 agents). We have found that different scenarios are possible, from chaotic to non-chaotic motion with cluster formation as function of the σ parameter and depending on the initial conditions imposed to the system. The present contribution aim is to show how a realistic though minimalist nonlinear dynamics model can be used to describe the market competition (companies, brokers, decision makers) among other opinion maker communities.     

Epistemic Communities under Active Inference

The spread of ideas is a fundamental concern of today’s news ecology. Understanding the dynamics of the spread of information and its co-option by interested parties is of critical importance. Research on this topic has shown that individuals tend to cluster in echo-chambers and are driven by confirmation bias. In this paper, we leverage the active inference framework to provide an in silico model of confirmation bias and its effect on echo-chamber formation. We build a model based on active inference, where agents tend to sample information in order to justify their own view of reality, which eventually leads to them to have a high degree of certainty about their own beliefs. We show that, once agents have reached a certain level of certainty about their beliefs, it becomes very difficult to get them to change their views. This system of self-confirming beliefs is upheld and reinforced by the evolving relationship between an agent’s beliefs and observations, which over time will continue to provide evidence for their ingrained ideas about the world. The epistemic communities that are consolidated by these shared beliefs, in turn, tend to produce perceptions of reality that reinforce those shared beliefs. We provide an active inference account of this community formation mechanism. We postulate that agents are driven by the epistemic value that they obtain from sampling or observing the behaviours of other agents. Inspired by digital social networks like Twitter, we build a generative model in which agents generate observable social claims or posts (e.g., ‘tweets’) while reading the socially observable claims of other agents that lend support to one of two mutually exclusive abstract topics. Agents can choose which other agent they pay attention to at each timestep, and crucially who they attend to and what they choose to read influences their beliefs about the world. Agents also assess their local network’s perspective, influencing which kinds of posts they expect to see other agents making. The model was built and simulated using the freely available Python package pymdp. The proposed active inference model can reproduce the formation of echo-chambers over social networks, and gives us insight into the cognitive processes that lead to this phenomenon.     

Mechanistic origin of dragon-kings in a population of competing agents

We analyze the mechanistic origins of the extreme behaviors that arise in an idealized model of a population of competing agents, such as traders in a market. These extreme behaviors exhibit the defining characteristics of ‘dragon-kings’. Our model comprises heterogeneous agents who repeatedly compete for some limited resource, making binary choices based on the strategies that they have in their possession. It generalizes the well-known Minority Game by allowing agents whose strategies have not made accurate recent predictions, to step out of the competition until their strategies improve. This generates a complex dynamical interplay between the number V of active agents (mimicking market volume) and the imbalance D between the decisions made (mimicking excess demand). The wide spectrum of extreme behaviors which emerge, helps to explain why no unique relationship has been identified between the price and volume during real market crashes and rallies.     

Cooperative behavior in evolutionary snowdrift game with bounded rationality

An evolutionary snowdrift game (SG) that incorporates bounded rationality and limited information in the evolutionary process is proposed and studied. Based on SG in a well-mixed population and defining the winning action at a turn to be the one that gets a higher payoff, the most recent m winning actions can be used as a public information based on which the competing agents decide their next actions. This defines a strategy pool from which each agent picks a number of strategies as their tool in adapting to the competing environment. The payoff parameter r in SG serves to set the maximum number of winners per turn. Due to the bounded rationality and limited information, the cooperative frequency shows steps and plateaux as a function of r and these features tend to be smoothed out as m increases. These features are results of an interplay between a restricted subset of m-bit histories that the system can visit at a value of r and the limited capacity that agents can adapt. The standard deviation in the number of agents taking the cooperative action is also studied. For general values of r, our model generates a realization of the binary-agent-resource model. The idea of introducing bounded rationality into a two-person game to realize the minority game or binary-agent-resource model could be a useful tool for future research.     

Uncertainty about fundamentals and herding behavior in the FOREX market

It is traditionally assumed in finance models that the fundamental value of assets is known with certainty. Although this is an appealing simplifying assumption it is by no means based on empirical evidence. A simple heterogeneous agent model of the exchange rate is presented. In the model, traders do not observe the true underlying fundamental exchange rate and as a consequence they base their trades on beliefs about this variable. Despite the fact that only fundamentalist traders operate in the market, the model belongs to the heterogeneous agent literature, as traders have different beliefs about the fundamental rate.     

Economic modelling with low-cognition agents

The standard socio-economic model (SSSM) postulates very considerable cognitive powers on the part of its agents. They are able to gather all relevant information in any given situation, and to take the optimal decision on the basis of it, given their tastes and preferences. This behavioural rule is postulated to be universal. The concept of bounded rationality relaxes this somewhat, by permitting agents to have access to only limited amounts of information. But agents still optimise subject to their information set and tastes.Empirical work in economics over the past 20 years or so has shown that in general these behavioural postulates lack empirical validity. Instead, agents appear to have limited ability to gather information, and use simple rules of thumb to process the information which they have in order to take decisions.Building theoretical models on these realistic foundations which give better accounts of empirical phenomena than does the SSSM is an important challenge to both economists and econophysicists. Considerable progress has already been made in a short space of time, and examples are given in this paper.     

Kinetic theory of situated agents applied to pedestrian flow in a corridor

A situated agent-based model for simulation of pedestrian flow in a corridor is presented. In this model, pedestrians choose their paths freely and make decisions based on local criteria for solving collision conflicts. The crowd consists of multiple walking agents equipped with a function of perception as well as a competitive rule-based strategy that enables pedestrians to reach free access areas. Pedestrians in our model are autonomous entities capable of perceiving and making decisions. They apply socially accepted conventions, such as avoidance rules, as well as individual preferences such as the use of specific exit points, or the execution of eventual comfort turns resulting in spontaneous changes of walking speed. Periodic boundary conditions were considered in order to determine the density-average walking speed, and the density-average activity with respect to specific parameters: comfort angle turn and frequency of angle turn of walking agents. The main contribution of this work is an agent-based model where each pedestrian is represented as an autonomous agent. At the same time the pedestrian crowd dynamics is framed by the kinetic theory of biological systems.     

Myopic random walkers and exclusion processes: Single and multispecies

A motility mechanism based on a simple exclusion process, where the probability of movement of an agent depends on the number of unoccupied nearest-neighbor sites is considered. Such interacting agents are termed myopic. This problem is an extension of the famous blind or myopic ant in a labyrinth problem. For the interacting agent models considered here, each agent plays the role of an ant in a labyrinth, where the paths of allowed sites though the labyrinth consist of sites not occupied by other agents. We derive a nonlinear diffusion equation for the average occupancy of the discrete agent-based model for myopic agents. In contrast, interacting blind agents have a constant probability of movement to each of their nearest-neighbor sites, giving rise to a linear diffusion equation. Insight into the various terms in the nonlinear diffusion coefficient is obtained from a study of multiple subpopulations of interacting myopic agents, where an advection–diffusion equation for each subpopulation is derived, and from tracking an individual agent within the crowd, where a motility coefficient is extracted. Averaged discrete simulation data compares very well with the solution to the continuum models. We also compare the behavior of myopic and blind agents. The myopic motility mechanism is biologically motivated to emulate information an individual cell gathers from environment cues. The multispecies model developed and investigated here assists with the interpretation of experimental data involving the tracking subpopulations of cells within a total cell population.     

Network evolution based on minority game with herding behavior

The minority game (MG) is used as a source of information to design complex networks where the nodes represent the playing agents. Differently from classical MG consisting of independent agents, the current model rules that connections between nodes are dynamically inserted or removed from the network according to the most recent game outputs. This way, preferential attachment based on the concept of social distance is controlled by the agents wealth. The time evolution of the network topology, quantitatively measured by usual parameters, is characterized by a transient phase followed by a steady state, where the network properties remain constant. Changes in the local landscapes around individual nodes depend on the parameters used to control network links. If agents are allowed to access the strategies of their network neighbors, a feedback effect on the network structure and game outputs is observed. Such effect, known as herding behavior, considerably changes the dependence of volatility σ on memory size: it is shown that the absolute value of σ as well as the corresponding value of memory size depend both on the network topology and on the way along which the agents make their playing decisions in each game round.     

Measuring Dynamics in Evacuation Behaviour with Deep Learning

Bounded rationality is one crucial component in human behaviours. It plays a key role in the typical collective behaviour of evacuation, in which heterogeneous information can lead to deviations from optimal choices. In this study, we propose a framework of deep learning to extract a key dynamical parameter that drives crowd evacuation behaviour in a cellular automaton (CA) model. On simulation data sets of a replica dynamic CA model, trained deep convolution neural networks (CNNs) can accurately predict dynamics from multiple frames of images. The dynamical parameter could be regarded as a factor describing the optimality of path-choosing decisions in evacuation behaviour. In addition, it should be noted that the performance of this method is robust to incomplete images, in which the information loss caused by cutting images does not hinder the feasibility of the method. Moreover, this framework provides us with a platform to quantitatively measure the optimal strategy in evacuation, and this approach can be extended to other well-designed crowd behaviour experiments.     

Dynamic aggregation evolution of competitive societies of cooperative and noncooperative agents

We propose an evolution model of cooperative agent and noncooperative agent aggregates to investigate the dynamic evolution behaviors of the system and the effects of the competing microscopic reactions on the dynamic evolution. In this model, each cooperative agent and noncooperative agent are endowed with integer values of cooperative spirits and noncooperative spirits, respectively. The cooperative spirits of a cooperative agent aggregate and the noncooperative spirits of a noncooperative agent aggregate change via four competing microscopic reaction schemes: the win-win reaction between two cooperative agents, the lose-lose reaction between two noncooperative agents, the win-lose reaction between a cooperative agent and a noncooperative agent (equivalent to the migration of spirits from cooperative agents to noncooperative agents), and the cooperative agent catalyzed decline of noncooperative spirits. Based on the generalized Smoluchowski's rate equation approach, we investigate the dynamic evolution behaviors such as the total cooperative spirits of all cooperative agents and the total noncooperative spirits of all noncooperative agents. The effects of the three main groups of competition on the dynamic evolution are revealed. These include: (i) the competition between the lose-lose reaction and the win-lose reaction, which gives rise to respectively the decrease and increase in the noncooperative agent spirits; (ii) the competition between the win-win reaction and the win-lose reaction, which gives rise to respectively the increase and decrease in the cooperative agent spirits; (iii) the competition between the win-lose reaction and the catalyzed-decline reaction, which gives rise to respectively the increase and decrease in the noncooperative agent spirits.     

The importance of timescales: simple models for economic markets

This paper considers a simple model of an economy. The economy consists of agents. Each agent produces exactly one good. The good is sold on the market and the agent uses the resulting money to buy many other goods. All agents have the goal to maximize their own utility, which consists of a positive contribution from consumption, and a negative contribution from work. The problem for the agent thus is to balance work and consumption. In contrast to many other economic models, this model prescribes the process in all completeness. The paper looks both at analytical solutions and at simulation results. A particularly important results is that a well-defined market only emerges when prices adapt on a much slower time scale than consumption. This makes clear that a functioning market does not just emerge by itself.     

Social learning with bounded confidence and heterogeneous agents

This paper investigates an opinion formation model in social networks with bounded confidence and heterogeneous agents. The network topologies are shaped by the homophily of beliefs, which means any pair of agents are neighbors only if their belief difference is not larger than a positive constant called the bound of confidence. We consider a model with both informed agents and uninformed agents, the essential difference between which is the informed agents have access to outside signals which are function of the underlying true state of the social event concerned. More precisely, the informed agents update their beliefs by combining the Bayesian posterior beliefs based on their private observations and weighted averages of the beliefs of their neighbors. The uninformed agents update their beliefs simply by linearly combining the beliefs of their neighbors. We find that the whole group can learn the true state only if the bound of confidence is larger than a positive threshold which is related to the population density. Furthermore, simulations show that the proportion of informed agents required for collective learning decreases as the population density increases. By tuning the learning speed of informed agents, we find the following: the higher the speed, the shorter the time needed for the whole group to achieve a steady state, and on the other hand, the higher the speed, the lower the proportion of agents with successful learning — there is a trade-off.     

Entropy,Information, and the Updating of Probabilities

This paper is a review of a particular approach to the method of maximum entropy as a general framework for inference. The discussion emphasizes pragmatic elements in the derivation. An epistemic notion of information is defined in terms of its relation to the Bayesian beliefs of ideally rational agents. The method of updating from a prior to posterior probability distribution is designed through an eliminative induction process. The logarithmic relative entropy is singled out as a unique tool for updating (a) that is of universal applicability, (b) that recognizes the value of prior information, and (c) that recognizes the privileged role played by the notion of independence in science. The resulting framework—the ME method—can handle arbitrary priors and arbitrary constraints. It includes the MaxEnt and Bayes’ rules as special cases and, therefore, unifies entropic and Bayesian methods into a single general inference scheme. The ME method goes beyond the mere selection of a single posterior, and also addresses the question of how much less probable other distributions might be, which provides a direct bridge to the theories of fluctuations and large deviations.     

Birth and death in a continuous opinion dynamics model

We here discuss the process of opinion formation in an open community where agents are made to interact and consequently update their beliefs. New actors (birth) are assumed to replace individuals that abandon the community (deaths). This dynamics is simulated in the framework of a simplified model that accounts for mutual affinity between agents. A rich phenomenology is presented and discussed with reference to the original (closed group) setting. Numerical findings are supported by analytical calculations.     

Asset price dynamics in a financial market with heterogeneous trading strategies and time delays

In this paper we present a continuous time dynamical model of heterogeneous agents interacting in a financial market where transactions are cleared by a market maker. The market is composed of fundamentalist, trend following and contrarian agents who process market information with different time delays. Each class of investors is characterized by path dependent risk aversion. We also allow for the possibility of evolutionary switching between trend following and contrarian strategies. We find that the system shows periodic, quasi-periodic and chaotic dynamics as well as synchronization between technical traders. Furthermore, the model is able to generate time series of returns that exhibit statistical properties similar to those of the S&P 500 index, which is characterized by excess kurtosis, volatility clustering and long memory.     

Self-organized networks of competing boolean agents

A model of Boolean agents competing in a market is presented where each agent bases his action on information obtained from a small group of other agents. The agents play a competitive game that rewards those in the minority. After a long time interval, the poorest player's strategy is changed randomly, and the process is repeated. Eventually the network evolves to a stationary but intermittent state where random mutation of the worst strategy can change the behavior of the entire network, often causing a switch in the dynamics between attractors of vastly different lengths.     

Modeling and simulating human teamwork behaviors using intelligent agents

Among researchers in multi-agent systems there has been growing interest in using intelligent agents to model and simulate human teamwork behaviors. Teamwork modeling is important for training humans in gaining collaborative skills, for supporting humans in making critical decisions by proactively gathering, fusing, and sharing information, and for building coherent teams with both humans and agents working effectively on intelligence-intensive problems. Teamwork modeling is also challenging because the research has spanned diverse disciplines from business management to cognitive science, human discourse, and distributed artificial intelligence. This article presents an extensive, but not exhaustive, list of work in the field, where the taxonomy is organized along two main dimensions: team social structure and social behaviors. Along the dimension of social structure, we consider agent-only teams and mixed human–agent teams. Along the dimension of social behaviors, we consider collaborative behaviors, communicative behaviors, helping behaviors, and the underpinning of effective teamwork—shared mental models. The contribution of this article is that it presents an organizational framework for analyzing a variety of teamwork simulation systems and for further studying simulated teamwork behaviors.     

How do ensembles occupy space?

To find an answer to the title question, an attractiveness function between agents and locations is introduced yielding a phenomenological but generic model for the search for optimal distributions of agents over space. Agents can be seen as, e.g., members of biological populations like colonies of bacteria, swarms, and so on. The global attractiveness between agents and locations is maximized causing (self-propelled) `motion' of agents and, eventually, distinct distributions of agents over space. At the same token spontaneous changes or `decisions' are realized via competitions between agents as well as between locations. Hence, the model's solutions can be considered a sequence of decisions of agents during their search for a proper location. Depending on initial conditions both optimal as well as suboptimal configurations can be reached. For the latter early decision-making are important for avoiding possible conflicts: if the proper moment is missed, then only a few agents can find an optimal solution. Indeed, there is a delicate interplay between the values of the attractiveness function and the constraints as can be expressed by distinct terms of a potential function containing different Lagrange parameters. The model should be viewed as a top-down approach as it describes the dynamics of order parameters, i.e. macroscopic variables that reflect affiliations between agents and locations. The dynamics, however, is modified via so-called cost functions that are interpreted in terms of affinity levels. This interpretation can be seen as an original step towards an understanding of the dynamics at the underlying microscopic level. When focusing on the agent, one may say that the dynamics of an order parameter shows the evolution of an agent's intrinsic `map' for solving the problem of space occupation. Importantly, the dynamics does not necessarily distinguish between evolving (or moving) agents and evolving (or moving) locations though agents are more likely to be actors than the locations. Put differently, an order parameter describes an internal map which is linked to the expectation of an agent to find a certain location. Owing to the dynamical representation, we can therefore follow up the change of these maps over time leading from uncertainty to certainty.     

Learning, competition and cooperation in simple games

The minority model was introduced to study the competition between agents with limited information. It has the remarkable feature that, as the number of strategies available to the agents increases, the collective gain made by the agents is reduced. This crowd effect arises from the fact that only a minority can profit at each moment, while all agents make their choices using the same input. We show that the properties of the model change drastically if the agents make choices based on their individual stories, keeping all remaining rules unaltered. This variation reduces the intrinsic frustration of the model, and improves the tendency towards cooperation and self organization. We finally study the stable mixing of individual and collective behavior. Received 30 June 1999 and Received in final form 27 September 1999