NOMAD is an agent-based model of firm location choice between two aggregate regions (“near” and “off”) under logistics uncertainty. Firms occupy sites characterised by attractiveness and logistics risk, earn a risk-adjusted payoff that depends on regional costs (wages plus congestion) and an individual risk-tolerance trait, and update location choices using aspiration-based satisficing rules with switching frictions. Logistics risk evolves endogenously on occupied sites through a region-specific absorption mechanism (good/bad events that reduce/increase risk), while congestion feeds back into regional costs via regional shares and local crowding. Runs stop endogenously once the near-region share becomes quasi-stable after burn-in, and the model records time series and quasi-stable outcomes such as near/off composition, switching intensity, costs, average risk, and average risk tolerance.

CROwd Simulation of Situated individuals represents a modern generation simulation as a (social) scientific tool for understanding crowd behaviour. The CROSS model represents individuals in a crowd as social-cognitive agents that are affected by their social and physical surroundings and produce behaviour and behaviour patterns.

This is an agent-based model that captures the dynamic processes related to moving from an educational system where the school a student attends is based on assignment to a neighborhood school, to one that gives households more choice among existing and newly formed public schools.

An empirical ABM of smallholder decisions in times of drought stress.

We model the relationship between natural resource user´s individual time preferences and their use of destructive extraction method in the context of small-scale fisheries.

A NetLogo ABM developed to explore unarmed resistance to an active shooter. The landscape is a generalized open outdoor area. Parameters enable the user to set shooter armament and control for assumptions with regard to shooter accuracy.

We extend the Flache-Mäs model to incorporate the location and dyadic communication regime of the agents in the opinion formation process. We make spatially proximate agents more likely to interact with each other in a pairwise communication regime.

The wisdom of the crowd refers to the phenomenon in which a group of individuals, each making independent decisions, can collectively arrive at highly accurate solutions—often more accurate than any individual within the group. This principle relies heavily on independence: if individual opinions are unbiased and uncorrelated, their errors tend to cancel out when averaged, reducing overall bias. However, in real-world social networks, individuals are often influenced by their neighbors, introducing correlations between decisions. Such social influence can amplify biases, disrupting the benefits of independent voting. This trade-off between independence and interdependence has striking parallels to ensemble learning methods in machine learning. Bagging (bootstrap aggregating) improves classification performance by combining independently trained weak learners, reducing bias. Boosting, on the other hand, explicitly introduces sequential dependence among learners, where each learner focuses on correcting the errors of its predecessors. This process can reinforce biases present in the data even if it reduces variance. Here, we introduce a new meta-algorithm, casting, which captures this biological and computational trade-off. Casting forms partially connected groups (“castes”) of weak learners that are internally linked through boosting, while the castes themselves remain independent and are aggregated using bagging. This creates a continuum between full independence (i.e., bagging) and full dependence (i.e., boosting). This method allows for the testing of model capabilities across values of the hyperparameter which controls connectedness. We specifically investigate classification tasks, but the method can be used for regression tasks as well. Ultimately, casting can provide insights for how real systems contend with classification problems.

Decision-makers often have to act before critical times to avoid the collapse of ecosystems using knowledge \textcolor{red}{that can be incomplete or biased}. Adaptive management may help managers tackle such issues. However, because the knowledge infrastructure required for adaptive management may be mobilized in several ways, we study the quality and the quantity of knowledge provided by this knowledge infrastructure. In order to analyze the influence of mobilized knowledge, we study how the following typology of knowledge and its use may impact the safe operating space of exploited ecosystems: 1) knowledge of the past based on a time series distorted by measurement errors; 2) knowledge of the current systems’ dynamics based on the representativeness of the decision-makers’ mental models of the exploited ecosystem; 3) knowledge of future events based on decision-makers’ likelihood estimates of extreme events based on modeling infrastructure (models and experts to interpret them) they have at their disposal. We consider different adaptive management strategies of a general regulated exploited ecosystem model and we characterize the robustness of these strategies to biased knowledge. Our results show that even with significant mobilized knowledge and optimal strategies, imperfect knowledge may still shrink the safe operating space of the system leading to the collapse of the system. However, and perhaps more interestingly, we also show that in some cases imperfect knowledge may unexpectedly increase the safe operating space by suggesting cautious strategies.
The code enables to calculate the safe operating spaces of different managers in the case of biased and unbiased knowledge.

This agent-based model explores the dynamics between human behavior and vaccination strategies during COVID-19 pandemics. It examines how individual risk perceptions influence behaviors and subsequently affect epidemic outcomes in a simulated metropolitan area resembling New York City from December 2020 to May 2021.

Agents modify their daily activities—deciding whether to travel to densely populated urban centers or stay in less crowded neighborhoods—based on their risk perception. This perception is influenced by factors such as risk perception threshold, risk tolerance personality, mortality rate, disease prevalence, and the average number of contacts per agent in crowded settings. Agent characteristics are carefully calibrated to reflect New York City demographics, including age distribution and variations in infection probability and mortality rates across these groups. The agents can experience six distinct health statuses: susceptible, exposed, infectious, recovered from infection, dead, and vaccinated (SEIRDV). The simulation focuses on the Iota and Alpha variants, the dominant strains in New York City during the period.

We simulate six scenarios divided into three main categories:
1. A baseline model without vaccinations where agents exhibit no risk perception and are indifferent to virus transmission and disease prevalence.
…