MASTOC-LLM extends the classic Multi-Agent System Tragedy of the Commons (MASTOC) model by replacing hard-coded behavioral rules with autonomous decision-making powered by large language models (LLMs). Three heterogeneous agents manage herds of cows on a shared grassland commons. Each tick, an agent receives a structured prompt describing current resource levels, its own herd size, peer behavior, and — optionally — a rolling memory of recent rounds and messages from neighboring agents. The LLM returns a stocking decision (add, remove, or hold cows) together with a natural-language rationale and, when communication is enabled, a short message to broadcast to peers.

The model is designed to test whether LLM agents spontaneously develop Ostrom-style common-pool resource governance (mutual monitoring, graduated sanctions, graduated rule revision) or instead fall into identifiable failure modes. Preliminary experiments with Claude Haiku 4.5, GPT-5.4-mini, and DeepSeek R1:32b have revealed four recurring collapse patterns — Cooperative Paralysis, Defection Cascade, Overshoot-Panic, and Hybrid Architecture Failure — whose onset timing is sensitive to memory length, inter-agent communication, and the post-training alignment approach of the underlying model.

MASTOC-LLM is intended as a laboratory for generative agent-based modelling (GABM) methodology: it provides a clean, well-understood commons baseline against which LLM behavioral hypotheses can be systematically tested and compared across models, parameter sweeps, and alignment regimes.

MOSAIC is an agent-based NetLogo model of decentralized mission coordination among heterogeneous robots operating under partial observability, limited energy, spatially variable risk, dynamic communication, and individual and cooperative task requirements. Robots discover tasks locally, exchange task information through temporary communication links, submit capability-, energy-, deadline-, and risk-aware bids, compete for individual contracts, and form temporary coalitions for cooperative tasks.

The model integrates decentralized auctions, greedy capability-based coalition formation, contract release and reassignment, four reward regimes, reputation, adaptive bidding strategies, failure traceability, and mission-, network-, information-, inequality-, and coalition-level metrics. It operates without a centralized mission planner or global combinatorial assignment solver.

Seven paired-seed BehaviorSpace experiments comprising 690 official simulation runs evaluate baseline mission viability, reward regimes, communication structure, capability heterogeneity, cooperative-task demand, reputation and adaptive strategies, and mission-incentive strength. The results indicate that structural coordination capacity—particularly information reach, capability compatibility, and feasible coalition construction—has a stronger effect on mission completion than increasing incentive intensity within the tested architecture and parameter ranges.

…

A stylized scale model to codesign with villagers an agent-based model of bushmeat hunting in the periphery of Korup National Park (Cameroon)

This multi-model (i.e. a model composed of interacting submodels) is a multi-level representation of a collective motion phenomenon. It was designed to study the impact of the mutual influences between individuals and groups in collective motion.

The model explores how two types of information - social (in the form of pheromone trails) and private (in the form of route memories) affect ant colony level foraging in a variable enviroment.

The CHIME ABM explores information distribution networks and agents’ protective decision making in the context of hurricane landfall.

This code simulates the WiFi user tracking system described in: Thron et al., “Design and Simulation of Sensor Networks for Tracking Wifi Users in Outdoor Urban Environments”. Testbenches used to create the figures in the paper are included.

The purpose of this model is to explore the effects of different power structures on a cross-functional team’s prosocial decision making. Are certain power distributions more conducive to the team making prosocial decisions?

In the face of the COVID-19 pandemic, public health authorities around the world have experimented, in a short period of time, with various combinations of interventions at different scales. However, as the pandemic continues to progress, there is a growing need for tools and methodologies to quickly analyze the impact of these interventions and answer concrete questions regarding their effectiveness, range and temporality.

COMOKIT, the COVID-19 modeling kit, is such a tool. It is a computer model that allows intervention strategies to be explored in silico before their possible implementation phase. It can take into account important dimensions of policy actions, such as the heterogeneity of individual responses or the spatial aspect of containment strategies.

In COMOKIT, built using the agent-based modeling and simulation platform GAMA, the profiles, activities and interactions of people, person-to-person and environmental transmissions, individual clinical statuses, public health policies and interventions are explicitly represented and they all serve as a basis for describing the dynamics of the epidemic in a detailed and realistic representation of space.

…

The fight against poverty is an urgent global challenge. Microinsurance is promoted as a valuable instrument for buffering income losses due to health or climate-related risks of low-income households in developing countries. However, apart from direct positive effects they can have unintended side effects when insured households lower their contribution to traditional arrangements where risk is shared through private monetary support.

RiskNetABM is an agent-based model that captures dynamics between income losses, insurance payments and informal risk-sharing. The model explicitly includes decisions about informal transfers. It can be used to assess the impact of insurance products and informal risk-sharing arrangements on the resilience of smallholders. Specifically, it allows to analyze whether and how economic needs (i.e. level of living costs) and characteristics of extreme events (i.e. frequency, intensity and type of shock) influence the ability of insurance and informal risk-sharing to buffer income shocks. Two types of behavior with regard to private monetary transfers are explicitly distinguished: (1) all households provide transfers whenever they can afford it and (2) insured households do not show solidarity with their uninsured peers.

The model is stylized and is not used to analyze a particular case study, but represents conditions from several regions with different risk contexts where informal risk-sharing networks between smallholder farmers are prevalent.

…