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.

The purpose of this model is to explain the post-disaster recovery of households residing in their own single-family homes and to predict households’ recovery decisions from drivers of recovery. Herein, a household’s recovery decision is repair/reconstruction of its damaged house to the pre-disaster condition, waiting without repair/reconstruction, or selling the house (and relocating). Recovery drivers include financial conditions and functionality of the community that is most important to a household. Financial conditions are evaluated by two categories of variables: costs and resources. Costs include repair/reconstruction costs and rent of another property when the primary house is uninhabitable. Resources comprise the money required to cover the costs of repair/reconstruction and to pay the rent (if required). The repair/reconstruction resources include settlement from the National Flood Insurance (NFI), Housing Assistance provided by the Federal Emergency Management Agency (FEMA-HA), disaster loan offered by the Small Business Administration (SBA loan), a share of household liquid assets, and Community Development Block Grant Disaster Recovery (CDBG-DR) fund provided by the Department of Housing and Urban Development (HUD). Further, household income determines the amount of rent that it can afford. Community conditions are assessed for each household based on the restoration of specific anchors. ASNA indexes (Nejat, Moradi, & Ghosh 2019) are used to identify the category of community anchors that is important to a recovery decision of each household. Accordingly, households are indexed into three classes for each of which recovery of infrastructure, neighbors, or community assets matters most. Further, among similar anchors, those anchors are important to a household that are located in its perceived neighborhood area (Moradi, Nejat, Hu, & Ghosh 2020).

The model generates disaggregated traffic flows of pedestrians, simulating their daily mobility behaviour represented as probabilistic rules. Various parameters of physical infrastructure and travel behaviour can be altered and tested. This allows predicting potential shifts in traffic dynamics in a simulated setting. Moreover, assumptions in decision-making processes are general for mid-sized cities and can be applied to similar areas.

Together with the model files, there is the ODD protocol with the detailed description of model’s structure. Check the associated publication for results and evaluation of the model.

Installation
Download GAMA-platform (GAMA1.8.2 with JDK version) from https://gama-platform.github.io/. The platform requires a minimum of 4 GB of RAM.
…

This is a tool to explore the effects of groups´ spatial segregation on the emergence of opinion polarization. It embeds two opinion formation models: a model of negative (and positive) social influence and a model of persuasive argument exchange.

This model represents informal information transmission networks among medieval Genoese investors used to inform each other about cheating merchants they employed as part of long-distance trade operations.

The model demonstrates how non-instantaneous sampling techniques produce bias by overestimating the number of counted animals, when they move relative to the person counting them.

This documentation provides an overview and explanation of the NetLogo simulation code for modeling skilled workers’ migration in Iran. The simulation aims to explore the dynamics of skilled workers’ migration and their transition through various states, including training, employment, and immigration.

The flow of elite and talent migration, or “brain drain,” is a complex issue with far-reaching implications for developing countries. The decision to migrate is made due to various factors including economic opportunities, political stability, social factors and personal circumstances.
Measuring individual interests in the field of immigration is a complex task that requires careful consideration of various factors. The agent-based model is a useful tool for understanding the complex factors that are involved in talent migration. By considering the various social, economic, and personal factors that influence migration decisions, policymakers can provide more effective strategies to retain skilled and talented labor and promote sustainable growth in developing countries. One of the main challenges in studying the flow of elite migration is the complexity of the decision-making process and a set of factors that lead to migration decisions. Agent-based modeling is a useful tool for understanding how individual decisions can lead to large-scale migration patterns.

This project combines game theory and genetic algorithms in a simulation model for evolutionary learning and strategic behavior. It is often observed in the real world that strategic scenarios change over time, and deciding agents need to adapt to new information and environmental structures. Yet, game theory models often focus on static games, even for dynamic and temporal analyses. This simulation model introduces a heuristic procedure that enables these changes in strategic scenarios with Genetic Algorithms. Using normalized 2x2 strategic-form games as input, computational agents can interact and make decisions using three pre-defined decision rules: Nash Equilibrium, Hurwicz Rule, and Random. The games then are allowed to change over time as a function of the agent’s behavior through crossover and mutation. As a result, strategic behavior can be modeled in several simulated scenarios, and their impacts and outcomes can be analyzed, potentially transforming conflictual situations into harmony.

Organisms, Individuals and Organizations face the dilemma of exploration vs. exploitation
Identifying the optimal trade-off between the two is a challenge
Too much exploration (e.g. gaining new knowledge) can be detrimental to day-to-day survival and too much exploitation (applying existing knowledge) could be detrimental to long term survival esp. if conditions change over time

The purpose of the model is to investigate how the amount of resources acquired (wealth/success) is related to persistence with the strategy of local exploration under different resource distributions, availability of resources over time and cost of relocation

The model is designed to simulate the behavior and decision-making processes of individuals (agents) in a social network. It aims to represent the changes in individual probability to take any action based on changes in attributes. The action is anything that can be reasonably influenced by the three influencing methods implemented in this model: peer pressure, social media, and state campaigns, and for which the user has a decision-making model. The model is implemented in the multi-agent programmable environment NetLogo 6.3.0.