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.

SERDUX-MARCIM simulates the propagation of a cyberattack over the computational network of an organization in the maritime sector, at the strategic level of decision-making. It is the instantiation of ABM-MARCIM, the agent-based model of the MARCIM framework for the modeling and simulation of maritime cyberdefense.

Every computational asset of the target organization – servers, endpoints, routers, gateways, vessel systems, radars – is an agent that occupies one of six states at each time step: Susceptible, Exposed, Resistant, Degraded, Unavailable or Destroyed, the initials of which give the model its name. The states Degraded, Unavailable and Destroyed are associated with the D5 cyberattack effects (disrupt, degrade, deny, destroy, deceive) as a function of the degree and the duration of the attack. Thirteen transitions between states are admissible.

Unlike a conventional agent-based model, the local update function is not an individual behavioral rule. It is a system of six ordinary differential equations with eight time-dependent transition rates – propagation, cyberattack (degraded), cyberattack (unavailable), cyberattack (destroyed), recovery, sanitation, loss of resistance, and unavailability by other causes. The values of those rates derive from the capabilities of the target organization, the capabilities of the attacker, and the degree and duration of the cyberattack, computed through a cyber risk approach aligned with the OWASP Risk Rating Methodology, the ISACA categorization of security controls and the IMO Guidelines on Maritime Cyber Risk Management.

…

This model aims to study the dynamic propagation of individual behaviour within social networks, focusing on how normative expectations (NE) and experiential expectations (EE) jointly influence behavioural decisions. It also explores the long-term effects of different intervention scenarios (such as enhancing visibility, considering indirect social links, and education) on behavioural propagation patterns and the overall behaviour of the group.
The model was developed in NetLogo 6.4. It generates simulated groups based on large-scale survey data, utilizing NetLogo’s CSV, Table, and Matrix extensions. The model also employs the NW extension to enable network analysis functionality.
The model is designed for research “Shaping social norms to promote individual response behavior in public crises: An agent-based modeling approach” in Journal of Cleaner Production, Volume 554, 8 April 2026, 148014
https://doi.org/10.1016/j.jclepro.2026.148014

This model simulates the form and function of an idealised estuary with associated barrier-spit complex on the north east coast of New Zealand’s North Island (from Bream Bay to central Bay of Plenty) during the years 2010 - 2050 CE. It combines variables from social, ecological and geomorphic systems to simulate potential directions of change in shallow coastal systems in response to external forcing from land use, climate, pollution, population density, demographics, values and beliefs. The estuary is over 1000Ha, making it a large estuary according to Hume et al. (2007) - there are 12 large estuaries in the Auckland region alone (Suyadi et al., 2019). The model was developed as part of Andrew Allison’s PhD Thesis in Geography from the School of Environment and Institute of Marine Science, University of Auckland, New Zealand. The model setup allows for alteration of geomorphic, ecological and social variables to suit the specific conditions found in various estuaries along the north east coast of New Zealand’s North Island.
This model is not a predictive or forecasting model. It is designed to investigate potential directions of change in complex shallow coastal systems. This model must not be used for any purpose other than as a heuristic to facilitate researcher and stakeholder learning and for developing system understanding (as per Allison et al., 2018).

The Megafaunal Hunting Pressure Model (MHPM) is an interactive, agent-based model designed to conduct experiments to test megaherbivore extinction hypotheses. The MHPM is a model of large-bodied ungulate population dynamics with human predation in a simplified, but dynamic grassland environment. The overall purpose of the model is to understand how environmental dynamics and human predation preferences interact with ungulate life history characteristics to affect ungulate population dynamics over time. The model considers patterns in environmental change, human hunting behavior, prey profitability, herd demography, herd movement, and animal life history as relevant to this main purpose. The model is constructed in the NetLogo modeling platform (Version 6.3.0; Wilensky, 1999).

The purpose of the Credit and debt market of low-income families model is to help the user examine how the financial market of low-income families works.

The model is calibrated based on real-time data which was collected in a small disadvantaged village in Hungary it contains 159 households’ social network and attributes data.
The simulation models the households’ money liquidity, expenses and revenue structures as well as the formal and informal loan institutions based on their network connections. The model forms an intertwined system integrated in the families’ local socioeconomic context through which families handle financial crises and overcome their livelihood challenges from one month to another.
The simulation-based on the abstract model of low-income families’ financial survival system at the bottom of the pyramid, which was described in following the papers:

…

This model implements a coupled opinion-mobility agent-based framework in NetLogo, extending Attraction-Repulsion Model (ARM) dynamics with endogenous migration in continuous 2D space.

Each agent has an opinion s in [0,1] and a spatial position (x,y). Agents interact locally within an interaction radius, with exposure-controlled interaction probability. Opinion updates follow ARM rules: attraction for small opinion distance and repulsion for large distance (tolerance threshold T). After social interaction, agents move according to a social-force mechanism that balances attraction to similar neighbors and avoidance of dissimilar neighbors, controlled by orientation bias (approaching goods vs leaving bads). The model also includes an optional exposure-mobility coupling setting.

Main outputs include polarization (P), spatial assortativity (Moran’s I), mixed-neighbor fraction (f_mix), and good-component count (N_g). The model is designed to study phase behavior of polarization and segregation under mobility and tolerance heterogeneity.

…

This model presents an autonomous, two-lane driving environment with a single lane-closure that can be toggled. The four driving scenarios - two baseline cases (based on the real-world) and two experimental setups - are as follows:

• Baseline-1 is where cars are not informed of the lane closure.
• Baseline-2 is where a Red Zone is marked wherein cars are informed of the lane closure ahead.
• Strategy-1 is where cars use a co-operative driving strategy - FAS. ^[1]
• Strategy-2 is a variant of Strategy-1 and uses comfortable deceleration values instead of the vehicle’s limit.
  …

The Hohokam Trade Networks Model focuses on key features of the Hohokam economy to explore how differences in trade network topologies may show up in the archaeological record. The model is set in the Phoenix Basin of central Arizona, AD 200-1450.

The model simulates the spread of a virus through a synthetic network with a degree distribution calibrated on close-range contact data. The model is used to study the macroscopic consequences of cross-individual variability in close-range contact frequencies and to assess whether this variability can be exploited for effective intervention targeting high-contact nodes.