This model simulates a simple aquatic ecosystem containing fish and food. It explores how individual interactions such as movement, feeding, and reproduction shape the population dynamics of fish over time.

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.

This repository contains the Python implementation of an agent-based model investigating how localized boundary-crossing dynamics generate large-scale connectivity in structured multi-attractor landscapes.

Agents evolve in a continuous two-dimensional environment composed of attractor basins. A fraction of agents exhibits exploratory higher-mobility dynamics, while the remaining agents remain locally constrained. The model analyzes how localized configurational transitions accumulate into transition networks that progressively integrate the explored state space.

The repository includes:

…

Exploring learned cooperation, coevolution and free-riding. Learning is achieved through Multi-Agent Deep Reinforcement Learning (MADRL) in an ecological environment. The environment emits no other than sparse reproduction rewards. No reward shaping, no explicit cooperation signal.

Interest-based compound economies generate monotonically increasing wealth inequality through multiplicative accumulation dynamics, yet the conditions under which gift-based reciprocal exchange outperforms such systems in collective well-being remain unquantified. We present Zensei Wago (全生和合), a seven-layer agent-based model comparing a Gift Resource Circulation (GRC) economy with a Compound Interest Circulation (CIC) economy under identical initial conditions. Across N = 5000 Monte Carlo replications (T = 700 ticks, N = 100 agents), GRC produced significantly higher collective resonance than CIC (p < 0.001, Cohen’s d = +0.171), above a critical prosocial threshold pm ≈ 0.698. Cohen’s d grows monotonically with duration — d = +1.943 at T = 1500 and d = +4.126 at T = 3000 — driven primarily by structural collapse of CIC resonance as inequality exceeds a critical Gini threshold (G > 0.333), while GRC resonance remains stable. The gift mechanism further decouples collective well-being from distributional outcomes, generating resonance through relational quality rather than material redistribution. Network topology analysis across seven configurations — combining a Watts-Strogatz rewiring sweep and a T = 1500 longitudinal replication — reveals that ring topology maximises GRC advantage (d = +1.17), that most topology-dependent reversals are transient (sparse and small-world both transition to significantly positive by T = 1500), and that a critical rewiring threshold of p ≈ 0.10–0.20 separates GRC-advantaged from GRC-disadvantaged network configurations. Scale-free networks remain persistently adverse (d = -7.24*), requiring structural redesign for gift-economy viability.

This computational model accompanies the article “The Informational Assumptions of Schelling Segregation: An Agent-Based Decomposition of Cue Inference, Cultural Schemas, and Residential Sorting.” It implements an agent-based model in which agents infer latent neighborhood-type classes from noisy non-demographic cues through schema-specific diagnostic mappings, update beliefs, and relocate when satisfaction on a preferred latent class falls below a threshold.

The model serves as a mechanism-isolation device for studying the informational architecture underlying Schelling-style residential sorting. It includes the principal sweep configuration (14,400 runs across a seven-parameter grid), a disagreement-metric sub-sweep with permutation-minimized Jensen-Shannon divergence recorded natively, controls (positive, negative, and frozen-belief), a paired-seed cue-channel perturbation experiment, and selected-cell sensitivity sweeps for cue persistence and home-biased mobility.

The full ODD protocol, parameter manifests, deterministic seed schedules, processed outputs, regenerable figure scripts, the verification test suite, and the satisfaction-mapping audit document are included. Every reported run is deterministic given a (config, seed) pair, and an included audit script verifies bit-for-bit replay on sampled runs.

This model is a part of an ongoing research project on Multiagent Reinforcement Learning (MARL). The ODD protocol is included in the model. In this version of the model, Proximal Policy Optimization (PPO) is designed in the agent behaviors. It also includes a designed experiment in its Behavior Space which is used in the Response Surface Methodology and training of an Artificial Neural Network (ANN) based Recommender System.

An agent-based microsimulation of insecticide-treated net (ITN) distribution and adoption in Kenya (2003–2024), integrating the Theory of Planned Behaviour, Rogers diffusion, Weibull net decay, and a GPS-based two-layer social network. 8,561 household agents calibrated via Approximate Bayesian Computation to six DHS/MIS survey waves, achieving 2.42 pp mean absolute error on Kenya-level ownership. The analysis chain supports mechanism counterfactuals and policy experiments on equity outcomes of ITN distribution strategies.

The current model is designed to examine whether—and under what conditions—minority influence can generate social change. Specifically, the model assesses whether empirically validated psychological mechanisms of indirect minority influence, operating in combination, can produce system-level social change, defined as the initial minority opinion becoming the majority position. Notably, this model formalizes Moscovici’s (1976) genetic model of social influence using agent-based modeling.

This is a model that explores how a few fishermen sharing a common fishery learn their harvesting strategies under different incentive settings, and how individual greed, cooperation, and sustainability penalties shape resource depletion and the tragedy of the commons.