The TERROIR agent-based model was built for the multi-level analysis of biomass and nutrient flows within agro-sylvo-pastoral villages in West Africa. It explicitly takes into account both human organization and spatial extension of such flows.

This model visualizes gradient descent optimization - the fundamental algorithm used to train neural networks and other machine learning models. Agents represent different optimization algorithms searching for the minimum of a loss landscape (the “error surface” that ML models try to minimize during training).

The model demonstrates how different optimizer types (SGD, Momentum with different parameters) behave on various loss landscapes, from simple bowls to the notoriously difficult Rosenbrock “banana valley” function. This helps build intuition about why certain optimization algorithms work better than others for different problem geometries.

HOW IT WORKS

…

This computational model is an agent-based model (ABM) developed to investigate how repeated failures of emerging niches accumulate and influence the trajectory of socio-technical transitions. Built in AnyLogic 8.7.11, the model simulates the dynamic interactions between a dominant regime and sequential niche entrants within a two-dimensional practice space. It models alignment, movement, and competition based on technological maturity and market penetration. The model utilizes a reinforcing feedback structure linking consumer support, output, resource accumulation, and capacity development (Physical and Institutional Capacity). A complete model specification following the ODD+D (Overview, Design concepts, Details, and Decision) protocol is included in the documentation.

AgModel is an agent-based model of the forager-farmer transition. The model consists of a single software agent that, conceptually, can be thought of as a single hunter-gather community (i.e., a co-residential group that shares in subsistence activities and decision making). The agent has several characteristics, including a population of human foragers, intrinsic birth and death rates, an annual total energy need, and an available amount of foraging labor. The model assumes a central-place foraging strategy in a fixed territory for a two-resource economy: cereal grains and prey animals. The territory has a fixed number of patches, and a starting number of prey. While the model is not spatially explicit, it does assume some spatiality of resources by including search times.

Demographic and environmental components of the simulation occur and are updated at an annual temporal resolution, but foraging decisions are “event” based so that many such decisions will be made in each year. Thus, each new year, the foraging agent must undertake a series of optimal foraging decisions based on its current knowledge of the availability of cereals and prey animals. Other resources are not accounted for in the model directly, but can be assumed for by adjusting the total number of required annual energy intake that the foraging agent uses to calculate its cereal and prey animal foraging decisions. The agent proceeds to balance the net benefits of the chance of finding, processing, and consuming a prey animal, versus that of finding a cereal patch, and processing and consuming that cereal. These decisions continue until the annual kcal target is reached (balanced on the current human population). If the agent consumes all available resources in a given year, it may “starve”. Starvation will affect birth and death rates, as will foraging success, and so the population will increase or decrease according to a probabilistic function (perturbed by some stochasticity) and the agent’s foraging success or failure. The agent is also constrained by labor caps, set by the modeler at model initialization. If the agent expends its yearly budget of person-hours for hunting or foraging, then the agent can no longer do those activities that year, and it may starve.

Foragers choose to either expend their annual labor budget either hunting prey animals or harvesting cereal patches. If the agent chooses to harvest prey animals, they will expend energy searching for and processing prey animals. prey animals search times are density dependent, and the number of prey animals per encounter and handling times can be altered in the model parameterization (e.g. to increase the payoff per encounter). Prey animal populations are also subject to intrinsic birth and death rates with the addition of additional deaths caused by human predation. A small amount of prey animals may “migrate” into the territory each year. This prevents prey animals populations from complete decimation, but also may be used to model increased distances of logistic mobility (or, perhaps, even residential mobility within a larger territory).

…

This is a relatively simple foraging-radius model, as described first by Robert Kelly, that allows one to quantify the effect of increased logistical mobility (as represented by increased effective foraging radius, r_e) on the likelihood that 2 randomly placed central place foragers will encounter one another within 5000 time steps.

ReMoTe-S is an agent-based model of the residential mobility of Swiss tenants. Its goal is to foster a holistic understanding of the reciprocal influence between households and dwellings and thereby inform a sustainable management of the housing stock. The model is based on assumptions derived from empirical research conducted with three housing providers in Switzerland and can be used mainly for two purposes: (i) the exploration of what if scenarios that target a reduction of the housing footprint while accounting for households’ preferences and needs; (ii) knowledge production in the field of residential mobility and more specifically on the role of housing functions as orchestrators of the relocation process.

The HUMan impact on LANDscapes (HUMLAND) model has been developed to track and quantify the intensity of different impacts on landscapes at the continental level. This agent-based model focuses on determining the most influential factors in the transformation of interglacial vegetation with a specific emphasis on burning organized by hunter-gatherers. HUMLAND integrates various spatial datasets as input and target for the agent-based model results. Additionally, the simulation incorporates recently obtained continental-scale estimations of fire return intervals and the speed of vegetation regrowth. The obtained results include maps of possible scenarios of modified landscapes in the past and quantification of the impact of each agent, including climate, humans, megafauna, and natural fires.

The MML is a hybrid modeling environment that couples an agent-based model of small-holder agropastoral households and a cellular landscape evolution model that simulates changes in erosion/deposition, soils, and vegetation.

This is based off my previous Profiler tutorial model, but with an added tutorial on converting it into a model usable with BehaviorSpace, and creating a BehaviorSpace experiment.

The Agent-Based Model for Multiple Team Membership (ABMMTM) simulates design teams searching for viable design solutions, for a large design project that requires multiple design teams that are working simultaneously, under different organizational structures; specifically, the impact of multiple team membership (MTM). The key mechanism under study is how individual agent-level decision-making impacts macro-level project performance, specifically, wage cost. Each agent follows a stochastic learning approach, akin to simulated annealing or reinforcement learning, where they iteratively explore potential design solutions. The agent evaluates new solutions based on a random-walk exploration, accepting improvements while rejecting inferior designs. This iterative process simulates real-world problem-solving dynamics where designers refine solutions based on feedback.

As a proof-of-concept demonstration of assessing the macro-level effects of MTM in organizational design, we developed this agent-based simulation model which was used in a simulation experiment. The scenario is a system design project involving multiple interdependent teams of engineering designers. In this scenario, the required system design is split into three separate but interdependent systems, e.g., the design of a satellite could (trivially) be split into three components: power source, control system, and communication systems; each of three design team is in charge of a design of one of these components. A design team is responsible for ensuring its proposed component’s design meets the design requirement; they are not responsible for the design requirements of the other components. If the design of a given component does not affect the design requirements of the other components, we call this the uncoupled scenario; otherwise, it is a coupled scenario.