The Global Centre for Risk and Innovation (GCRI)
Agent-based models represent many individual actors, such as households, firms, vehicles or cells, following rules and interacting, so that system patterns emerge from the bottom up. They suit questions where heterogeneity, adaptation and interaction matter, and they are hard to validate. This page sets out how agent-based and complexity models are built, documented and tested, and what evidence serves the institutions that use them.
GCRI · In development · service level R0
Agent-based models are simulations of many individual actors following rules and interacting, from which system patterns emerge. They suit questions where heterogeneity, adaptation and interaction matter, and because they have many parameters and little data, they need standard documentation and validation against several observed patterns.
For agent-based and complexity models, in GCRI's spatial-intelligence reading, the map opens on true-colour imagery, population density and night lights with relief in 3D, framed on North America; every other NASA layer, live feed and forecast waits in the layer list.
The interactive map needs JavaScript; the layer list and sources are below.
Layers 3
Hazards and exposure, for this page
Keyboard, with the map focused
Hold Ctrl or ⌘ and scroll to zoom
Latest available
NASA GIBS / EOSDIS · ASTER GDEM (METI, NASA) · Blue Marble (NASA) · © OpenStreetMap contributors (ODbL) · Natural Earth
Loading the latest NASA imagery.
Drag to move. Ctrl or ⌘ and scroll, pinch or the buttons to zoom; hover for the values under the pointer; reset returns to the region.
For agent-based and complexity models, in GCRI's spatial-intelligence reading, the map opens on true-colour imagery, population density and night lights with relief in 3D, framed on North America; every other NASA layer, live feed and forecast waits in the layer list.Satellite and model products as NASA publishes them, each with its own date; indicators, not assessments of any place; claim ceiling INTEGRATED.
Layers and dates
Behaviour, adaptation and emergence in social-technical systems. The dossier below sets out how the risk develops, what is measured, where the data come from, how it is modelled and who decides.
Agents differ in resources, behaviour and location. Averages can hide the variation that drives outcomes such as who is harmed by a shock.
Agents respond to each other and to policy, so the system changes as it is studied, which equation-based models may miss.
Simple rules can produce unexpected collective patterns, such as tipping points or segregation, which are useful but need careful explanation.
Agent-based models often have more parameters than data can constrain, so documentation and pattern-oriented validation are essential.
Each term as this page uses it. Where a standard-setting body defines the term, the wording follows that body.
| Term | Definition |
|---|---|
| Agent | An autonomous entity in a model, such as a person, household or firm, with its own state and rules. |
| Pattern-oriented modelling | Validating a model by testing whether it reproduces several observed patterns at once. |
| Stochastic replicate | One of several runs of a model with different random seeds. |
Each indicator carries its unit, its cadence, the kind of source behind it and the evidence label a value would carry. No values are shown: a programme supplies them under its own authority.
| Indicator | Unit | Cadence | Source type | Label |
|---|---|---|---|---|
| Models documented with the ODD protocol | percent of published models | per review | Literature review | OBSERVED |
| Patterns reproduced in validation | count of target patterns | per model | Validation studies | ESTIMATED |
| Parameters calibrated to data | percent of parameters | per model | Model documentation | DECLARED |
| Code published | percent of published models | per review | Code archives | OBSERVED |
| Sensitivity analysis performed | percent of published models | per review | Literature review | OBSERVED |
| Runs per experiment | count of stochastic replicates | per experiment | Model documentation | DECLARED |
The topic as a computational graph: named sources feed the methods, an evidence review gates the outputs, and the decisions stay with the authorities that hold them.
Fig. 02Nodes carry their type, two metric rows and a state; connectors show what feeds what; dashed frames name the owners; the timeline is the cadence. Design of the pipeline, not a record of a run.
Named public datasets and institutional records, each with its licence class. Sources by agreement are used only under their terms and never republished.
| Source | What it gives | Licence class |
|---|---|---|
| CoMSES Net Computational Model Library | Archive of agent-based models with code and ODD documentation | Open, licence per model |
| ODD protocol publications | Documentation standard for agent-based models | Licence per publication |
| NetLogo and Mesa | Open-source agent-based modelling platforms | Open source, GPL and Apache licences |
| Journal of Artificial Societies and Social Simulation | Peer-reviewed agent-based research | Open access, CC BY |
| Pattern-oriented modelling literature | Validation approaches for complex models | Licence per publication |
Systems Science and Modelling prepares evidence. The decisions below are taken by the authorities named, under their own rules, and never by the platform.
| Who decides | The question | Evidence this page serves |
|---|---|---|
| Modellers | How to document, calibrate and validate agent-based models | Templates; validation designs |
| Institutions using simulation results | When agent-based results are credible enough to inform decisions | Validation and sensitivity evidence |
| Journals and reviewers | What documentation to require | Documentation evidence |
In complex systems, simple local rules can produce large-scale patterns: traffic jams form without a cause at the front; panic buying spreads through observation of others; markets can swing as traders follow trends. These effects arise from interaction, feedback and heterogeneity, which equation-based models that assume average behaviour may miss.
Agents adapt. People change behaviour in response to risk, warnings and the actions of others; firms adjust strategies; institutions change rules. Adaptation can stabilise systems or create new risks, as when protection encourages development in hazard areas.
Complex systems can show tipping points, where small changes cause abrupt shifts, and path dependence, where history constrains future possibilities.
Computational models of interacting agents, using complex systems methods. Many parameters can be hard to estimate.
Surveys, experiments and observations to ground agent rules. Behaviour in crises is hard to observe in advance.
Testing whether models reproduce several observed patterns at once. Matching patterns does not prove mechanisms.
Exploring outcome ranges across rules and parameters, using computing resources.
Agent-based models are powerful for exploring mechanisms but difficult to calibrate and validate. Results are presented as explorations, not predictions, unless validation supports more.
The current service level is R0: information and enquiries. An enquiry receives a reference and names the next step; applications, payments, compute allocations and registrations open only when their team, terms and review capacity exist.