Diffusion Model Analyst and Pathfinder is an interactive browser simulator for modeling how outreach spreads, stabilizes, or collapses in a collective. It accompanies the paper State-Dependent Participation Diffusion on Adaptive Influence Networks and turns the paper’s agent-based diffusion model into a visual, explorable system.
The simulator runs entirely in the browser as a single HTML file. There are no dependencies, no build step, and no server required.
Link: Diffusion Model Analyst and Pathfinder
-
Forward prediction mode Configure a collective’s network, events, visibility, capacity, and mentoring structure, then simulate how impact and participation evolve across a timeframe.
-
Single-network sample runs Press Run sample to generate one randomized social network and watch the collective grow, plateau, or collapse week by week.
-
Monte Carlo ensemble simulations Press Run simulation to run many randomized trials and view average weekly outcomes, uncertainty bands, survival probability, and final turnout ranges.
-
Reverse planning mode Set a target end-of-term attendance level, budget constraints, and intervention limits. The planner searches for a lightest-touch strategy that reaches the target.
-
Plan testing After a reverse-mode plan is generated, test it using either one randomized sample or a full ensemble simulation.
-
Interactive network visualization Each core network node represents a simulated agent. Color shows motivation, the loyalty arc shows accumulated commitment, and hover readouts show individual motivation, loyalty, fatigue, attendance, and status.
-
Public-body visualization layer The surrounding faint dots represent the wider population body outside the modeled core network. Drawn-in public members show broader reach from events and acquaintance pathways without changing the underlying behavioral equations.
Download or clone the repository, then open:
index.html
in any modern browser.
No installation is required.
The simulator is also available at:
https://hgd-dev.github.io/DMAP/
Use this mode to explore how a collective behaves under different initial conditions and interventions.
Main controls include:
- Agents / network size — number of simulated core network agents.
- Regime — collapse-prone or stabilizing dynamics.
- Weeks to Simulate — timeframe for running the network
- Current active membership — initialization of member count
- Motivation distribution — initialization of motivation across current members
- Commitment/loyalty distribution — initialization of commitment or loyalty across current members
- Fatigue distribution — initialization of fatigue across current members
- Group size — how students cluster into communities.
- Close ties — within-group influence connections.
- Bridge ties — cross-group influence connections.
- Extra acquaintance ties — additional weak long-range ties.
- Public population — size of the wider visual collective body around the core network.
- Events — week and intensity of enthusiasm pulses.
- Visibility — promotion / expressive reach.
- Capacity — effective room or format capacity.
- Scaffolding — early mentoring support across selected weeks.
- Trials — number of randomized runs used in ensemble simulation.
Use Run sample to animate one randomized network.
Use Run simulation to average many randomized networks and view the expected weekly outcome.
Use this mode to ask:
What is the lightest intervention plan that reaches a desired attendance target?
Set:
- target steady attendance,
- main controls
Then press Find a plan.
If a feasible plan is found, you can test it with:
- Run sample plan — one randomized animated realization.
- Run simulation plan — aggregate average across randomized networks.
- Node color represents motivation on a plasma scale from low to high.
- Outer ring marks a core simulated network student.
- Amber arc around a node represents loyalty.
- Small faint public dots represent the wider student body outside the simulated core network.
- Lavender public dots represent students drawn in through events or acquaintances.
- Central hub represents the club; its size and glow scale with turnout.
- Chart line shows attendance over time.
- Orange uncertainty band appears in ensemble mode and shows the middle 50% of trial outcomes.
- Timeline slider lets you inspect either one sampled run or the average ensemble state week by week.
The simulator implements the behavioral equations from State-Dependent Participation Diffusion on Adaptive Influence Networks as an interactive companion.
The following are visualization or interface layers rather than changes to the model:
- the public-body dots surrounding the core network,
- the sample-vs-ensemble interface,
- the averaged timeline display,
- hover readouts and visual encodings,
- the reverse-mode button workflow.
The close-tie and bridge-tie controls affect the randomized graph ensemble. They change how the social network is sampled, not the core behavioral update equations.
.
├── index.html
├── README.md
├── Paper/
│ ├── State-Dependent Participation Diffusion on Adaptive Influence Networks - Hudson Dong.pdf
│ ├── State-Dependent Participation Diffusion on Adaptive Influence Networks - Hudson Dong.tex
├── ODD Protocol/
│ ├── ODD Protocol for State-Dependent Participation Diffusion on Adaptive Influence Networks - Hudson Dong.pdf
│ └── ODD Protocol for State-Dependent Participation Diffusion on Adaptive Influence Networks - Hudson Dong.tex
├── Figures/
│ ├── Figure1_heatmap.pdf
│ └── Figure2_trajectories.pdf
├── Supplements/
│ ├── Color Readout Supplement - H. Dong.pdf
│ └── Color Readout Supplement - H. Dong.tex
└── LICENSE
Based on:
Hudson Dong. State-Dependent Participation Diffusion on Adaptive Influence Networks.
- MIT License
This simulator is intended for qualitative exploration of diffusion regimes, not calibrated empirical forecasting. The outputs should be interpreted as structural tendencies of the model under randomized social-network conditions.