Naclist
DOCUMENTATION / 文档

From city fabric
to transmission chains.

A field guide to the model, the interface, and the limits of what a synthetic outbreak can tell us.

01

What this project is

The public framework presents one equilibrated synthetic city with 10,000 synthetic residents. Every resident has a household, demographic identity, activity schedule, workplace or school, road-constrained mobility and time-stamped contacts.

The epidemic layer is deliberately downstream. It does not rebuild the city when Target R0 changes; it discards only disease state and replays transmission over the same society.

ILLUSTRATIVENo profile is calibrated to a named pathogen.
02

Causal architecture

01Urban structureroads · blocks · buildings
↓
02Synthetic societyhouseholds · jobs · schools
↓
03Daily behaviouractivities · routes · healthcare
↓
04Temporal contacts7 semantic layers · 15 min
↓
05Disease dynamicsS → E → P → I → R
WEB SUBSTRATEThe browser loads the same exported 10K city geometry used locally: 6,256 roads, 400 blocks, 3,130 building footprints, 31 healthcare facilities and 14 days of hourly agent positions. The compressed bundle is about 7 MB.
03

Public model scale

PUBLIC MODEL10,000 agents

Individual points, live movement and rapid parameter comparison. Designed for explanation and network analysis.

04

Three transmission mechanisms

These are generic mechanism profiles, not named diseases. Natural-history parameters stay locked in the public interface.

●

Close-range respiratory

Repeated household, school and workplace encounters dominate.

HH 1.7 · SCH 1.2 · WRK 1.0
◎

Shared-air respiratory

Crowded indoor, community and transport mixing receives more weight.

COM 1.2 · TRN 1.1 · WRK 1.1
◇

Contact-mediated

Persistent close-contact environments matter more than transient mixing.

HH 1.5 · SCH 1.0 · HCR 1.1
05

How to operate the sandbox

  1. 1Load the fixed 10K city and Agent substrate.
  2. 2Choose one generic transmission mechanism.
  3. 3Set Target R0 and initial seed count.
  4. 4Press Apply & Restart. Changing R0 alone never regenerates the city.
  5. 5Use Play, Pause, step and timeline controls to inspect spread.
  6. 6Compare Target R0, Expected network R0, Realized R and current Rt separately.
  7. 7Download the current run JSON or upload an earlier 10K run.
  8. 8Open Transmission Network to inspect Agents, earliest edge times and alternative colour schemes.
06

How to read the result

One stochastic realization is an example, not an estimate. A reliable comparison uses repeated runs and reports medians, IQRs and simulation intervals. The contact-layer chart contains realized transmissions only.

Target R0Requested experiment strength
Expected network R0Calibrated expectation on the substrate
Realized RObserved offspring among completed cases
RtTime-varying effective reproduction
07

Website deployment boundary

GitHub Pages serves the bilingual interface, the authentic compact 10K city substrate and deterministic browser execution. The small backend stores uploaded run JSON files for up to 24 hours; it is not presented as a large-scale simulation service.

GitHub Pages UI
      │ WebSocket / HTTPS
      ▼
FastAPI run store ── 24-hour JSON retention
      │
      └── incremental SEIR · blocks · tree · layers