Skip to content

Latest commit

 

History

History
80 lines (57 loc) · 4.11 KB

README.md

File metadata and controls

80 lines (57 loc) · 4.11 KB
Raw

EpiRust

An agent-based epidemiology simulation framework built in Rust

Build Status

This is a monorepo containing code for the EpiRust framework -

  1. engine - The main EpiRust engine, which can be run as a CLI app
  2. orchestrator - Coordinator for multiple engines to run large scale simulations

Note: We are no longer maintaining the Web UI for EpiRust which allows you to trigger simulations and grid visualization for upto 1000 agents. Hence, it has been removed from here. Please refer sims2020 branch for web.

Quickstart

The quickest way to run the application is using the CLI interface on the engine. Follow the docker or local setup to get the engine up and running.

Docker

Prerequisites:

Build EpiRust engine docker image
docker build -t epirust-engine -f ./engine-app/Dockerfile .
Running EpiRust engine docker container
docker run --rm --name epirust-engine epirust-engine /bin/sh -c './epirust-engine -c config/default.json && ls *.csv'
  • Copy the listed CSV produced from the container to the host machine
docker cp epirust-engine:/home/nonroot/app/<CSV_FILE> .
  • Go to the visualization section to plot the copied CSV file

Local setup

Prerequisites:

  • Install Rust and Cargo (version 1.40.0 or above). Refer to https://rustup.rs/ for rust installation
  • Install cmake (version 3.16.4 or later). Installation instructions will depend on your platform. On MacOS you can install it using brew install cmake. Refer to https://cmake.org/download/
  • The following dependencies are needed on Debian: ca-certificates curl file build-essential autoconf automake autotools-dev libtool xutils-dev cmake pkg-config libfreetype6-dev libfontconfig1-dev xclip
  • Python 3 (for visualization and charting). We also need the pandas and matplotlib libraries which can be installed using pip or conda.
    • pip install pandas matplotlib

Running standalone simulations

To run in standalone mode, run this:

    RUST_LOG=info cargo run --bin engine-app -- -i <engine-id> -c path/to/simulation-config

Visualization:

  • After the simulation is run, it will generate a CSV file. We can plot this using a simple script included in the engine/plot directory
    • Ensure you're in the engine directory
    • Run python plot/plot.py --data-path <PATH_TO_CSV_FILE> - this will plot the csv you provide.

Sample output: Epicurves plot

Running multi-region simulations

You can run EpiRust for multi-region simulations (e.g. to simulate a group of cities). This will require the orchestrator, and each engine representing a region. The architecture page on the wiki describes how this works at a high level.

Steps for running a multi-region simulation:

  1. Prepare a configuration. A configuration is a json file which consists of two parts:
    • The simulation configuaration for each engine. (Note that currently there is some repetition needed in the configuration for each engine)
    • The travel plan, which defines a matrix containing the daily travellers from one region to another. You can take a look at the orchestrator/config directory for examples of the configuration.
  2. To start with KAFKA on local, follow this
  3. To start with MPI on local, follow this
  4. To start with MPI on cloud, follow this
  5. To start with KAFKA on cloud, you have to set up KAFKA cluster on cloud. (follow this)

It will generate output CSV and JSON files which you can use to for analysis and charting.

License

EpiRust is an open source project licensed under AGPL v3