Atomica
Atomica is a simulation engine for compartmental models. It can be used to simulate disease epidemics, health care cascades, and many other things.
For detailed documentation, visit https://atomica.tools/docs
Installation
Atomica requires Python 3.10 or later and is distributed via PyPI. To install, run
pip install atomica
Git installation
If you want to install a different branch of Atomica, or plan to make changes to the Atomica source code, you will need to install Atomica via Git rather than via PyPI. This can be performed using
git clone https://github.com/atomicateam/atomica.git
cd atomica
pip install -e .
Running tests
Atomica includes a suite of tests, some of which get automatically run and others that are used manually. The automated test suite can be executed with pytest, and can be run from within an isolated environment using tox. To use the tests, you will need to follow the steps above to perform a 'Git installation' because the tests are not included in the PyPI distribution. After installation, you can run individual test scripts from the tests directory with commands like:
python tests/testworkflow.py
Note that many of the tests open matplotlib figures as part of the test. If the test script is run on a machine without a display available, the error
_tkinter.TclError: couldn't connect to display "localhost:0.0"
will be raised. In that case, simply set the matplotlib backend to agg which allows the calls to succeed with a display present. For example, run
export MPLBACKEND=agg
python tests/testworkflow.py
To run the automated suite, install the test dependencies using
pip install -e ".[test]"
which will install the additional development dependencies. Then, to run the automated suite, from the root directory (the one containing README.md) run:
pytest
To run the tests in an isolated virtual environment, from the root directory, run
tox
If you don't have tox, install it using pip install tox. To test against a specific Python version, pass it as an argument, e.g.
tox -e py312
Claude Code integration
Atomica ships an MCP server (atomica.mcp) that exposes tools for querying framework and databook files and a set of built-in workflow skills (MCP prompts). The tools let Claude read compartments, parameters, transitions, and variable metadata directly from .xlsx framework files, as well as read and write data in .xlsx databook files. The skills guide Claude through multi-step workflows such as producing a structured summary of a framework.
To register the server with Claude Code, run the following from within your project directory:
claude mcp add atomica -- uv run python -m atomica.mcp
Once added, the tools and skills are available automatically in any Claude Code session for that project. If you use Atomica across multiple projects and want the MCP server available in all of them without repeating the claude mcp add step, register it at the user level instead:
claude mcp add -s user atomica -- uv run python -m atomica.mcp
The Atomica MCP will then be used whenever you are working within a project that has atomica as a dependency.
Adding custom skills
Skills are plain Markdown files in atomica/mcp/skills/. To add a new workflow, drop a .md file into that directory — it is registered as an MCP prompt automatically when the server starts, with no code changes required. The first # Heading line becomes the prompt description shown in the MCP client.
Troubleshooting
Installation fails due to missing numpy
If running pip install -e . in a new environment, numpy must be installed prior to scipy. In some cases,
installing numpy may fail due to missing compiler options. In that case, you may wish to install numpy via Anaconda
(by installing Python through Anaconda, and using conda install numpy scipy matplotlib). In general, our experience
has been that it is easier to set up the C binaries for numpy and the QT dependencies for matplotlib via Anaconda
rather than doing this via the system, which involves different steps on every platform.
Figure plotting hangs
On some systems, the default matplotlib backend may hang - this is not an issue with atomica. To resolve, try changing the backend by including
import matplotlib
matplotlib.use("Qt5Agg")
at the very start of your script. A different backend may be required for your system. You can make the change persistant by setting the backend in your matplotlibrc file.
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Filter files by name, interpreter, ABI, and platform.
If you're not sure about the file name format, learn more about wheel file names.
Copy a direct link to the current filters
File details
Details for the file atomica-1.32.1.tar.gz.
File metadata
- Download URL: atomica-1.32.1.tar.gz
- Upload date:
- Size: 1.6 MB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via:
twine/6.2.0 CPython/3.14.6
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
2bb1d42e1f36c56cdfdb71126d1a7c79c5197bad88dac4057cb7ff8506b72a1b
|
|
| MD5 |
21bb2fc0d9726ef7f5fda4700e0338ce
|
|
| BLAKE2b-256 |
47000bbfb2fdd7ea312041c3b9b67d16175798ba574f31e1878a244155efc72d
|
File details
Details for the file atomica-1.32.1-py3-none-any.whl.
File metadata
- Download URL: atomica-1.32.1-py3-none-any.whl
- Upload date:
- Size: 1.7 MB
- Tags: Python 3
- Uploaded using Trusted Publishing? No
- Uploaded via:
twine/6.2.0 CPython/3.14.6
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
e3c817bf3e1b37793ef6283cd89688fb3863545e4753995a29068e82e60056af
|
|
| MD5 |
38c02f936b69a847ab0a6feafb8b9651
|
|
| BLAKE2b-256 |
523fba41679cb24439aca5aee56915d01ecf877ea5f51a6936b8d7abde41a1d7
|
