PySDM
open-atmos.github.io
- Free plan
- Linux
- Self-hosted
What PySDM is
PySDM is a free, GPL v3 open-source package for simulating particle populations as they grow or break apart. Its development focuses on atmospheric cloud physics, including aerosol, cloud, and rain microphysics. The package implements the Super-Droplet Method Monte Carlo algorithm for collisional growth. Documented compute backends include multithreaded CPU, CUDA GPU, and JAX, with the JAX backend still early in development. Examples cover condensation, coagulation, breakup, drop freezing, isotopic fractionation, aqueous chemistry, and seeding in box, parcel, single-column, and two-dimensional prescribed-flow setups. Install PySDM from PyPI with pip or from a local Git checkout. The GPU backend requires NVIDIA hardware and a CUDA driver. Example notebooks can launch through Google Colab or mybinder.org, but are not included in the examples package wheels. PyMPDATA supplies advection in the single-column and two-dimensional examples.
Who it suits
PySDM suits developers and researchers building dispersed-phase simulations, especially atmospheric cloud physics models. It is also useful for users who need CPU, GPU, or documented example configurations.
Good to know, and what to watch
- Uses the Super-Droplet Method algorithm
- Provides CPU and CUDA backends
- Covers varied microphysics processes
- Installable from PyPI with pip
- Released under GPL v3
- JAX backend is early in development
- GPU backend requires NVIDIA hardware and CUDA
- Example notebooks are absent from package wheels
- Atmospheric cloud physics is the development focus
Plans
Every plan and pricePySDM
Free plan
The verdict
PySDM provides an open-source simulation foundation with multiple compute backends and broad atmospheric examples. Hardware requirements and the maturity of the JAX backend should guide implementation choices.
What the maker publishes
- Authoring tools
- Yes source
- Compute backends
- The documentation describes a multithreaded CPU backend, a CUDA GPU backend, and a JAX backend that is early in development. source
- Core method
- It implements the Super-Droplet Method Monte Carlo algorithm for representing collisional growth. source
- Data export
- Yes source
- Delivery mode
- hybrid source
- Free plan
- Yes source
- GPU requirement
- The GPU backend requires NVIDIA hardware and a CUDA driver. source
- Installation
- PySDM is available from PyPI and can be installed with pip; the documentation also gives instructions for installing from a local Git checkout. source
- Intended users
- The package is intended as a building block for simulation systems involving dispersed phases, with development focused on atmospheric cloud physics applications. source
- Lesson library
- Yes source
- License
- PySDM and PySDM-examples are free/libre open-source software released under the GNU GPL v3 license. source
- Model setups
- Examples cover box, parcel, single-column, and two-dimensional prescribed-flow environments. source
- Notebook access
- Example notebooks have badges for launching them on Google Colab or mybinder.org, and the notebooks themselves are not included in the examples package wheels. source
- Processes
- The example gallery covers condensation, collisional coagulation and breakup, drop freezing, isotopic fractionation, aqueous chemistry, and seeding. source
- Purpose
- PySDM simulates particle populations undergoing diffusional and collisional growth and breakage, with a focus on atmospheric cloud physics and aerosol, cloud, and rain microphysics. source
- Related software
- PyMPDATA is used for advection in the single-column and two-dimensional prescribed-flow examples. source
- Support
- The project recommends GitHub Discussions for help understanding, using, and extending PySDM, and its GitHub issue tracker for reporting incorrect or undocumented behavior. source
Checked against the maker's pages on 11 Oct 2026.