quickstart.rst

Python quickstart

Install DEME in an environment with a compatible NVIDIA driver and CUDA runtime, then verify the package and version:

python -c "import deme; print(deme.__version__)"

The following complete example creates a material, a bounded domain, and one spherical clump, then advances the simulation:

.. literalinclude:: examples/sphere_drop.py
   :language: python
   :linenos:

Run it with:

python docs/python/examples/sphere_drop.py

The solver constructor initializes CUDA worker resources, so even importing successfully is not sufficient to run a simulation without a visible, supported NVIDIA GPU. See CUDA device selection when the process can see more than one GPU.

The example uses DoDynamicsThenSync because the position is read immediately afterward. For longer simulations, asynchronous DoDynamics calls can overlap host work; synchronize before reading results or exiting.

More Python demos

The repository's python/demos directory includes ports of DEMdemo_SingleSphereCollide, DEMdemo_BallDrop, and DEMdemo_Centrifuge. They demonstrate cohesive contact, runtime particle insertion, a meshed projectile, and a rotating analytical drum:

python python/demos/single_sphere_collide.py --smoke-test
python python/demos/ball_drop.py --smoke-test
python python/demos/centrifuge.py --smoke-test

Run from a checkout with the Python package installed. Each script supports --help, --device, --duration, and --output-dir. See python/demos/README.md for full runs, output visualization, and differences from the larger C++ examples. Smoke runs still require a GPU and first-time kernel compilation.

Where to go next

• Solver lifecycle explains which operations belong before and after Initialize().
• Retrieving simulation data covers host-returning tracker methods and direct retrieval into CUDA arrays.
• Python API overview maps common tasks to the Python objects and methods that implement them.