Skip to content

Latest commit

 

History

History
55 lines (33 loc) · 2.02 KB

README.md

File metadata and controls

55 lines (33 loc) · 2.02 KB

DLA Simulation

A basic diffusion-limited aggregation simulation on a square grid.

Installation

The most portable way to run this simulation is Docker. A public image is available on Docker Hub at wbonelli/dla.

To use Python instead of Docker, make a fresh environment with your tool of choice (e.g. venv, Anaconda), then install the packages in requirements.txt (e.g., pip install -r requirements.txt).

Usage

At its simplest, the script can be invoked with:

python dla.py

By default a 100x100 grid will be used, with 100 particles (walkers), and sticking probability unity. Each of these parameters can be provided explicitly as well:

  • --side (-s): The length of a side of the square grid
  • --mass (-m): The number of walkers to attach to the cluster
  • --prob (-p): The probability of a walker to stick to an adjacent cluster cell

For instance, to use a 200x200 grid with 1000 walkers and a 50/50 chance of sticking:

python dla.py -s 200 -m 1000 -p 0.5

Optimizations

None, save for boundary-aware step selection, i.e., not wasting cycles bumping into walls.

Outputs

A GUID is generated each time the script is invoked. This GUID is written with status messages to stdout and forms the stem of the names of the output files produced by each invocation.

When a cluster is completed and a trial ends, the cluster's mass (M) and mean radius of gyration (R) are calculated, printed, and persisted to file:

  • {guid}.png: a plot of the fully grown cluster (with M and R overlaid)
  • {guid}.csv: a 1-line, 2-value CSV file whose first value is M, second is R

These quantities are helpful in estimating fractal dimension.