dda-bench

Benchmark tool for cross comparison of DDA codes.

Install (Users)

pip install dda-bench

Install (Development)

git clone https://github.com/Argencle/dda-bench.git
cd dda-bench
pip install -e .

Important

This package does not ship external DDA solvers. You must provide executables and point your config to valid paths.

Full configuration reference:

• docs/configuration.md

Paper Context

This repository is linked to, and was developed as part of, a benchmark paper.

The dataset used for that paper is available in:

• https://github.com/Argencle/data-dda-benchmark-paper.git

To reproduce the equivalent parameter agreement between codes reported in Appendix C of the paper, patches from that data repository must be applied to ADDA and DDSCAT before running dda-bench.

The corresponding benchmark command set is provided in:

• dda_bench/examples/DDA_commands

Run benchmark/comparison

dda-bench --init

This creates starter files in current directory:

• DDA_commands
• dda_codes.json
• bin/

The starter dda_codes.json points to executables under bin/. --init copies only bin/diel/* and bin/*.dat/bin/*.par support files, not solver executables.

Then run:

dda-bench

dda-bench without options expects:

• ./DDA_commands
• ./dda_codes.json

Override with your own files:

dda-bench --commands /path/to/DDA_commands --code-config /path/to/dda_codes.json

Other options:

dda-bench --output outputs --omp 1 --clean

Output

The command writes:

• dda_bench.log
• dda_bench.errors.log
• per-case results.json under output directory
• summary.csv in output directory