COOM Configuration¶

Description¶

A collection of scalable product configuration problems specified in the COOM language created to be solved with the COOM Suite.
This collection includes benchmark instances, scripts to generate more instances, and everything required to run the benchmarks using the potassco-benchmark-tool.

Usage¶

The COOM Suite can be installed using pip. It is recommended to create a clean python environment beforehand, e.g., using conda.

conda create -n <enc-name> python=3.14
conda activate <env-name>
pip install coomsuite

Instances can then be solved using:

coomsuite solve <instance>

For more information on how to use the COOM Suite check the documentation.

All instances and encodings were tested using clingo >=5.8.0

Scripts to generate new instances can be found in the directory of the specific domain, e.g., benchmarks/restaurant/create_instance.sh

The benchmark-tool can also be installed using pip:

pip install potassco-benchmark-tool

To run the benchmarks, the following steps have to be performed.
Replace laptop with cluster if you want to run the benchmarks on the cluster instead of locally.

Generate the benchmark folder structure and scripts
```
btool gen runscripts/runscript-all-laptop.xml
```

Run the benchmarks

./output/coom-benchmark-project/laptop/start.py

Verify results

btool verify output/coom-benchmark-project/laptop/results

Evaluate and convert the results

btool eval runscripts/runscript-all-laptop.xml | btool conv -o result-all.xlxs

Additional scripts, e.g., for creating plots, can be found in the COOM Suite Benchmarks repository.

Metadata¶

Technical Details¶

Type: Configuration, Multi-shot (wip)
Format: COOM (translations to ASP through COOM Suite)
Tested With: clingo>=5.8

Instances¶

Generation: Each domain has its own script, one real-world anonymized domain (space-collider)
Details:
Multiple domains with about 3-50 instances each (work in progress)

Data & Access Data & Access (Confidentiality)¶

Status: Public
License: MIT
Sensitivity: Mostly artificial instances, real-world problems are already anonymized

Source & Literature¶

Repository/ZIP:
https://github.com/potassco/coom-benchmarks
Reference:
https://arxiv.org/abs/2504.00013

Contact¶

Name: Nicolas Rühling
Email: nruehling@uni-potsdam.de

Miscellaneous¶

Other Notes:
The corresponding clingo encoding can be found in the COOM Suite: https://github.com/potassco/coom-suite/blob/master/src/coomsuite/encodings/encoding-base-clingo.lp

Download¶

In the future you will be able to download the benchmark set here.