Welcome to the repository for Answer Set Programming (ASP) benchmarks, curated by the Potassco group.
Our goal is to provide solver developers and researchers with a diverse, high-quality suite of challenges—from classic combinatorial problems to modern industrial applications.
We believe that great benchmarks often go unshared because the "cost" of formal documentation is too high. We want to change that. We value your contribution even if it is incomplete. If you have a set of instances and a basic encoding but haven't written a paper or a generator script yet—submit it anyway! We can help refine the metadata later.
The submission process takes less than 5 minutes and happens entirely through GitHub:
- Open an Issue: Go to the [Issues] tab and click "New Issue".
- Choose the Template: Select "New Benchmark Submission".
- Fill what you can: Complete the fields you have information for. If you don't have the information (yet), just leave them blank!
- Provide the Data:
- Link to a GitHub repo or DOI.
- or simply attach a (not tooo big)
.zipfile directly to the issue.
We recognize that industrial or real-world data is often sensitive. We support three levels of access:
| Access Level | Description |
|---|---|
| Public | The encoding and instances are hosted here or linked for everyone. |
| Restricted | We host the encoding and metadata only. Users must contact you for the instances. |
| Embargoed | We index the benchmark but keep the details hidden until a specified date (e.g., after a paper is published). |
While we primarily focus on ASP-Core-2 and clingo compatible encodings, we welcome benchmarks in aspif, DLV, or even DIMACS formats.
Need help? If you have a problem you want to share but aren't sure how to format it for the collection, just [open a general issue] and we will be happy to assist you with the "onboarding" of your data.
The documentation can be accessed here or build and hosted locally using zensical.
pip install zensical
zensical serve