Size-Generalization

Experiments accompanying “Accelerating Data-Driven Algorithm Selection for Combinatorial Partitioning Problems.”

1. Quick start

You can set up an isolated environment (Python ≥ 3.10) with:

python -m venv .venv          # or: python3 -m venv .venv
source .venv/bin/activate     # Windows: .venv\Scripts\activate

And install all runtime dependencies with:

pip install -r requirements.txt

Usage

We organize all max-cut algorithms' implementation in the max-cut folder and clustering algorithms' implementation in the clustering folder. We also include in the plots folder the exact plotting code used in our paper.

We include all datasets we used for the clustering experiments under the clustering folder. All max-cut data are generated using NetworkX generators, which we include as part of our dependencies.