Similarity-based Portfolio Construction for Black-box Optimization - Reproducibility

# Reproducibility files for 'Similarity-based Portfolio Construction for Black-box Optimization' All files required for reproducing the results shown in the paper can be found in 'reproducibility.zip'. # Reproducibility instructions Follow these steps to generate plots and tables: 1. **Prepare the environment:** - Make sure you have all necessary dependencies installed. - If a `requirements.txt` file is present, install the required packages by running: ```bash pip install -r requirements.txt ``` 2. **Generate initial data:** - Run the following script to generate the initial data: ```bash python run_algos_many_affine.py ``` 3. **Compute ELA features:** - Calculate Exploratory Landscape Analysis (ELA) features by executing: ```bash python compute_ela.py ``` 4. **Process features and cosine similarity:** - Open and run the notebook `process_features_cosine_sim.ipynb` to process the features and compute cosine similarity. 5. **Compute SBP (Surrogate Benchmark Problems):** - Run the following commands to compute SBP for different dimensions: ```bash python compute_SBP.py --dim=2 --k=10 --reps=50 python compute_SBP.py --dim=5 --k=10 --reps=50 python compute_SBP.py --dim=10 --k=10 --reps=50 ``` 6. **Run experiments:** - Start the experiments for different configurations by running: ```bash python run_experiments.py --dim=2 --sbs=modcma --feat_type=ela --tops=20 python run_experiments.py --dim=5 --sbs=modcma --feat_type=ela --tops=20 python run_experiments.py --dim=10 --sbs=modcma --feat_type=ela --tops=20 python run_experiments.py --dim=2 --sbs=modcma --feat_type=perf --tops=20 python run_experiments.py --dim=5 --sbs=modcma --feat_type=perf --tops=20 python run_experiments.py --dim=10 --sbs=modcma --feat_type=perf --tops=20 ``` 7. **Generate plots and tables:** - Use the notebook `process_plots_tables.ipynb` to generate the final plots and tables.