Description:
This package contains the datasets, experimental results and reference sets of the paper <i>A benchmark generator for ensemble-based discrete optimization</i>. The following files are included:
<ul>
<li>File <i>test_instances.zip</i>: contains the test instances (TP1 to TP10) used in the experiments. For each test instance, three files are available: TP_k_F_m.txt, which contains the earnings of each decision variable for each scenario; TP_k_F_m_probabilities.txt, which contains the probability of occurrence for each scenario; and TP_k_F_m_weights.txt, which contains the generated weights for each decision variable. The value of k ranges from 1 to 10, and m ranges from 2 to 4. </li>
<li>File <i>experimental_results.zip</i>: contains the experimental results. There is a subfolder for each test instance, and within each subfolder, there are 31 files for each algorithm, containing the non-dominated solutions obtained in each of the 31 independent repetitions.</li>
<li>File <i>reference_sets.zip</i>: contains the reference sets, which are the best-known solutions for each test instance. Each file is structured as follows: each row contains an N-sized decision vector separated by semicolons, and the last values correspond to the objective function values. For example, if it is a bi-objective problem, consider the last two values; if it is a three-objective problem, consider the last three values, and so on.</li>
</ul>