FFAProject

How to run Hill Climber experiment

Download requirements

• pip install -r requirements.txt

Change Config.py file

Important variables for Hill Climber experiment:

• runs: number of runs
• maxFunctionEvaluations: maximum number of function evaluations per run
• poolProcessed: number of processors used on machine
• resultFolder: folder to save results
• intermediateFolder: folder to save intermediate results during runs.

Important variables for PPA experiment:

• runs: number of runs
• maxFunctionEvaluations: maximum number of function evaluations per run
• poolProcessed: number of processors used on machine
• populationSize: number of individuals selected per generation
• maxOffspring: maximum number of offspring per individual
• relativeMaxSwaps: maximum number of swaps relative to the permutation list size
• resultFolder: folder to save results
• intermediateFolder: folder to save intermediate results during runs.

Change main.py file

• Choose library, the current library (full_full_library.txt) is the one used in the (original) experiment
• Choose the Experiments.runHillClimberComparisonExperiment(library, sortedInstances) function for hill climber experiment.

Now it should run

• When the run needs to be stopped/paused, it will continue once you start again. The intermediateResults folder will be used to resume.
• IMPORTANT NOTE: When you restart a run, empty the files/taken.txt file. (Do not delete it)
• The results for the experiment for EvoStar can be found in files/output/hc2.

Results

• All results should be visible in the resultFolder chosen. To create a single file with all results, run the Util.createAllResultsFile function, with the same resultFolder as argument. An example is the files/output/hc2/allResults.csv file.
• Results of Weise et al.'s original experiment can also be found for comparison: files/weiseResultsOriginal.csv.

How to create figures

• Run makeFigures.py
• You can also choose to run other instances, by changing the instances variable. I hope it works with more or less than 6 instances, but not sure.

How to create summary table

• Run makeTables.py
• This will create three files in main directory: own_results.csv, weise_results and comparison_results.csv
• The first is a summary of our own results. For this the files/output/hc2/allResults.csv file is used.
• The second is a summary of Weise et al.'s results. For this the files/weiseResultsOriginal.csv file is used.
• The third is a comparison with Weise et al.'s original results. For this the same files are used.
• Both result files can be found in files.