MTSP-NAGA-II

This is the code of course project in Advanced Artificial Intelligence. This project uses NSGA-II to solve the problem of Multiple Traveling Salesmen Problem. The baseline code is GA-for-mTSP and we finetuned its parameters to make it more suitable for our experiment. The population select algorithm of our code was mainly based on the NSGA-II and the enhanced GA algorithm is mainly based on the arifield’s work. For more information, please visit the repository.

Contribution

Instruction manual

Repeate our experiment result

1. Change the dir to baseline/code/, and run the shell repeat_test.sh, and you will get the baseline result saved in the baseline_run_data.json
   
   
2. Change the dir ro mtsp_nsga_ii/code/, and run the python file repeat_test.py, and you will get the improved GA result saved in the ours_run_data.json
   
   
3. Move the two .json file you get in the previous step to summary_figure/ and run polt_figure.pyand summary.py , and you will get the result figure and table of this two algorithm we represent in our report.

Note: It needs a lot of time to run the experiment, as we repeat 30 times in each dataset.

Run the enhanced GA algorithm

Change the dir to mtsp_nsga_ii/code/

 >python main.py 
 usage main.py --problem [--traveller] [--population] [--generations] [--mutation]
 optional arguments:
         --problem      problem name
         --traveller    number of travellers,default 5
         --population   number of population,default 100
         --generations  number of generations, default 200
         --mutation     nutation rate, default 0.2