EAlib

Members：

Name	Students numbers	email addresse
Zhang Kaiyan	1160300607	[email protected]
Wu Jiaming	1160300612	[email protected]
Li Yangfan	1160300627	[email protected]
Wang Zhongyuan	1160300629	[email protected]
Sun Runxin	1160300911	[email protected]

Video Here:

• OneDriver
• YouTube

Get Started

numpy = 1.16.4 is required.

Just One Line Code

$ python setup.py install && python test/EAlibTest.py -filename=tsp/eil51.tsp && python test/visual.py -filename=eil51.tsp

---

You can also clone the code from GitHub, and run the commands as follows in terminal:

$ git clone https://github.com/iseesaw/EAlib
$ cd EAlib
$ python setup.py install

Then you can run our best Evolutionary Computation algorithm on the tsp problems under the tsp directory:

$ python test/EAlibTest.py -filename=tsp/eil51.tsp

The results are saved in ouput/ei151.tsp.json.

You can visual the evolutionary process:

$ python test/visual.py -filename=ei151.tsp

Tutorials

• Usage

  You can use our python library, EAlib:

  from EAlib.ea import Naive_Ea
  Naive_Ea(filename="tsp/att48.tsp")

• Run Scripts

  • Grid Search, which is used to find the best topk combinations.

    bash search.sh

    This script is running 4 x 4 x 3 = 48 algorithms on nine datasets(Not including pr2392).

    You can get the results in output/grid_search

  • Run top3 best algorithms 10 datasets with 4 population sizes.

    bash run.sh selection_func crossover_func mutation_func

    You can find the functions' name in Mapping parameters to functions.

    And the results are save in output/{filename}.{selection_func}.{crossover_func}.{mutation_func}.{unit_num}.{max_gen}.json.

    At the end, you can get 4 x 10 = 40 results on the algorithm.

  • Visualization

    python test\visual.py -filename=att48.tsp.basic_selection.basic_crossover.basic_mutation.50.100

    parameter	type	default	usage
    filename	str	None	output result file name
    tsp	str	None (default get from filename)	tsp problem file
    pause	float	0.5	time to pause

  • **Run Best algorithm on ten dataset for N iterations **

    bash run_best.sh

Appendix

Here comes the information about EAlibTest.py.

• Arguments

parameter	type	default	help
-filename	str	tsp/att48.tsp	tsp problem file (under the tsp dir)
--selection	str	basic	[Optional] basic, elitism, tournament, fitnetss
--crossover	str	basic	[Optional] basic, order, cycle, edge
--mutation	str	basic	[Optional] basic, insert, scramble, inversion, swap
--unit_num	int	20	unit num of the population
--max_gen	int	100	max generations
--prob_c	float	0.5	probability of crossover
--prob_m	float	0.3	probability of mutation
--print_every	int	50	how many times to print the best one
--output_dir	str	output	output the best one record

• Mapping parameter to functions

      selections = {
          "basic": basic_selection,
          "rank": rank_based,
          "tournament": tournament_selection,
          "fitnetss": fitnetss_proporitional
      }
  
      mutations = {
          "basic": basic_mutation,
          "insert": insert_mutation,
          "scramble": scramble_mutation,
          "inversion": inversion_mutation,
          "swap": swap_mutation
      }
  
      crossovers = {
          "basic": basic_crossover,
          "order": Order_Crossover,
          "cycle": Cy_cle_Crossover,
          "edge": Edge_Recombination,
          "pmx": PMX_Crossover
      }

Files Structure

EAlib
|-- EAlib # Our Core Algorithms
|   |-- __init__.py
|   |-- basic # basic classes
|   |   |-- __init__.py
|   |   |-- individual.py
|   |   |-- population.py
|   |   `-- tspproblem.py
|   |-- ea # main function
|   |   |-- __init__.py
|   |   `-- naive_ea.py
|   |-- log # log
|   |   |-- __init__.py 
|   |   |-- log.conf
|   |   `-- log.py
|   |-- operators
|   |   |-- __init__.py
|   |   |-- crossover.py
|   |   |-- mutation.py
|   |   `-- selection.py
|   `-- utils
|       |-- __init__.py
|       `-- dataloader.py
|-- LICENSE
|-- README.md
|-- log.log # running log
|-- history.log # more log to show in the history
|-- output # output result files
|   |-- best_algorithm.json  #### the log for the best algorithm ###
|   |-- eil51.tsp
|   |-- eil51.tsp.json # output file while you run the first command above
|   `-- top5_algorithms.json ### the log for results of 3（5） algorithms ###
|-- run_best.sh  # script
|-- run_grid_search.sh # script 
|-- run_top3.sh # script
|-- setup.py
|-- test  # experiments
|   |-- EAlibTest.py
|   |-- TSPProblemTest.py
|   |-- analyze.py
|   |-- dataloader.log
|   |-- dataloadertest.py
|   `-- visual.py
`-- tsp # tsp problems