|
crossover = 0.2 probability = 0.0001 |
|
crossover = 0.67 probability = 0.001 |
|
crossover = 0.99 probability = 0.01 |
|
crossover = 0.2 probability = 0.0001 |
|
crossover = 0.67 probability = 0.001 |
|
crossover = 0.99 probability = 0.01 |
|
crossover = 0.2 probability = 0.0001 |
|
crossover = 0.67 probability = 0.001 |
|
crossover = 0.99 probability = 0.01 |
|
crossover = 0.2 probability = 0.0001 |
|
crossover = 0.67 probability = 0.001 |
|
crossover = 0.99 probability = 0.01 |
The graph above show the evolotion of algorithm for this data. It has data for max., ave. and min.
Two graphs below the path. First one is if you go with order. Second one is our solution.
eil76
There are 76 cities. Optimum path length is 426. 20%(6) of 30 runs passed 75%(473). 18(60%) of 30 runs passed 70%(501).
The worst case is 811 which corresponds to 66%. The best case is 442 which corresponds to 83%. I made 10000 generation.
Population size is 100. Chromosome length is 76, So speed is proportional to 500000*76.
opt 538
13 %75 mark 717
21 %70 mark 768
worst 811
best 647
speed 500000*76
The graph above show the evolotion of algorithm for this data. It has data for max., ave. and min.
Two graphs below the path. First one is if you go with order. Second one is our solution.
lin105
There are 105 cities. Optimum path length is 14379. 3%(6) of 30 runs passed 10%(473). 9(30%) of 30 runs passed 15%(501).
The worst case is 33852 which corresponds to 42. The best case is 21972 which corresponds to 65%. I made 10000 generation.
Population size is 100. Chromosome length is 105, So speed is proportional to 500000*105.
opt 14379
1 %60 mark 23965
9 %50 mark 28758
worst 33852
best 21972
speed 500000*105
The graph above show the evolotion of algorithm for this data. It has data for max., ave. and min.
Two graphs below the path. First one is if you go with order. Second one is our solution.
The algorithm works very well with dataset1. It works ok with second dataset. But, it shows a bad performance with dataset3.
The reason is chromosome length is high. It needs more generation to approach to optimum.
1. http://www.obitko.com/tutorials/genetic-algorithms/crossover-mutation.php
question.cpp
report