Question:
You need to carry out a short investigation into the effect of different starting solutions and tabu list lengths for runs of a fixed number of iterations when using Tabu Search for a TSP, using the software provided in the course (Which is in additional file ) Zip

Before you start I suggest that you make some practice runs of Tabu Search using the software, so as to become thoroughly familiar with it.  Then you will need to make several runs “for real”.

You should conduct your experiments for one set of data: use city125.txt.

Use the following settings:
Delay factor: Use a value of 1 or more in order to observe how a search is progressing using the map. When you have decided what runs you want to record, set the delay to zero and the program will run more quickly. Results are updated in the output window as normal, but the map only shows the initial and best solutions.

Initial solution: Use Successive Inclusion or Random.

Starting city:   For every run the starting city should be a number between 00 and 99 formed from any two consecutive digits of the last 5 digits of your library card number.For example, if your library card number is 32012345, you could use starting cities 12, 23, 34 or 45. [If repetitions of digits mean that this process does not give four different starting cities, then you can get more starting cities by choosing any you wish between 100 and 125].
( in this case: 4 starting cities are  27,73,34,48)

Neighbourhood option: reverse a route section

Improvement option: Tabu Search

Text period: The Output window displays information every time it finds a new best solution plus the current solution every n iterations, where n is the value of the text period you have input. You are free to choose, but n = 50 may be appropriate for your experiments.

Tabu Search Settings – No. Moves: Fix at 500
Tabu Search Settings – List Length: Your choice

You will need to make several runs of the program, but it is quick, so this should not take very long.  You will need to save the results from each run from the Output window in a uniquely named file.  If you use the same name the new results will overwrite the old.

Please check the starting city everytime in output

a) With “Successive Inclusion” as the initial solution, choose 4 different values for the list length and record the solutions to the TSP from 4 different starting cities. Repeat with “Random” as the initial solution, using the same list lengths and starting cities.

Summarise your results in a table and a suitable chart or graph. Comment on the results and give any reasons why you think the different input settings have affected the results in the ways observed.
(20% marks)

b) From your results, comment on the relative importance of the following decisions in specifying the input settings in order to get a good solution from the Tabu Search algorithm to the particular TSP problem: i) the initial city used, ii) the initial solution method, iii) the length of the Tabu List.
(20% marks)

c) Recommend the best set of input settings to use for this problem. Use the output from a single run with your recommended set of input settings to show how the current value of the total distance depends on the iteration number and at what iteration the best solution is found.

Place an order with similar or related instructions NOW