ODP-IP
Exact Algorithms for "Complete Set Partitioning" (i.e., for "Coalition Structure Generation"). This project was copied from https://github.com/trahwan/ODP-IP and then fixed to work with maven
- Import as maven project (eg in Eclipse)
- Run solveParticularProblem/SolveParticularProblem
You will get a GUI that is set up already to run the example file that is explained below. Press Run. The GUI will show the optimal solution.
Details
-
This code contains implementations of the following algorithms: IP, ODP-IP, DP, ODP, and IDP (which is the size version of ODP)
-
For more on the optimization problem itself, or the algorithms, see: http://www.cs.ox.ac.uk/files/5616/CS-RR-13-09.pdf
-
The two main, executable java classes are the following (both come with a graphical user interface (GUI)):
-
SolveRandomProblems.java: This allows the user to experiment with the different algorithms with randomly-generated problem instances.
-
SolverParticularProblem.java: This allows the user to solver a particular problem, where the values of different coalitions (i.e., subsets) are stored in a file. Below are more details on the format of the file.
-
-
file format:
- if you want to read the subset values from a file, it should be stored in a text file, where every line contains the value of one subset. Here is an example for the contnt of such a file, given 4 agents (i.e., 4 elements):
0
30
40
50
25
60
55
90
45
80
70
120
80
100
115
140
In a file containing values just like the above, the value in line x represents the value of the subset that is represented by the true bits in the number (x-1).
For example, assume we have the following four agents (or elements): {a1,a2,a3,a4}. The value at line 6 is 60 (see above). This represents the value of the subset that is represented by the true bits in number (6-1), which is: 0101. This subset is: {a1,a3}. Based on this, the above list of values represent the following problem:
0 is the value of the subset represented by 0 = 0000 which is {}
30 is the value of the subset represented by 1 = 0001 which is {a1}
40 is the value of the subset represented by 2 = 0010 which is {a2}
50 is the value of the subset represented by 3 = 0011 which is {a1,a2}
25 is the value of the subset represented by 4 = 0100 which is {a3}
60 is the value of the subset represented by 5 = 0101 which is {a1,a3}
55 is the value of the subset represented by 6 = 0110 which is {a2,a3}
90 is the value of the subset represented by 7 = 0111 which is {a1,a2,a3}
45 is the value of the subset represented by 8 = 1000 which is {a4}
80 is the value of the subset represented by 9 = 1001 which is {a1,a4}
70 is the value of the subset represented by 10 = 1010 which is {a2,a4}
120 is the value of the subset represented by 11 = 1011 which is {a1,a2,a4}
80 is the value of the subset represented by 12 = 1100 which is {a3,a4}
100 is the value of the subset represented by 13 = 1101 which is {a1,a3,a4}
115 is the value of the subset represented by 14 = 1110 which is {a2,a3,a4}
140 is the value of the subset represented by 15 = 1111 which is {a1,a2,a3,a4}
The optimal solution to this problem is the following partition: {{a1}, {a2}, {a3,a4}}. The value of this partition is: v({a1})+v({a2})+v({a3,a4}) = 150
License: All source files are made available under the terms of the GNU General Public License (GPL).