3.2 k-cluster problem

Given a graph G = (V,E) the k-cluster problem consists of determining a subset S V of k vertices such that the sum of the weights of the edges between vertices in S is maximized.

Letting n = |V | denote the number of vertices, and wij denote the edge weight for ij E and wij = 0 for ij∕∈E, the problem can be modeled as the following 0-1 quadratic problem:

maximize 1
2xTWx
subject to ixi = k, x ∈{0,1}n.
where W = (wij)ij is the weighted adjacency matrix of the graph G.
  3.2.1 k-cluster heuristic
  3.2.2 k-cluster instances and conversion