PEACE logo

PEACE (Parameterized and Exact Algorithms for Cluster Editing) is a project for clustering experimental data using a weighted graph derived from pairwise similarity information.

The underlying mathematical model is the NP-complete Cluster Editing problem. Similar models are used in the clustering algorithms CLICK and CAST. Advantages are that you do not have to specify cluster centers, the number of clusters need not be given in advance, and solutions are definite. We compute exact solutions for this model, mainly with fixed-parameter algorithms. These algorithms employ both parameter-independent and parameter-dependent reduction rules to cut down the size of the instance initially and inside the search tree. PEACE is implemented in C++. You can download the source code here.

The Integer Linear Programming code has been provided by Gunnar Klau at the Centrum Wiskunde & Informatica, Amsterdam. The Linux executable can be downloaded here. You need CPLEX to run this code. See here for further information.

See the Experimental Data page if you want to download or cite experimental data concerning PEACE.