The clustering process starts with an iterative expansion
and inflation operation on the associated matrix
M of G, as described in Section 1.2. The nodes that
are attracted together are clustered into one cluster. At
the same time, all the connected edges between the
nodes are eliminated. When the clustering result at one
level is achieved, G has been transformed into G' with
fewer nodes and fewer edges. The reduced graph G' is
a hierarchy netlist one level higher than the pre-clustering
graph G. The clustering process is iteratively
repeated on the