Scotch Brand 5.1.10 - Manual (Page 62)

pass=nbr
Set the maximum number of optimization passes performed by the algo-
rithm. The Fiduccia-Mattheyses algorithm stops as soon as a pass has
not yielded any improvement of the cost function, or when the maximum
number of passes has been reached. Value −1 stands for an infinite num-
ber of passes, that is, as many as needed by the algorithm to converge.
g Gibbs-Poole-Stockmeyer method. This method has only one parameter.
pass=nbr
Set the number of sweeps performed by the algorithm.
h Greedy-graph-growing method. This method has only one parameter.
pass=nbr
Set the number of runs performed by the algorithm.
m Multi-level method. The parameters of the multi-level method are listed be-
low.
asc=strat
Set the strategy that is used to refine the partitions obtained at ascend-
ing levels of the uncoarsening phase by projection of the bipartitions
computed for coarser graphs. This strategy is not applied to the coarsest
graph, for which only the low strategy is used.
low=strat
Set the strategy that is used to compute the partition of the coarsest
graph, at the lowest level of the coarsening process.
rat=rat
Set the threshold maximum coarsening ratio over which graphs are no
longer coarsened. The ratio of any given coarsening cannot be less that
0.5 (case of a perfect matching), and cannot be greater than 1.0. Coars-
ening stops when either the coarsening ratio is above the maximum coars-
ening ratio, or the graph has fewer vertices than the minimum number
of vertices allowed.
type=type
Set the type of matching that is used to coarsen the graphs. type is h for
heavy-edge matching, or s for scan (first-fit) matching.
vert=nbr
Set the threshold minimum graph size under which graphs are no longer
coarsened. Coarsening stops when either the coarsening ratio is above
the maximum coarsening ratio, or the coarsened graph would have fewer
vertices than the minimum number of vertices allowed.
x Exactifying method.
z Zero method. This method moves all of the vertices to the first part. Its
main use is to stop the bipartitioning process, if some condition is true, when
mapping onto variable-sized architectures (see section 3.1.7).
62