Rank individuals by nondominating sorted front first and by hypervolume contribution or crowding distance second.
Ties are broken randomly by adding random noise of relative magnitude
.Machine$double.eps * 2^10
to points.
overallRankMO(fitness, sorting = "crowding", ref.point)
[integer]
vector of ranks with length ncol(fitness)
, lower ranks are
associated with individuals that tend to dominate more points and that tend to
have larger crowding distance or hypervolume contribution.
[matrix]
fitness matrix, one column per individual.
[character(1)]
one of "domhv"
or "crowding"
(default).
[numeric]
reference point for hypervolume, must be given
if sorting
is "domhv"
.