1071 / 1161
Pareto front navigation in multi-criteria optimization?
To be able to navigate through the landscape we need library of plans
“as fine as possible” resolution of the landscape (= many plans)
All ‘corner’ plans should be part of the library with enough data points
along
the Pareto surface (so among all individual Pareto fronts),
so that any interpolated plan should be as close as possible to an
already calculated plan
Pareto front navigation works fine for fluence optimization
as long as the landscape is defined with enough detail