Scaling of the objective function values

Source: R/scale_objectives.R

Performs scaling of the objective function values for the MOEADr package

scale_objectives(Y, Yt, scaling, eps = 1e-16, ...)

Arguments

Y

matrix of objective function values for the incumbent solutions

Yt

matrix of objective function values for the candidate solutions

scaling

list containing the scaling parameters (see moead() for details).

eps

tolerance value for avoiding divisions by zero.

...

other parameters (included for compatibility with generic call)

Value

List object containing scaled objective function value matrices Y and Yt, as well as estimates of the "ideal" point minP`` and "nadir" point maxP`.

Details

This routine scales the matrices of objective function values for the current (Yt) and candidate (Y) solutions. The following methods are currently available:

  • scaling$name = "none": no scaling

  • scaling$name = "simple": simple linear scaling between estimated ideal and nadir points, calculated from the available points in Y and Yt at each iteration.

References

F. Campelo, L.S. Batista, C. Aranha (2020): The MOEADr Package: A Component-Based Framework for Multiobjective Evolutionary Algorithms Based on Decomposition. Journal of Statistical Software doi:10.18637/jss.v092.i06