“…It is assumed that the objective vectors are arranged in such a way that F (1) represents the vector being the closest to Fc = p -1 kF (k) (i.e., the geometric center of {F (k) }k = 1,…,p). The design xF (1) = U * (F (1) ) is found in a conventional way, i.e., through direct optimization; however, to reduce the cost, FBO is used whenever possible [65]. Having xF (1) , the inverse sensitivity matrix J x (F (1) ) is estimated using the procedure outlined in Section II.C.…”