A unified approach for characterizing Pareto optimal solutions of multiobjective optimization problems: The hyperplane method (Q1123814): Difference between revisions

A scalarization method for obtaining Pareto optimal solutions to a multiobjective optimization problem (MOP) is proposed in the paper. MOP consists in minimizing a k-dimensional vector function f on a set X. The scalarization problem - called 'hyperplane problem' by the authors - is introduced as follows: \[ HP(Q,t_ k):\quad \min_{x\in X}z\quad s.t.\quad QF(f(x))\leq D(z,t_ k) \] where \(z\in R^ 1\); \(t_ k=(t_ 1,...,t_{k-1})\in R^{k-1}\); \(t_ i>0\), \(\sum^{k}_{1}t_ i=1\); Q is a \(k\times k\) matrix whose elements \(q_{ij}\geq 0\) and at least one element in any row is positive; F(\(\cdot)\) and D(\(\cdot)\) are k- dimensional vector functions whose components are continuously differentiable; in addition, the components of F(\(\cdot)\) are strictly monotone increasing. The relationships between an optimal solution of \(HP(I,t_ k)\) and a strong Pareto optimal solution of the MOP are characterized. The relations between the \(HP(Q,t_ k)\) and well-known scalarizing problems such as the weighting problem, the weighting minimax problem, the weighting \(\ell_ p\)-norm problem and so on are considered.