A compromise procedure for the multiple objective linear fractional programming problem (Q761358)

Consider the following multiple objective fractional programming (MOFP) problem \[ V\max \{F(x)=[\ell_ 1(x)/m_ 1(x),\quad \ell_ 2(x)/m_ 2(x),...,\ell_ p(x)/m_ p(x)]^ T: x\in X\} \] where (a) \(p\geq 2\) and natural, (b) \(\ell_ k\), \(m_ k\) \((k=1,2,...,p)\) real-valued and continuous functions on X, (c) \(m_ k(x)>0\) \((k=1,2,...,p)\) for \(x\in X\), (d) X is a nonempty, convex and compact set in \({\mathbb{R}}^ n.\) Solving the MOFP problem means finding the set of all its efficient solutions, i.e. \(X_ F=\{\bar x\in X:\) there is no \(x\in X\) with F(x)\(\geq F(\bar x)\}\). A concept of multiple objective programming (MOP) problem corresponding to MOFP is introduced and some relations between these problems are examined. In particular it is shown that if in the MOFP problem its objective functions have a constant sign on the whole set X, then \(X_ F\subset X_{\psi}\) where \(X_{\psi}\) is the set of efficient solutions of the appropriate corresponding MOP problem. Finally, based on these results, a compromise procedure for MOLFP problems is proposed which is illustrated by a numerical example.