ELSEVIER
Fuzzy Sets and Systems 78 (1996) 231-241
sets and systems
Possible and necessary efficiency in possibilistic multiobjective
linear programming problems and possible efficiency test
Masahiro Inuiguchi*, Masatoshi Sakawa
Department of lndustrial and Systems Engineering, Faculty of Engineering, Hiroshima University, 4-1, Kagamivama l-chome,
Higashi-Hiroshima, 724, Japan
Abstract
In this paper, the concept of efficient solutions to the conventional multiobjective linear programming problems is
extended to the fuzzy (possibilistic) coefficients case. Two kinds of efficient solution sets, i.e., a set of possibly efficient
solutions and a set of necessarily efficient solutions, are defined as fuzzy sets whose membership grades represent the
possibility and necessity degrees to which the solution is efficient. A test to check the possible efficiency is discussed when
a feasible solution is given. To do this, we first consider the interval case, where all fuzzy (possibilistic) coefficients
degenerate into interval coefficients. In this case, a set of possibly efficient solutions degenerates into a usual (crisp) set.
A necessary and sufficient condition of the possible efficiency for the interval case is presented. This condition shows that
the possible efficiency is checked by solving a system of linear inequalities. Extending this result to the fuzzy (possibilistic)
case, the degree of possibility efficiency is obtained by solving a nonlinear programming problem. The nonlinear
programming problem is solved by the simplex and bisection methods.
Keywords: Multiobjective linear programming; Efficiency; Possibility; Necessity; Simplex method
1. Introduction
We usually face some difficulties when a real
world problem is formulated to a mathematical
programming problem. One of the difficulties is
caused by the uncertainty in knowledge, informa-
tion and/or decision maker's preference; for
example, it is difficult to identify the coefficient
under the ambiguous information such as "the
* Corresponding author.
profit rate will be almost 10 ($/min)" and to fix the
target value under the vague aspiration such as "we
want the profit substantially larger than 10 mil-
lion". If such parameters are determined as real
numbers appropriately, the formulated problem of-
ten has no feasible solution or yields an improper
solution. In these cases, we have to investigate the
causes, change the determined real numbers, solve
the reformulated problem and repeat this proced-
ure until we obtain the satisficing solution. Thus,
this requires a formidable effort and plenty of time.
Fuzzy programming is proposed in order to deal
with such uncertainties [2,4,5, 10, 12, 14, 15]. In
0165-0114/96/$15.00 © 1996 - Elsevier Science B.V. All rights reserved
SSDI 0165-01 14(95)00169-7