# Fuzzy programming

(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)

Fuzzy programming deals with mathematical programming problems under non-probabilistic uncertainty. The idea of fuzzy programming was first given by R.E. Bellman and L.A. Zadeh [a1] and then developed by H. Tanaka [a2] and H.-J. Zimmermann [a3]. Those approaches treat soft constraints and vagueness of aspiration levels of objective function values and are called flexible programming. Later, C.V. Negoita, S. Minoiu and E. Stan [a4] dealt with the ambiguity of coefficients in mathematical programming problems by fuzzy set theory (cf. also Non-precise data). This fuzzy programming is called robust programming. D. Dubois [a5] proposed the treatment of fuzzy coefficients in fuzzy programming problems based on possibility theory. Fuzzy programming based on possibility theory is called possibilistic programming. Until now (2000), a lot of fuzzy programming approaches have been proposed, as shown in [a6], [a7], [a8].

The traditional linear programming problem can be written as (cf. also Linear programming) In this problem, the decision variable vector is imposed to satisfy the constraints and satisfies but is never considered as a candidate for the solution even when is extremely small. However, in the real world it frequently occurs that the right-hand values are not determined as rigorous restrictions and sometimes the right-hand values show the aspiration levels of the left-hand function values elicited from the decision maker. In such cases, the constraints should be treated "softly" . Fuzzy programming can treat such soft constraints. Let ; then the constraints of the above linear programming problem can be represented as . In fuzzy programming, replacing by a fuzzy set representing a "set of values approximately smaller than b" , one treats the soft constraints as , by controlling the membership value . Such a fuzzy set is called a fuzzy constraint.

The coefficients and in the above linear programming problem are supposed to be given as real numbers. However, in real world problems one's knowledge is often insufficient to determine the coefficients as real numbers, even when the possible ranges of coefficients is known, such as "ci is about 3" . Such possible ranges can be modelled by fuzzy sets. Let and be fuzzy sets representing possible ranges of and . Then the possible range of is obtained as the fuzzy set defined by the membership function: The possibility and necessity degrees of are defined by:  The possible range of and the possibility and necessity degrees of are defined similarly, where is a fuzzy set.

Soft constraints with ambiguous coefficients are treated as and , where and are constants given by the decision maker. In order to treat the objective function with ambiguous coefficients, a fuzzy set of satisfactory values for the decision maker is established. Such a fuzzy set is called a fuzzy goal. Then the objective function is treated as or . Namely, in the former case, the problem is formulated as It is known that such problems can be reduced to linear programming problems under certain reasonable assumptions that are applicable to real world problems [a8].

How to Cite This Entry:
Fuzzy programming. Encyclopedia of Mathematics. URL: http://encyclopediaofmath.org/index.php?title=Fuzzy_programming&oldid=13383
This article was adapted from an original article by Masahiro Inuiguchi (originator), which appeared in Encyclopedia of Mathematics - ISBN 1402006098. See original article