Skip to content

Advertisement

Open Access

On sequences of fuzzy sets and fuzzy set-valued mappings

Fixed Point Theory and Applications20132013:327

https://doi.org/10.1186/1687-1812-2013-327

Received: 19 June 2013

Accepted: 6 November 2013

Published: 2 December 2013

Abstract

Based on level sets of fuzzy sets, we propose definitions of limits of sequences of fuzzy sets, and limits and derivatives of fuzzy set-valued mappings. Then, their properties are derived. Limits of sequences of fuzzy sets, and limits and derivatives of fuzzy set-valued mappings are fuzzified ones of them for crisp sets, where ‘crisp’ means ‘non-fuzzy’.

MSC:03E72, 90C70.

Keywords

sequence of fuzzy setsfuzzy set-valued mappingfuzzy set-valued analysis

1 Introduction and preliminaries

The usefulness and importance of limits of sequences of crisp sets, and limits (continuity) and derivatives of crisp set-valued mappings have been recognized in many areas, for example, variational analysis, set-valued optimization, stability theory, sensitivity analysis, etc. For details, see, for example, [16]. The concept of limits of sequences of crisp sets is interesting and important for itself, and it is necessary to introduce the concepts of limits and derivatives of crisp set-valued mappings. Typical and important applications of them are (i) set-valued optimization and (ii) stability theory and sensitivity analysis for mathematical models. For the case (ii), consider the following system. Some mathematical model outputs the set of optimal values W ( u ) R and the set of optimal solutions S ( u ) R n for a given input parameter u R m . Then W and S are crisp set-valued mappings. Stability theory deals with the continuity of W and S . Sensitivity analysis deals with the derivative of W .

In this article, limits of sequences of fuzzy sets, and limits and derivatives of fuzzy set-valued mappings are considered. They are generalizations of them for crisp sets. The aim of this article is to propose those concepts and to investigate their properties systematically.

Some research works deal with limits of sequences of fuzzy numbers or fuzzy sets with bounded supports [79], while few research works deal with limits of sequences of fuzzy sets. In addition, some research works deal with limits (continuity) and derivatives of fuzzy number or fuzzy set with bounded support-valued mappings [7, 10, 11], while few research works deal with limits (continuity) and derivatives of fuzzy set-valued mappings. Furthermore, their approaches need some assumptions that level sets of fuzzy sets are nonempty and compact. Our new approach in this article, however, does not need those assumptions. Limits of sequences of fuzzy sets, and limits and derivatives of fuzzy set-valued mappings can be expected to be useful and important for (i) fuzzy set-valued optimization, (ii) stability theory and sensitivity analysis for fuzzy mathematical models, etc. For the case (ii), consider the following system. Some fuzzy mathematical model outputs the fuzzy set of optimal values on R , W ˜ ( u ) , and the set of optimal solutions S ( u ) R n for a given input parameter u R m . Then W ˜ is a fuzzy set-valued mapping, and S is a crisp set-valued mapping. Then we will be able to deal with the continuity of W ˜ and S , and the derivative of W ˜ . This means that our proposing approaches give useful tools for stability theory and sensitivity analysis in fuzzy set theory.

We use the following notations.

For a , b R { , } , we set [ a , b ] = { x R : a x b } , [ a , b [ = { x R : a x < b } , ] a , b ] = { x R : a < x b } , and ] a , b [ = { x R : a < x < b } .

For a λ [ 0 , 1 ] , λ Λ , we define λ Λ a λ = inf λ Λ a λ and λ Λ a λ = sup λ Λ a λ , where λ Λ a λ = inf λ Λ a λ = 1 and λ Λ a λ = sup λ Λ a λ = 0 if Λ = .

Let be the set of all natural numbers, and we set
N = { N N : N N  is finite } = { subsequences of  N  containing all  k  beyond some  k 0 } , N = { N N : N  is infinite } = { all subsequences of  N } .

A subsequence of a sequence { x k } k N is represented as { x k } k N for some N N . We write lim k , lim k , or lim k N when k in , but lim k N or lim k N in the case of the convergence of { x k } k N for some N in N or N .

For a set C R n , let cl ( C ) be the closure of C.

Let C ( R n ) , K ( R n ) , and CK ( R n ) be sets of all closed, convex, and closed convex subsets of R n , respectively.

1.1 Limits of sequences of sets

First, we slightly review the definitions of limits of sequences of sets and their properties.

Definition 1.1 (Definition 4.1 in [3])

For a sequence { C k } k N of subsets of R n , its lower limit is defined as the set
lim inf k C k = { x R n : N N , x k C k ( k N )  with  x k N x } ,
and its upper limit is defined as the set
lim sup k C k = { x R n : N N , x k C k ( k N )  with  x k N x } .
The limit of { C k } k N is said to exist if lim inf k C k = lim sup k C k , and its limit is defined as the set
lim k C k = lim sup k C k = lim inf k C k .

Example 1.1 For a , b , c , d R with a < b < c < d , set A = [ a , c ] and B = [ b , d ] . For each k N , let C k = A if k is odd, and let C k = B if k is even. In this case, we have lim inf k C k = A B = [ b , c ] and lim sup k C k = A B = [ a , d ] .

Proposition 1.1 (Exercise 4.2(b) in [3])

For a sequence { C k } k N of subsets of R n , lim inf k C k = N N cl ( k N C k ) and lim sup k C k = N N cl ( k N C k ) .

Proposition 1.2 (Exercise 4.3 in [3])

For sequences { C k } k N , { C k 1 } k N , { C k 2 } k N of subsets of R n and C R n , the following statements hold.
  1. (i)

    If C k (which means that C k C k + 1 ), then lim k C k = cl ( k N C k ) .

     
  2. (ii)

    If C k (which means that C k C k + 1 ), then lim k C k = k N cl ( C k ) .

     
  3. (iii)

    If C k 1 C k C k 2 , k N and C k 1 C , C k 2 C , then C k C .

     

Proposition 1.3 (Proposition 4.4 in [3])

For sequences { C k } k N , { D k } k N of subsets of R n and C R n , the following statements hold.
  1. (i)

    lim inf k C k , lim sup k C k C ( R n ) .

     
  2. (ii)

    If cl ( C k ) = cl ( D k ) , k N , then lim inf k C k = lim inf k D k and lim sup k C k = lim sup k D k .

     
  3. (iii)

    If C k = C , k N , then lim k C k = cl ( C ) .

     

Definition 1.2 (Example 4.13 in [3])

For nonempty closed sets C , D R n , the Hausdorff distance between C and D is defined as
ρ ( C , D ) = sup x R n | d C ( x ) d D ( x ) | ,

where d C ( x ) = inf y C y x , d D ( x ) = inf y D y x and is the Euclidean norm. A sequence { C k } k N of nonempty closed subsets of R n is said to converge to a nonempty closed set C R n with respect to ρ if ρ ( C k , C ) 0 .

Proposition 1.4 (Example 4.13 in [3])

Let X R n be a bounded set, and let C k X , k N and C X be nonempty closed sets. Then { C k } k N converges to C in the sense of Definition  1.1 if and only if { C k } k N converges to C with respect to ρ.

Proposition 1.5 (Proposition 4.15 in [3])

If { C k } k N K ( R n ) , then lim inf k C k K ( R n ) .

From Definition 1.1, the following Proposition 1.6 can be obtained.

Proposition 1.6 Let C k R n , k N , and let C = lim k C k . In addition, let N N . Then C = lim k N C k .

1.2 Limits of set-valued mappings

Next, we give the definitions of limits of set-valued mappings and their properties.

A mapping F such that F ( x ) R m for each x R n is called a set-valued mapping from R n to R m , and it is denoted by F : R n R m . The set-valued mapping F is said to be closed-valued, convex-valued, or closed convex-valued if F ( x ) C ( R m ) , F ( x ) K ( R m ) , or F ( x ) CK ( R m ) for any x R n , respectively.

Definition 1.3 (See p.152 in [3])

Let F : R n R m , and let x ¯ R n . The lower limit of F when x x ¯ is defined as the set
lim inf x x ¯ F ( x ) = x k x ¯ lim inf k F ( x k ) ,
and the upper limit of F when x x ¯ is defined as the set
lim sup x x ¯ F ( x ) = x k x ¯ lim sup k F ( x k ) ,
where x k x ¯ and x k x ¯ mean the intersection and the union with respect to any sequence { x k } k N R n such that x k x ¯ , respectively. The limit of F when x x ¯ is said to exist if lim inf x x ¯ F ( x ) = lim sup x x ¯ F ( x ) , and its limit is defined as the set
lim x x ¯ F ( x ) = lim inf x x ¯ F ( x ) = lim sup x x ¯ F ( x ) .
Example 1.2 For a , b , c , d R with a < b < c < d , set A = [ b , c ] and B = [ a , d ] . Let F , G : R R be set-valued mappings defined as
F ( x ) = { A if  x 0 , B if  x > 0 , G ( x ) = { A if  x < 0 , B if  x 0

for each x R . In this case, we have lim inf x 0 F ( x ) = lim inf x 0 G ( x ) = A = [ b , c ] and lim sup x 0 F ( x ) = lim sup x 0 G ( x ) = B = [ a , d ] .

Definition 1.4 (Definition 5.4 in [3])

Let F : R n R m , and let x ¯ R n . The set-valued mapping F is said to be lower semicontinuous at x ¯ if
lim inf x x ¯ F ( x ) F ( x ¯ ) ,
and F is said to be upper semicontinuous at x ¯ if
lim sup x x ¯ F ( x ) F ( x ¯ ) .
The set-valued mapping F is said to be continuous at x ¯ if F is both lower and upper semicontinuous at x ¯ , that is,
lim x x ¯ F ( x ) = F ( x ¯ ) .

Example 1.3 Consider the set-valued mappings F and G defined in Example 1.2. Since lim inf x 0 F ( x ) = [ b , c ] [ b , c ] = F ( 0 ) and lim sup x 0 F ( x ) = [ a , d ] [ b , c ] = F ( 0 ) , F is lower semicontinuous at 0 but not upper semicontinuous at 0. Since lim sup x 0 G ( x ) = [ a , d ] [ a , d ] = G ( 0 ) and lim inf x 0 G ( x ) = [ b , c ] [ a , d ] = G ( 0 ) , G is upper semicontinuous at 0 but not lower semicontinuous at 0.

From Definition 1.3, the following Proposition 1.7 can be obtained.

Proposition 1.7 Let F : R n R m , and let x ¯ R n . Then lim x x ¯ F ( x ) = F ( x ¯ ) if and only if lim k F ( x k ) = F ( x ¯ ) for any sequence { x k } k N R n such that x k x ¯ .

1.3 Derivatives of set-valued mappings

In this subsection, we give the definition of derivatives of set-valued mappings and their properties.

A set C R n is called a cone if 0 C and λ x C for any x C and any λ 0 .

Definition 1.5 (Definition 3.4 in [2])

Let S R n , and let x 0 S . A vector d R n is called a tangent vector of S at x 0 if there exists a sequence { x k } k N S , which converges to x 0 , and a sequence { t k } k N R of positive real numbers such that
lim k t k ( x k x 0 ) = d .

The set of all tangent vectors of S at x 0 is called the tangent cone of S at x 0 , and it is denoted by T ( S ; x 0 ) .

A tangent cone in Definition 1.5 is also called a contingent cone. For details of tangent cones, see, for example, [1, 12].

Example 1.4 Let α ] 0 , 1 ] , and let S = { ( x , y ) R 2 : min { α ( x 3 x ) , ( 2 α ) ( x 3 x ) } y max { α ( x 3 x ) , ( 2 α ) ( x 3 x ) } } . In addition, let x 0 = ( 0 , 0 ) S . In this case, we have T ( S ; x 0 ) = { ( x , y ) R 2 : min { α x , ( 2 α ) x } y max { α x , ( 2 α ) x } } .

Proposition 1.8 (Theorem 3.7 in [2])

Let S R n , and let x 0 S . Then T ( S ; x 0 ) is a closed cone. Furthermore, if S is convex, then T ( S ; x 0 ) is also convex.

Proposition 1.9 (Theorem 3.8 in [2])

Let S , Q R n with S Q , and let x 0 S . Then T ( S ; x 0 ) T ( Q ; x 0 ) .

For a set-valued mapping F : R n R m , the set
Graph ( F ) = { ( x , y ) R n × R m : y F ( x ) }

is called the graph of F.

Definition 1.6 (Definition 4.8 in [2], Definition 5.1.1 in [1])

Let F : R n R m , and let ( x 0 , y 0 ) Graph ( F ) . Then the set-valued mapping D F ( x 0 , y 0 ) : R n R m defined as
D F ( x 0 , y 0 ) ( u ) = { v R m : ( u , v ) T ( Graph ( F ) ; ( x 0 , y 0 ) ) }

for each u R n is called the contingent derivative of F at ( x 0 , y 0 ) .

From Definition 1.6, it can be seen that
Graph ( D F ( x 0 , y 0 ) ) = T ( Graph ( F ) ; ( x 0 , y 0 ) ) .

Example 1.5 Let α ] 0 , 1 ] , and let F : R R be a set-valued mapping defined as F ( x ) = [ min { α ( x 3 x ) , ( 2 α ) ( x 3 x ) } , max { α ( x 3 x ) , ( 2 α ) ( x 3 x ) } ] . Then ( 0 , 0 ) Graph ( F ) . From Example 1.4, we have D F ( 0 , 0 ) ( u ) = [ min { α u , ( 2 α ) u } , max { α u , ( 2 α ) u } ] for each u R .

From Definition 1.6 and Proposition 1.8, the following Proposition 1.10 can be obtained.

Proposition 1.10 Let F : R n R m , and let ( x 0 , y 0 ) Graph ( F ) . Then the following statements hold.
  1. (i)

    D F ( x 0 , y 0 ) is closed-valued.

     
  2. (ii)

    If Graph ( F ) K ( R n × R m ) , then D F ( x 0 , y 0 ) is closed convex-valued.

     

1.4 Fuzzy sets

We investigate properties of level sets of fuzzy sets, which are necessary to consider limits of sequences of fuzzy sets, and limits and derivatives of fuzzy set-valued mappings.

We identify each fuzzy set s ˜ on R n with its membership function, and s ˜ is interpreted as s ˜ : R n [ 0 , 1 ] . Let F ( R n ) be the set of all fuzzy sets on R n .

For s ˜ F ( R n ) and α [ 0 , 1 ] , the set
[ s ˜ ] α = { x R n : s ˜ ( x ) α }

is called the α-level set of s ˜ .

For a crisp set S R n , the function c S : R n { 0 , 1 } defined as
c S ( x ) = { 1 if  x S , 0 if  x S

for each x R n is called the indicator function of S. Whenever we consider c S as a fuzzy set, c S : R n { 0 , 1 } is interpreted as c S : R n [ 0 , 1 ] .

It is known as the decomposition theorem that s ˜ F ( R n ) can be represented as
s ˜ = sup α ] 0 , 1 ] α c [ s ˜ ] α

(see, for example, [13]).

A fuzzy set s ˜ F ( R n ) is said to be closed if s ˜ is upper semicontinuous. A fuzzy set s ˜ F ( R n ) is closed if and only if [ s ˜ ] α C ( R n ) , α ] 0 , 1 ] .

A fuzzy set s ˜ F ( R n ) is said to be convex if
s ˜ ( λ x + ( 1 λ ) y ) s ˜ ( x ) s ˜ ( y )

for any x , y R n and any λ [ 0 , 1 ] . Namely, s ˜ F ( R n ) is said to be convex if s ˜ is a quasiconcave function, and s ˜ is convex if and only if [ s ˜ ] α K ( R n ) , α ] 0 , 1 ] .

Let CF ( R n ) , KF ( R n ) , and CKF ( R n ) be sets of all closed, convex, and closed convex fuzzy sets on R n , respectively.

A fuzzy set s ˜ F ( R n ) is called a fuzzy cone if [ s ˜ ] α R n , α ] 0 , 1 ] are cones.

For s ˜ F ( R n ) , the set
hypo ( s ˜ ) = { ( x , α ) R n × [ 0 , 1 ] : α s ˜ ( x ) }

is called the fuzzy hypograph of s ˜ .

In order to investigate relationships between fuzzy sets and their level sets, we set
S ( R n ) = { { S α } α ] 0 , 1 ] : S α R n , α ] 0 , 1 ] ,  and  S β S γ  for  β , γ ] 0 , 1 ]  with  β < γ } ,
and define M : S ( R n ) F ( R n ) as
M ( { S α } α ] 0 , 1 ] ) = sup α ] 0 , 1 ] α c S α
for each { S α } α ] 0 , 1 ] S ( R n ) . For { S α } α ] 0 , 1 ] S ( R n ) and x R n , it can be seen that
M ( { S α } α ] 0 , 1 ] ) ( x ) = sup α ] 0 , 1 ] α c S α ( x ) = sup { α ] 0 , 1 ] : x S α } ,
where sup = 0 . By using M, the decomposition theorem can be represented as
s ˜ = M ( { [ s ˜ ] α } α ] 0 , 1 ] )

for s ˜ F ( R n ) .

Proposition 1.11 (Proposition 3 in [14])

Let { S α } α ] 0 , 1 ] , { T α } α ] 0 , 1 ] S ( R n ) . If S α T α , α ] 0 , 1 ] , then M ( { S α } α ] 0 , 1 ] ) M ( { T α } α ] 0 , 1 ] ) .

Proposition 1.12 (Proposition 4 in [14])

Let { S α } α ] 0 , 1 ] S ( R n ) , and let s ˜ = M ( { S α } α ] 0 , 1 ] ) . Then [ s ˜ ] α = β ] 0 , α [ S β for α ] 0 , 1 ] .

Definition 1.7 For s ˜ F ( R n ) , the fuzzy set
cl ( s ˜ ) = M ( { cl ( [ s ˜ ] α ) } α ] 0 , 1 ] )

is called the closure of s ˜ .

For a crisp set S R n , it can be seen that cl ( c S ) = c cl ( S ) . Thus, the closure for fuzzy sets is a generalization of the crisp one.

Proposition 1.13 (See pp.13-14 in [3])

Let s ˜ F ( R n ) , and let u ˜ CF ( R n ) be the smallest closed fuzzy set among closed fuzzy sets t ˜ CF ( R n ) such that s ˜ t ˜ . Then hypo ( u ˜ ) = cl ( hypo ( s ˜ ) ) .

Proposition 1.14 Let s ˜ F ( R n ) . Then the following statements hold.
  1. (i)

    [ cl ( s ˜ ) ] α = β ] 0 , α [ cl ( [ s ˜ ] β ) for α ] 0 , 1 ] .

     
  2. (ii)

    cl ( s ˜ ) CF ( R n ) .

     
  3. (iii)

    cl ( s ˜ ) is the smallest closed fuzzy set among closed fuzzy sets t ˜ CF ( R n ) such that s ˜ t ˜ .

     
Proof (i) follows from Proposition 1.12. (ii) follows from (i). In order to show (iii), let u ˜ CF ( R n ) be the smallest closed fuzzy set among closed fuzzy sets t ˜ CF ( R n ) such that s ˜ t ˜ , and we show that u ˜ = cl ( s ˜ ) . It follows that cl ( s ˜ ) CF ( R n ) from (ii), and that s ˜ cl ( s ˜ ) from the decomposition theorem and Proposition 1.11. Thus, we have u ˜ cl ( s ˜ ) . Assume that there exists x 0 R n such that u ˜ ( x 0 ) < cl ( s ˜ ) ( x 0 ) . We set α = cl ( s ˜ ) ( x 0 ) and β = u ˜ ( x 0 ) , and fix any γ ] β , α [ . Since α = cl ( s ˜ ) ( x 0 ) = sup { η ] 0 , 1 ] : x 0 cl ( [ s ˜ ] η ) } , it follows that
x 0 cl ( [ s ˜ ] γ ) .
(1.1)
Since β < γ , it follows that
( x 0 , γ ) hypo ( u ˜ ) = cl ( hypo ( s ˜ ) )
(1.2)

from Proposition 1.13. From (1.1), it follows that there exists { x k } k N [ s ˜ ] γ such that x k x 0 . Since s ˜ ( x k ) γ , k N , we have ( x k , γ ) hypo ( s ˜ ) , k N and ( x k , γ ) ( x 0 , γ ) cl ( hypo ( s ˜ ) ) , which contradicts (1.2). □

The following Example 1.6 shows that [ cl ( s ˜ ) ] α = cl ( [ s ˜ ] α ) does not hold in general.

Example 1.6 Let s ˜ F ( R ) be a fuzzy set defined as
s ˜ ( x ) = { max { | x | + 1 , 0 } if  x 0 , 0 if  x = 0

for each x R . In this case, we have cl ( s ˜ ) ( x ) = max { | x | + 1 , 0 } for each x R , and cl ( [ s ˜ ] 1 ) = { 0 } = [ cl ( s ˜ ) ] 1 .

From Propositions 1.11, 1.12, and 1.14, the following Proposition 1.15 can be obtained.

Proposition 1.15 Let { S α } α ] 0 , 1 ] S ( R n ) , and let s ˜ = M ( { S α } α ] 0 , 1 ] ) . Then cl ( s ˜ ) = M ( { cl ( S α ) } α ] 0 , 1 ] ) .

Proposition 1.16 Let { S α ( λ ) } α ] 0 , 1 ] S ( R n ) , λ Λ , and let s ˜ λ = M ( { S α ( λ ) } α ] 0 , 1 ] ) , λ Λ . In addition, let L α = λ Λ S α ( λ ) , U α = λ Λ S α ( λ ) for each α ] 0 , 1 ] , where L α = λ Λ S α ( λ ) = R n , U α = λ Λ S α ( λ ) = for each α ] 0 , 1 ] if Λ = . Then M ( { L α } α ] 0 , 1 ] ) = λ Λ s ˜ λ and M ( { U α } α ] 0 , 1 ] ) = λ Λ s ˜ λ .

Proof Fix any x R n .

First, assume that Λ = . Then it follows that L α = R n , U α = for each α ] 0 , 1 ] , and we have M ( { L α } α ] 0 , 1 ] ) ( x ) = 1 = λ Λ s ˜ λ ( x ) and M ( { U α } α ] 0 , 1 ] ) ( x ) = 0 = λ Λ s ˜ λ ( x ) .

Next, assume that Λ . We set β = M ( { L α } α ] 0 , 1 ] ) ( x ) = sup { α ] 0 , 1 ] : x L α } and γ = M ( { U α } α ] 0 , 1 ] ) ( x ) = sup { α ] 0 , 1 ] : x U α } .

From the definition of β, it follows that x L α , α ] 0 , β [ and x L α , α ] β , 1 ] . For any λ Λ and any α ] 0 , β [ , it follows that s ˜ λ ( x ) α since x S α ( λ ) . Thus, s ˜ λ ( x ) β for any λ Λ . For each α ] β , 1 ] , since x L α , there exists λ α Λ such that x S α ( λ α ) , and it follows that s ˜ λ α ( x ) α . Thus, for any ε > 0 , there exists λ 0 Λ such that s ˜ λ 0 ( x ) < β + ε . Therefore, we have β = λ Λ s ˜ λ ( x ) .

By the same way, we have γ = λ Λ s ˜ λ ( x ) . □

Proposition 1.17 Let { S α } α ] 0 , 1 ] , { T α } α ] 0 , 1 ] S ( R n ) . If S α = T α for any α ] 0 , 1 ] except for at most countable many α, then M ( { S α } α ] 0 , 1 ] ) = M ( { T α } α ] 0 , 1 ] ) .

Proof Assume that there exists x 0 R n such that M ( { S α } α ] 0 , 1 ] ) ( x 0 ) M ( { T α } α ] 0 , 1 ] ) ( x 0 ) . Without loss of generality, assume that M ( { S α } α ] 0 , 1 ] ) ( x 0 ) < M ( { T α } α ] 0 , 1 ] ) ( x 0 ) . We set β = M ( { S α } α ] 0 , 1 ] ) ( x 0 ) = sup { α ] 0 , 1 ] : x 0 S α } and γ = M ( { T α } α ] 0 , 1 ] ) ( x 0 ) = sup { α ] 0 , 1 ] : x 0 T α } . Since
β = sup { α ] 0 , 1 ] : x 0 S α } { x 0 S α , α ] 0 , β [ , x 0 S α , α ] β , 1 ]
and
γ = sup { α ] 0 , 1 ] : x 0 T α } { x 0 T α , α ] 0 , γ [ , x 0 T α , α ] γ , 1 ] ,

we have x 0 S α , x 0 T α , α ] β , γ [ and S α T α , α ] β , γ [ . Therefore, it is not true that S α = T α for any α ] 0 , 1 ] except for at most countable many α. □

2 Limits of sequences of fuzzy sets

In this section, we propose the definitions of limits of sequences of fuzzy sets based on their level sets, and investigate their properties.

The following Definition 2.1 is a fuzzified one of Definition 1.1.

Definition 2.1 Let { s ˜ k } k N F ( R n ) , and let
L α = lim inf k [ s ˜ k ] α , U α = lim sup k [ s ˜ k ] α
for each α ] 0 , 1 ] . The lower limit of { s ˜ k } k N is defined as the fuzzy set
lim inf k s ˜ k = M ( { L α } α ] 0 , 1 ] ) ,
and the upper limit of { s ˜ k } k N is defined as the fuzzy set
lim sup k s ˜ k = M ( { U α } α ] 0 , 1 ] ) .
The limit of { s ˜ k } k N is said to exist if lim inf k s ˜ k = lim sup k s ˜ k , and its limit is defined as the fuzzy set
lim k s ˜ k = lim inf k s ˜ k = lim sup k s ˜ k .

For crisp sets S k R n , k N , we set L = lim inf k S k , U = lim sup k S k , and set T = lim k S k if the limit of { S k } k N exists. Then it can be seen that lim inf k c S k = c L , lim sup k c S k = c U , and that lim k c S k = c T if the limit of { S k } k N exists. Thus, the lower limit, upper limit, and limit for sequences of fuzzy sets are generalizations of the crisp ones.

Example 2.1 Let s ˜ , t ˜ F ( R ) be fuzzy sets defined as
s ˜ ( x ) = { max { | x | + 1 , 0 } if  x 0 , 0 if  x = 0 , t ˜ ( x ) = max { | x | + 1 , 0 }
for each x R , and let { s ˜ k } k N , { t ˜ k } N , { u ˜ k } k N F ( R ) be sequences of fuzzy sets defined as
s ˜ k = s ˜ , t ˜ k = t ˜ , u ˜ k = { s ˜ if  k  is odd , t ˜ if  k  is even
for each k N . In this case, we have lim k [ s ˜ k ] α = lim k [ t ˜ k ] α = lim k [ u ˜ k ] α = [ α 1 , 1 α ] for α ] 0 , 1 [ , and lim k [ s ˜ k ] 1 = lim inf k [ u ˜ k ] 1 = and lim k [ t ˜ k ] 1 = lim sup k [ u ˜ k ] 1 = { 0 } . Thus, it follows that lim k s ˜ k = lim k t ˜ k = lim k u ˜ k = t ˜ . Therefore, we obtain the following statements.
  1. (i)

    lim inf k s ˜ k = lim inf k t ˜ k , while lim inf k [ s ˜ k ] 1 lim inf k [ t ˜ k ] 1 .

     
  2. (ii)

    lim sup k s ˜ k = lim sup k t ˜ k , while lim sup k [ s ˜ k ] 1 lim sup k [ t ˜ k ] 1 .

     
  3. (iii)

    lim inf k u ˜ k = lim sup k u ˜ k , while lim inf k [ u ˜ k ] 1 lim sup k [ u ˜ k ] 1 .

     
  4. (iv)

    lim k s ˜ k = lim k t ˜ k = lim k u ˜ k , while lim k [ s ˜ k ] 1 lim k [ t ˜ k ] 1 and there does not exist lim k [ u ˜ k ] 1 .

     

From Definition 1.1 and Propositions 1.3, 1.5, and 1.11, the following Proposition 2.1 can be obtained.

Proposition 2.1 Let { s ˜ k } k N F ( R n ) , and let L α = lim inf k [ s ˜ k ] α , U α = lim sup k [ s ˜ k ] α for each α ] 0 , 1 ] . Then the following statements hold.
  1. (i)

    { L α } α ] 0 , 1 ] , { U α } α ] 0 , 1 ] S ( R n ) .

     
  2. (ii)

    L α U α for α ] 0 , 1 ] .

     
  3. (iii)

    lim inf k s ˜ k lim sup k s ˜ k .

     
  4. (iv)

    L α , U α C ( R n ) for α ] 0 , 1 ] .

     
  5. (v)

    Let α ] 0 , 1 ] , and assume that [ s ˜ k ] α K ( R n ) , k N . Then L α K ( R n ) .

     

The following Proposition 2.2(i) is a fuzzified one of Proposition 1.3(i), (iii). The following Proposition 2.2(ii) is a fuzzified one of Proposition 1.5. They can be derived from Propositions 1.3, 1.12, and 2.1.

Proposition 2.2
  1. (i)

    Let { s ˜ k } k N F ( R n ) . Then lim inf k s ˜ k , lim sup k s ˜ k CF ( R n ) . If s ˜ k = u ˜ , k N for some u ˜ F ( R n ) , then lim k s ˜ k = cl ( u ˜ ) .

     
  2. (ii)

    Let { s ˜ k } k N KF ( R n ) . Then lim inf k s ˜ k CKF ( R n ) .

     

The following Proposition 2.3 is a fuzzified one of Proposition 1.1. It can be derived from the decomposition theorem and Propositions 1.1, 1.15, and 1.16.

Proposition 2.3 Let { s ˜ k } k N F ( R n ) . Then
lim inf k s ˜ k = N N cl ( k N s ˜ k ) and lim sup k s ˜ k = N N cl ( k N s ˜ k ) .

From Propositions 1.1, 1.15, 1.16, and 2.3, the following Proposition 2.4 can be obtained.

Proposition 2.4 Let { S α ( k ) } α ] 0 , 1 ] S ( R n ) , k N , and let s ˜ k = M ( { S α ( k ) } α ] 0 , 1 ] ) , k N . In addition, let L α = lim inf k S α ( k ) , U α = lim sup k S α ( k ) for each α ] 0 , 1 ] . Then lim inf k s ˜ k = M ( { L α } α ] 0 , 1 ] ) and lim sup k s ˜ k = M ( { U α } α ] 0 , 1 ] ) .

The following Proposition 2.5 is a fuzzified one of Proposition 1.3(ii). It can be derived from Propositions 1.3 and 2.4.

Proposition 2.5 Let { s ˜ k } k N , { t ˜ k } k N F ( R n ) , and assume that cl ( s ˜ k ) = cl ( t ˜ k ) , k N . Then lim inf k s ˜ k = lim inf k t ˜ k and lim sup k s ˜ k = lim sup k t ˜ k .

The following Proposition 2.6 is a fuzzified one of Proposition 1.2. It can be derived from Propositions 1.2, 1.11, 1.15, and 1.16.

Proposition 2.6 For { s ˜ k } k N , { s ˜ k 1 } k N , { s ˜ k 2 } k N F ( R n ) , the following statements hold.
  1. (i)

    If s ˜ k (which means that s ˜ k s ˜ k + 1 ), then lim k s ˜ k = cl ( k N s ˜ k ) .

     
  2. (ii)

    If s ˜ k (which means that s ˜ k s ˜ k + 1 ), then lim k s ˜ k = k N cl ( s ˜ k ) .

     
  3. (iii)

    If s ˜ k 1 s ˜ k s ˜ k 2 , k N and lim inf k s ˜ k 1 = lim inf k s ˜ k 2 , then lim inf k s ˜ k = lim inf k s ˜ k 1 = lim inf k s ˜ k 2 .

     
  4. (iv)

    If s ˜ k 1 s ˜ k s ˜ k 2 , k N and lim sup k s ˜ k 1 = lim sup k s ˜ k 2 , then lim sup k s ˜ k = lim sup k s ˜ k 1 = lim sup k s ˜ k 2 .

     
  5. (v)

    If s ˜ k 1 s ˜ k s ˜ k 2 , k N and lim k s ˜ k 1 = lim k s ˜ k 2 , then lim k s ˜ k = lim k s ˜ k 1 = lim k s ˜ k 2 .

     

The following Proposition 2.7 is a fuzzified one of Proposition 1.6. It can be derived from Definition 1.1 and Proposition 1.11.

Proposition 2.7 Let { s ˜ k } k N F ( R n ) , and let s ˜ = lim k s ˜ k . In addition, let N N . Then s ˜ = lim k N s ˜ k .

Throughout the rest of this section, let X R n be a compact set, and let
F ( R n ) = { s ˜ CF ( R n ) : sup x R n s ˜ ( x ) = 1  and  s ˜ ( x ) = 0 , x X } .
For s ˜ F ( R n ) and α [ 0 , 1 ] , we set
[ [ s ˜ ] ] α = { [ s ˜ ] α if  α ] 0 , 1 ] , cl ( { x R n : s ˜ ( x ) > 0 } ) if  α = 0 .

Let { s ˜ k } k N F ( R n ) , and let s ˜ F ( R n ) . Then { s ˜ k } k N is said to converge to s ˜ in the sense of Yoshida et al. [9] if ρ ( [ [ s ˜ k ] ] α , [ [ s ˜ ] ] α ) 0 for any α [ 0 , 1 ] except for at most countable many α, where ρ ( [ [ s ˜ k ] ] α , [ [ s ˜ ] ] α ) is the Hausdorff distance between [ [ s ˜ k ] ] α and [ [ s ˜ ] ] α defined in Definition 1.2. From Proposition 1.4, ρ ( [ [ s ˜ k ] ] α , [ [ s ˜ ] ] α ) 0 if and only if lim k [ [ s ˜ k ] ] α = [ [ s ˜ ] ] α in the sense of Definition 1.1.

The following Proposition 2.8 shows that the concept of the convergence for sequences of fuzzy sets in the sense of Definition 2.1 is weaker than that in the sense of Yoshida et al. [9].

Proposition 2.8 Let { s ˜ k } k N F ( R n ) , and let s ˜ F ( R n ) . If { s ˜ k } k N converges to s ˜ in the sense of Yoshida et al. [9], then lim k s ˜ k = s ˜ in the sense of Definition  2.1.

Proof { s ˜ k } k N converges to s ˜ in the sense of Yoshida et al. [9] if and only if lim k [ [ s ˜ k ] ] α = [ [ s ˜ ] ] α in the sense of Definition 1.1 for any α [ 0 , 1 ] except for at most countable many α. Then it follows that lim k [ s ˜ k ] α = lim inf k [ s ˜ k ] α = lim sup k [ s ˜ k ] α = [ s ˜ ] α for any α ] 0 , 1 ] except for at most countable many α. From Proposition 1.17 and the decomposition theorem, we have s ˜ = lim k s ˜ k = lim inf k s ˜ k = lim sup k s ˜ k in the sense of Definition 2.1. □

3 Limits of fuzzy set-valued mappings

In this section, we propose the definitions of limits of fuzzy set-valued mappings, and investigate their properties.

For a fuzzy set-valued mapping F ˜ : R n F ( R m ) and α ] 0 , 1 ] , we define the crisp set-valued mapping F α : R n R m as
F α ( x ) = [ F ˜ ( x ) ] α

for each x R n .

A fuzzy set-valued mapping F ˜ : R n F ( R m ) is said to be closed-valued, convex-valued, or closed convex-valued if F ˜ ( x ) CF ( R m ) , F ˜ ( x ) KF ( R m ) , or F ˜ ( x ) CKF ( R m ) for any x R n , respectively.

The following Definition 3.1 is a fuzzified one of Definition 1.3 (see Proposition 3.3 which is mentioned later).

Definition 3.1 Let F ˜ : R n F ( R m ) , and let x ¯ R n . In addition, let
L α ( x ¯ ) = lim inf x x ¯ F α ( x ) , U α ( x ¯ ) = lim sup x x ¯ F α ( x )
for each α ] 0 , 1 ] . The lower limit of F ˜ when x x ¯ is defined as the fuzzy set
lim inf x x ¯ F ˜ ( x ) = M ( { L α ( x ¯ ) } α ] 0 , 1 ] ) ,
and the upper limit of F ˜ when x x ¯ is defined as the fuzzy set
lim sup x x ¯ F ˜ ( x ) = M ( { U α ( x ¯ ) } α ] 0 , 1 ] ) .
The limit of F ˜ when x x ¯ is said to exist if lim inf x x ¯ F ˜ ( x ) = lim sup x x ¯ F ˜ ( x ) , and its limit is defined as the fuzzy set
lim x x ¯ F ˜ ( x ) = lim inf x x ¯ F ˜ ( x ) = lim sup x x ¯ F ˜ ( x ) .

For a crisp set-valued mapping F : R n R m and x ¯ R n , let L ( x ¯ ) = lim inf x x ¯ F ( x ) , U ( x ¯ ) = lim sup x x ¯ F ( x ) , and let T ( x ¯ ) = lim x x ¯ F ( x ) if the limit of F when x x ¯ exists. Then lim inf x x ¯ c F ( x ) = c L ( x ¯ ) , lim sup x x ¯ c F ( x ) = c U ( x ¯ ) , and lim x x ¯ c F ( x ) = c T ( x ¯ ) if the limit of F when x x ¯ exists. Thus, the lower limit, upper limit, and limit for fuzzy set-valued mappings are generalizations of the crisp ones.

Example 3.1 Let s ˜ , t ˜ , ˜ F ( R ) be fuzzy sets defined as
s ˜ ( x ) = { max { | x | + 1 , 0 } if  x 0 , 0 if  x = 0 , t ˜ ( x ) = max { | x | + 1 , 0 } , ˜ ( x ) = 0
for each x R , and let F ˜ : R F ( R ) be a fuzzy set-valued mapping defined as
F ˜ ( x ) = { s ˜ if  x B ( 1 , ε ) ( B ( 3 , ε ) Q ) , t ˜ if  x B ( 2 , ε ) ( B ( 3 , ε ) Q ) , ˜ otherwise
for each x R , where is the set of all rational numbers, and ε ] 0 , 1 2 [ , and B ( x ¯ , ε ) = { x R : | x x ¯ | < ε } for x ¯ R . Since
F α ( x ) = { [ α 1 , 0 [ ] 0 , 1 α ] if  x B ( 1 , ε ) ( B ( 3 , ε ) Q ) , [ α 1 , 1 α ] if  x B ( 2 , ε ) ( B ( 3 , ε ) Q ) , otherwise
for each x R when α ] 0 , 1 [ , and
F 1 ( x ) = { { 0 } if  x B ( 2 , ε ) ( B ( 3 , ε ) Q ) , otherwise
for each x R , we have lim x 1 F α ( x ) = lim x 2 F α ( x ) = lim x 3 F α ( x ) = [ α 1 , 1 α ] for α ] 0 , 1 [ , and lim x 1 F 1 ( x ) = lim inf x 3 F 1 ( x ) = and lim x 2 F 1 ( x ) = lim sup x 3 F 1 ( x ) = { 0 } . Thus, it follows that lim x 1 F ˜ ( x ) = lim x 2 F ˜ ( x ) = lim x 3 F ˜ ( x ) = t ˜ . Therefore, we obtain the following statements.
  1. (i)

    lim inf x 1 F ˜ ( x ) = lim inf x 2 F ˜ ( x ) , while lim inf x 1 F 1 ( x ) lim inf x 2 F 1 ( x ) .

     
  2. (ii)

    lim sup x 1 F ˜ ( x ) = lim sup x 2 F ˜ ( x ) , while lim sup x 1 F 1 ( x ) lim sup x 2 F 1 ( x ) .

     
  3. (iii)

    lim inf x 3 F ˜ ( x ) = lim sup x 3 F ˜ ( x ) , while lim inf x 3 F 1 ( x ) lim sup x 3 F 1 ( x ) .

     
  4. (iv)

    lim x 1 F ˜ ( x ) = lim x 2 F ˜ ( x ) = lim x 3 F ˜ ( x ) , while lim x 1 F 1 ( x ) lim x 2 F 1 ( x ) and there does not exist lim x 3 F 1 ( x ) .

     

The following Definition 3.2 is a fuzzified one of Definition 1.4.

Definition 3.2 Let F ˜ : R n F ( R m ) , and let x ¯ R n . The fuzzy set-valued mapping F ˜ is said to be lower semicontinuous at x ¯ if
lim inf x x ¯ F ˜ ( x ) F ˜ ( x ¯ ) ,
and F ˜ is said to be upper semicontinuous at x ¯ if
lim sup x x ¯ F ˜ ( x ) F ˜ ( x ¯ ) .
The fuzzy set-valued mapping F ˜ is said to be continuous at x ¯ if F ˜ is both lower and upper semicontinuous at x ¯ , that is,
lim x x ¯ F ˜ ( x ) = F ˜ ( x ¯ ) .
Example 3.2 Let s ˜ , t ˜ F ( R ) be fuzzy sets defined as s ˜ ( x ) = max { | x | + 1 2 , 0 } and t ˜ ( x ) = max { | x | + 1 , 0 } for each x R , and let F ˜ , G ˜ : R F ( R ) be fuzzy set-valued mappings defined as
F ˜ ( x ) = { s ˜ if  x 0 , t ˜ if  x > 0 , G ˜ ( x ) = { s ˜ if  x < 0 , t ˜ if  x 0

for each x R . Since lim inf x 0 F ˜ ( x ) = s ˜ s ˜ = F ˜ ( 0 ) and lim sup x 0 F ˜ ( x ) = t ˜ s ˜ = F ˜ ( 0 ) , F ˜ is lower semicontinuous at 0 but not upper semicontinuous at 0. Since lim sup x 0 G ˜ ( x ) = t ˜ t ˜ = G ˜ ( 0 ) and lim inf x 0 G ˜ ( x ) = s ˜ t ˜ = G ˜ ( 0 ) , G ˜ is upper semicontinuous at 0 but not lower semicontinuous at 0.

From Definition 1.3 and Propositions 1.3, 1.5, and 1.11, the following Proposition 3.1 can be obtained.

Proposition 3.1 Let F ˜ : R n F ( R m ) , and let x ¯ R n . In addition, let
L α ( x ¯ ) = lim inf x x ¯ F α ( x ) , U α ( x ¯ ) = lim sup x x ¯ F α ( x )
for each α ] 0 , 1 ] . Then the following statements hold.
  1. (i)

    { L α ( x ¯ ) } α ] 0 , 1 ] , { U α ( x ¯ ) } α ] 0 , 1 ] S ( R m ) .

     
  2. (ii)

    L α ( x ¯ ) cl ( F α ( x ¯ ) ) U α ( x ¯ ) for α ] 0 , 1 ] .

     
  3. (iii)

    lim inf x x ¯ F ˜ ( x ) cl ( F ˜ ( x ¯ ) ) lim sup x x ¯ F ˜ ( x ) .

     
  4. (iv)

    L α ( x ¯ ) C ( R m ) for α ] 0 , 1 ] .

     
  5. (v)

    Let α ] 0 , 1 ] , and assume that F α is convex-valued. Then L α ( x ¯ ) K ( R m ) .

     

From Propositions 1.12 and 3.1, the following Proposition 3.2 can be obtained.

Proposition 3.2 Let F ˜ : R n F ( R m ) , and let x ¯ R n . Then the following statements hold.
  1. (i)

    lim inf x x ¯ F ˜ ( x ) CF ( R m ) .

     
  2. (ii)

    If F ˜ is convex-valued, then lim inf x x ¯ F ˜ ( x ) CKF ( R m ) .

     

The following Proposition 3.3 shows that Definition 3.1 is a fuzzified one of Definition 1.3. It can be derived from Proposition 1.16.

Proposition 3.3 Let F ˜ : R n F ( R m ) , and let x ¯ R n . Then
lim inf x x ¯ F ˜ ( x ) = x k x ¯ lim inf k F ˜ ( x k ) and lim sup x x ¯ F ˜ ( x ) = x k x ¯ lim sup k F ˜ ( x k ) .

From Propositions 1.16, 2.4, and 3.3, the following Proposition 3.4 can be obtained.

Proposition 3.4 Let { S α ( x ) } α ] 0 , 1 ] S ( R m ) , x R n , and assume that F ˜ : R n F ( R m ) is defined as F ˜ ( x ) = M ( { S α ( x ) } α ] 0 , 1 ] ) for each x R n . In addition, let x ¯ R n , and let L α ( x ¯ ) = lim inf x x ¯ S α ( x ) , U α ( x ¯ ) = lim sup x x ¯ S α ( x ) for each α ] 0 , 1 ] . Then lim inf x x ¯ F ˜ ( x ) = M ( { L α ( x ¯ ) } α ] 0 , 1 ] ) and lim sup x x ¯ F ˜ ( x ) = M ( { U α ( x ¯ ) } α ] 0 , 1 ] ) .

The following Proposition 3.5 is a fuzzified one of Proposition 1.7. It can be derived from Proposition 3.3.

Proposition 3.5 Let F ˜ : R n F ( R m ) , and let x ¯ R n . Then lim x x ¯ F ˜ ( x ) = F ˜ ( x ¯ ) if and only if lim k F ˜ ( x k ) = F ˜ ( x ¯ ) for any sequence { x k } k N R n such that x k x ¯ .

4 Derivatives of fuzzy set-valued mappings

In this section, we propose the definition of derivatives of fuzzy set-valued mappings and investigate their properties.

The following Definition 4.1 is a fuzzified one of Definition 1.5.

Definition 4.1 Let s ˜ F ( R n ) , and let x 0 [ s ˜ ] 1 . Then the fuzzy set
T ˜ ( s ˜ ; x 0 ) = M ( { T ( [ s ˜ ] α ; x 0 ) } α ] 0 , 1 ] )

is called the fuzzy tangent cone or fuzzy contingent cone of s ˜ at x 0 .

Let S R n , and let x 0 S . Then it can be seen that
T ˜ ( c S ; x 0 ) = c T ( S ; x 0 ) .

Thus, the fuzzy tangent cone is a generalization of the crisp one.

Example 4.1 Let f : R R be a function defined as f ( x ) = ( x + 1 ) x ( x 1 ) = x 3 x for each x R , and let s ˜ F ( R 2 ) be a fuzzy set defined as, for each x = ( x , y ) R 2 ,
s ˜ ( x ) = max { 1 | f ( x ) | | y f ( x ) | + 1 , 0 }
if f ( x ) 0 , and
s ˜ ( x ) = { 1 if  y = 0 , 0 if  y 0
if f ( x ) = 0 . In addition, let x 0 = ( 0 , 0 ) [ s ˜ ] 1 . For each α ] 0 , 1 ] , since [ s ˜ ] α = { ( x , y ) R 2 : min { α ( x 3 x ) , ( 2 α ) ( x 3 x ) } y max { α ( x 3 x ) , ( 2 α ) ( x 3 x ) } } , it follows that T ( [ s ˜ ] α ; x 0 ) = { ( x , y ) R 2 : min { α x , ( 2 α ) x } y max { α x , ( 2 α ) x } } from Example 1.4. Therefore, for each x = ( x , y ) R 2 , we have
T ˜ ( s ˜ ; x 0 ) ( x ) = max { 1 | x | | y + x | + 1 , 0 }
if x 0 , and
T ˜ ( s ˜ ; x 0 ) ( x ) = { 1 if  y = 0 , 0 if  y 0

if x = 0 .

The following Proposition 4.1 is a fuzzified one of Proposition 1.8. It can be derived from Propositions 1.8 and 1.12.

Proposition 4.1 Let s ˜ F ( R n ) , and let x 0 [ s ˜ ] 1 . Then the following statements hold.
  1. (i)

    T ˜ ( s ˜ ; x 0 ) is a closed fuzzy cone.

     
  2. (ii)

    If s ˜ KF ( R n ) , then T ˜ ( s ˜ ; x 0 ) is a closed convex fuzzy cone.

     

The following Proposition 4.2 is a fuzzified one of Proposition 1.9. It can be derived from Propositions 1.9 and 1.11.

Proposition 4.2 Let s ˜ , t ˜ F ( R n ) with s ˜ t ˜ , and let x 0 [ s ˜ ] 1 . Then T ˜ ( s ˜ ; x 0 ) T ˜ ( t ˜ ; x 0 ) .

The following Definition 4.2 is a fuzzified one of the graph for crisp set-valued mappings.

Definition 4.2 Let F ˜ : R n F ( R m ) . The fuzzy set Graph ( F ˜ ) F ( R n × R m ) defined as
Graph ( F ˜ ) ( x , y ) = F ˜ ( x ) ( y )

for each ( x , y ) R n × R m is called the fuzzy graph of F ˜ .

From Definition 4.2, it can be seen that
[ Graph ( F ˜ ) ] α = Graph ( F α )

for α ] 0 , 1 ] .

Let F : R n R m , and assume that F ˜ : R n F ( R m ) is defined as F ˜ ( x ) = c F ( x ) for each x R n . Then it follows that
Graph ( F ˜ ) ( x , y ) = F ˜ ( x ) ( y ) = c F ( x ) ( y ) = c Graph ( F ) ( x , y )

for each ( x , y ) R n × R m . Thus, the fuzzy graph for fuzzy set-valued mappings is a generalization of the crisp one.

The following Definition 4.3 is a fuzzified one of Definition 1.6.

Definition 4.3 Let F ˜ : R n F ( R m ) , and let ( x 0 , y 0 ) [ Graph ( F ˜ ) ] 1 . Then the fuzzy set-valued mapping D F ˜ ( x 0 , y 0 ) : R n F ( R m ) such that
Graph ( D F ˜ ( x 0 , y 0 ) ) = T ˜ ( Graph ( F ˜ ) ; ( x 0 , y 0 ) )

is called the fuzzy contingent derivative of F ˜ at ( x 0 , y 0 ) .

From Definition 4.3, it can be seen that
D F ˜ ( x 0 , y 0 ) ( u ) ( v ) = Graph ( D F ˜ ( x 0 , y 0 ) ) ( u , v ) = T ˜ ( Graph ( F ˜ ) ; ( x 0 , y 0 ) ) ( u , v )

for each u R n and each v R m .

Let F : R n R m , and let ( x 0 , y 0 ) Graph ( F ) . Assume that F ˜ : R n F ( R m ) is defined as F ˜ ( x ) = c F ( x ) for each x R n . Then it follows that
Graph ( F ˜ ) ( x 0 , y 0 ) = c Graph ( F ) ( x 0 , y 0 ) = 1 ,
that is, ( x 0 , y 0 ) [ Graph ( F ˜ ) ] 1 , and that
Graph ( D F ˜ ( x 0 , y 0 ) ) = T ˜ ( Graph ( F ˜ ) ; ( x 0 , y 0 ) ) = T ˜ ( c Graph ( F ) ; ( x 0 , y 0 ) ) = c T ( Graph ( F ) ; ( x 0 , y 0 ) ) = c Graph ( D F ( x 0 , y 0 ) ) ,
that is,
Graph ( D F ˜ ( x 0 , y 0 ) ) = c Graph ( D F ( x 0 , y 0 ) ) .

Thus, the fuzzy contingent derivative for fuzzy set-valued mappings is a generalization of the crisp one.

Example 4.2 Let f : R R be the function defined in Example 4.1, and let F ˜ : R F ( R ) be a fuzzy set-valued mapping defined as, for each x R ,
F ˜ ( x ) ( y ) = max { 1 | f ( x ) | | y f ( x ) | + 1 , 0 }
for each y R if f ( x ) 0 , and
F ˜ ( x ) ( y ) = { 1 if  y = 0 , 0 if  y 0
for each y R if f ( x ) = 0 . In addition, let s ˜ F ( R 2 ) be the fuzzy set defined in Example 4.1. Then it follows that Graph ( F ˜ ) = s ˜ and ( 0 , 0 ) [ Graph ( F ˜ ) ] 1 . From Example 4.1, we have, for each u R and each v R ,
D F ˜ ( 0 , 0 ) ( u ) ( v ) = max { 1 | u | | v + u | + 1 , 0 }
if u 0 , and
D F ˜ ( 0 , 0 ) ( 0 ) ( v ) = { 1 if  v = 0 , 0 if  v 0 .

Proposition 4.3 Let F ˜ : R n F ( R m ) , and let ( x 0 , y 0 ) [ Graph ( F ˜ ) ] 1 . Then D F ˜ ( x 0 , y 0 ) ( u ) = M ( { D F α ( x 0 , y 0 ) ( u ) } α ] 0 , 1 ] ) for any u R n .

Proof Fix any u R n and any v R m . Then we have
D F ˜ ( x 0 , y 0 ) ( u ) ( v ) = Graph ( D F ˜ ( x 0 , y 0 ) ) ( u , v ) = T ˜ ( Graph ( F ˜ ) ; ( x 0 , y 0 ) ) ( u , v ) = sup { α ] 0 , 1 ] : ( u , v ) T ( [ Graph ( F ˜ ) ] α ; ( x 0 , y 0 ) ) } = sup { α ] 0 , 1 ] : ( u , v ) T ( Graph ( F α ) ; ( x 0 , y 0 ) ) } = sup { α ] 0 , 1 ] : ( u , v ) Graph ( D F α ( x 0 , y 0 ) ) } = sup { α ] 0 , 1 ] : v D F α ( x 0 , y 0 ) ( u ) } = M ( { D F α ( x 0 , y 0 ) ( u ) } α ] 0 , 1 ] ) ( v ) .

 □

The following Proposition 4.4 is a fuzzified one of Proposition 1.10. It can be derived from Proposition 4.1.

Proposition 4.4 Let F ˜ : R n F ( R m ) , and let ( x 0 , y 0 ) [ Graph ( F ˜ ) ] 1 . Then the following statements hold.
  1. (i)

    D F ˜ ( x 0 , y 0 ) is closed-valued.

     
  2. (ii)

    If Graph ( F ˜ ) KF ( R n × R m ) , then D F ˜ ( x 0 , y 0 ) is closed convex-valued.

     

5 Conclusions

In this article, we proposed definitions of limits of sequences of fuzzy sets, and limits and derivatives of fuzzy set-valued mappings based on level sets of fuzzy sets, and investigated their properties. They are fuzzified ones of them for crisp ones.

Derived results are very general in the sense that they deal with all fuzzy sets, especially fuzzy sets which are not support bounded.

Consider some fuzzy mathematical model whose optimal value/solution output is a fuzzy set for an input parameter. The concepts of limits of sequences of fuzzy sets, limits and derivatives of fuzzy set-valued mappings are necessary and important for stability theory and sensitivity analysis for such fuzzy mathematical models. Then derived results can be expected to be useful for them.

Declarations

Authors’ Affiliations

(1)
Graduate School of Science and Technology, Hirosaki University, Hirosaki, Japan
(2)
Faculty of Business Administration and Information Science, Kanazawa Gakuin University, Kanazawa, Japan

References

  1. Aubin JP, Frankowska H: Set-Valued Analysis. Birkhäuser, Basel; 1990.MATHGoogle Scholar
  2. Maeda T: Multiobjective Decision Making Theory and Economic Analysis. Makino-Syoten, Japan; 1996. (in Japanese)Google Scholar
  3. Rockafellar RT, Wets RJB: Variational Analysis. Springer, New York; 1998.View ArticleMATHGoogle Scholar
  4. Tanino T: Theory and applications of set-valued mappings, Part 1: fundamental properties of set-valued mappings. J. Jpn. Soc. Fuzzy Syst. 2001, 13: 11–19. (in Japanese)Google Scholar
  5. Tanino T: Theory and applications of set-valued mappings, Part 2: derivatives of set-valued mappings and applications to optimization. J. Jpn. Soc. Fuzzy Syst. 2001, 13: 146–154. (in Japanese)Google Scholar
  6. Tanino T: Theory and applications of set-valued mappings, Part 3: applications of set-valued mappings to dynamical systems, game theory and so on. J. Jpn. Soc. Fuzzy Syst. 2001, 13: 234–242. (in Japanese)Google Scholar
  7. Diamond P, Kloeden P: Metric Spaces of Fuzzy Sets: Theory and Applications. World Scientific, Singapore; 1994.View ArticleMATHGoogle Scholar
  8. Kurano M, Yasuda M, Nakagami J, Yoshida Y: Ordering of convex fuzzy sets - a brief survey and new results. J. Oper. Res. Soc. Jpn. 2000, 43: 138–148.MathSciNetView ArticleMATHGoogle Scholar
  9. Yoshida Y, Yasuda M, Nakagami J, Kurano M: A limit theorem in dynamic fuzzy systems with a monotone property. Fuzzy Sets Syst. 1998, 94: 109–119. 10.1016/S0165-0114(96)00214-XMathSciNetView ArticleMATHGoogle Scholar
  10. Furukawa N: Convexity and local Lipschitz continuity of fuzzy-valued mappings. Fuzzy Sets Syst. 1998, 93: 113–119. 10.1016/S0165-0114(96)00192-3MathSciNetView ArticleMATHGoogle Scholar
  11. Furukawa N: Mathematics of Fuzzy Optimization. Morikita-Syuppan, Japan; 1999. (in Japanese)Google Scholar
  12. Bazaraa MS, Goode JJ, Nashed MZ: On the cones of tangents with applications to mathematical programming. J. Optim. Theory Appl. 1974, 13: 11–19.MathSciNetView ArticleMATHGoogle Scholar
  13. Dubois D, Ostasiewicz W, Prade H: Fuzzy sets: history and basic notions. In Fundamentals of Fuzzy Sets. Edited by: Dubois D, Prade H. Kluwer Academic, Boston; 2000:21–124.View ArticleGoogle Scholar
  14. Kon, M: On degree of non-convexity of fuzzy sets. Sci. Math. Jpn. 76, 417–425 (in press)Google Scholar

Copyright

© Kon and Kuwano; licensee Springer. 2013

This article is published under license to BioMed Central Ltd. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.