Information Sciences 213 (2012) 84–93

Contents lists available at SciVerse ScienceDirect

Information Sciences
journal homepage: www.elsevier.com/locate/ins

Properties of Atanassov’s intuitionistic fuzzy relations

and Atanassov’s operators
Barbara Peß kala
University of Rzeszów, Institute of Mathematics, Rejtana 16a, 35-310 Rzeszów, Poland

a r t i c l e i n f o a b s t r a c t

Article history: The goal of this paper is to consider properties of Atanassov’s intuitionistic fuzzy relations
Received 11 March 2011 which were introduced by Atanassov in 1986. Fuzzy set theory turned out to be a useful
Received in revised form 19 May 2012 tool to describe situations in which the data are imprecise or vague. Atanassov’s intuition-
Accepted 25 May 2012
istic fuzzy set theory is a generalization of fuzzy set theory which was introduced by Zadeh
Available online 4 June 2012
in 1965. This paper is a continuation of examinations by Pe ßkala [22] on the interval-valued
fuzzy relations. We study standard properties of Atanassov’s intuitionistic fuzzy relations
Keywords:
in the context of Atanassov’s operators.
Fuzzy relations
Atanassov’s intuitionistic fuzzy relations
Ó 2012 Elsevier Inc. All rights reserved.
Atanassov’s operators

1. Introduction

The idea of a fuzzy relation was defined in [26]. An extension of fuzzy set theory is Atanassov’s intuitionistic fuzzy set
theory. An Atanassov’s intuitionistic fuzzy relation is a pair of fuzzy relations, namely a membership and a non-membership
function, which represent positive and negative aspects of the given information. These objects introduced by Atanassov and
originally called intuitionistic fuzzy relations were recently suggested to be called Atanassov’s intuitionistic fuzzy relations
[14]. Fuzzy sets and relations have many applications in diverse types of areas, for example in data bases, pattern recogni-
tion, neural networks, fuzzy modeling, economy, medicine, and multicriteria decision making. Atanassov’s fuzzy set theory is
also widely applied in solving real-life problems. An example of such application is the optimization in Atanassov’s intuition-
istic fuzzy environment (an extension of fuzzy optimization and an application of Atanassov’s intuitionistic fuzzy sets) where
by applying this concept it is possible to reformulate the optimization problem by using degrees of rejection of constraints
and values of the objective which are non-admissible. This concept allows one to define a degree of rejection which cannot
be simply a complement of the degree of acceptance. Atanassov’s intuitionistic fuzzy optimization problem may be applied
to nutrient medium optimization in bioprocess engineering and to decision making in biomedical engineering (cf. [1]). In
addition, the idea of a positive and negative information was confirmed by psychological investigations [8]. Moreover, mul-
tiattribute decision making using Atanassov’s intuitionistic fuzzy sets is possible (see [18,19]). If it comes to Atanassov’s
intuitionistic fuzzy relations, the effective approach to deal with decision making in medical diagnosis was proposed with
the use of composition of such objects (cf. [9]). Atanassov’s intuitionistic fuzzy relations (sets) are equivalent to some other
extensions of fuzzy relations (sets) (see [11]). Among others, Atanassov’s intuitionistic fuzzy relations are isomorphic to
interval-valued fuzzy relations. This fact was noticed by several authors [2,10,11]. Any interval-valued fuzzy set is defined
by an interval-valued membership function: a mapping from the given universe to the set of all closed subintervals of [0, 1]
(it means that information is incomplete). In this work we study properties of Atanassov’s intuitionistic fuzzy relations,
powers, lattice operations and Atanassov’s operators of these relations. Consideration of diverse properties of Atanassov’s

E-mail address: bpekala@univ.rzeszow.pl

0020-0255/$ - see front matter Ó 2012 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.ins.2012.05.024
 B. Peßkala / Information Sciences 213 (2012) 84–93 85

operators is interesting not only from a theoretical point of view but also for the applications. It is possible to use Atanassov’s
operators as parameters in Intuitionistic Fuzzy Systems. Furthermore, Atanassov’s intuitionistic fuzzy relations are applied in
classification and in decision making [4].
First, in this work we recall some concepts and results useful in our further considerations. Next, we study properties of
powers of Atanassov’s intuitionistic fuzzy relations. Finally, we consider standard properties of Atanassov’s intuitionistic fuz-
zy relations and we study connections of their properties with lattice operations and some Atanassov’s operators. We con-
sider preservation of some properties of Atanassov’s intuitionistic fuzzy relations by lattice operations and some Atanassov’s
operators.

2. Basic definitions

First we recall the notion of the lattice operations and the order in the family of Atanassov’s intuitionistic fuzzy relations.

Definition 1 (cf. [3]). Let X, Y – ;, R, Rd:X  Y ? [0, 1] be fuzzy relations fulfilling the condition

Rðx; yÞ þ Rd ðx; yÞ 6 1; ðx; yÞ 2 ðX  YÞ:

d
A pair q = (R, R ) is called an Atanassov’s intuitionistic fuzzy relation. The family of all Atanassov’s intuitionistic fuzzy rela-
tions described in the given sets X, Y is denoted by AIFR(X  Y). If X = Y we will write AIFR(X). The boundary elements in
AIFR(X  Y) are 1 = (1, 0) and 0 = (0, 1), where 0, 1 are the constant fuzzy relations.
Basic operations and relations for q = (R, Rd), r = (S, Sd) are the union and the intersection

q _ r ¼ ðmaxðR; SÞ; minðRd ; Sd ÞÞ; q ^ r ¼ ðminðR; SÞ; maxðRd ; Sd ÞÞ:

Example 1. Let X ¼ 3; q; r 2 AIFRðXÞ and

2 3
ð0:5; 0:5Þ ð0:5; 0:5Þ ð0:3; 0:7Þ
q¼6 7
4 ð0:5; 0:5Þ ð0:5; 0:5Þ ð0:3; 0:5Þ 5;
ð0:7; 0:3Þ ð0:5; 0:3Þ ð0:5; 0:5Þ
2 3
ð0:3; 0:5Þ ð0:5; 0:5Þ ð0:1; 0:7Þ
r¼6 7
4 ð0:5; 0:5Þ ð0:5; 0:5Þ ð0:4; 0:5Þ 5;
ð0:6; 0:3Þ ð0:5; 0:3Þ ð0:5; 0:5Þ
2 3
ð0:5; 0:5Þ ð0:5; 0:5Þ ð0:3; 0:7Þ
q _ r ¼ 4 ð0:5; 0:5Þ ð0:5; 0:5Þ ð0:4; 0:5Þ 7
6
5;
ð0:7; 0:3Þ ð0:5; 0:3Þ ð0:5; 0:5Þ
2 3
ð0:3; 0:5Þ ð0:5; 0:5Þ ð0:1; 0:7Þ
q^r¼6 7
4 ð0:5; 0:5Þ ð0:5; 0:5Þ ð0:3; 0:5Þ 5:
ð0:6; 0:3Þ ð0:5; 0:3Þ ð0:5; 0:5Þ
Similarly, for arbitrary set T – ;
! !
_ _ ^
qt ðx; yÞ ¼ Rt ðx; yÞ; Rdt ðx; yÞ ;
t2T t2T t2T
! !
^ ^ _
qt ðx; yÞ ¼ Rt ðx; yÞ; Rdt ðx; yÞ :
t2T t2T t2T

Moreover, order is defined by:

q 6 r () ðR 6 S; Sd 6 Rd Þ:
The pair (AIFR(X  Y), 6) is a partially ordered set. Operations _,^ are the supremum and the infimum in AIFR(X  Y), respec-
tively. As a result, the family (AIFR(X  Y), _, ^) is a lattice (for the notion of a lattice and other related concepts see [5]) which
is a consequence of the fact that ([0, 1], max, min) is a lattice. The lattice AIFR(X  Y) is complete. This fact follows from the
W V
notion of the supremum and infimum and from the fact that the values of fuzzy relations are from the interval [0, 1]
86 B. Peßkala / Information Sciences 213 (2012) 84–93

which, with the operations maximum and minimum, forms a complete lattice. As a result (AIFR(X  Y), _, ^) is a complete,
distributive lattice (see [10, Lemma 2.1]).
It was mentioned in the Introduction that Atanassov’s intuitionistic fuzzy relations are equivalent to some other exten-
sions of fuzzy relations (sets), i.e. interval-valued fuzzy relations.

Definition 2 (cf. [25,23]). Let Int([0, 1]) be the set of all closed subintervals of [0, 1]. An interval-valued fuzzy relation R
between universes X, Y is a mapping R : X  Y ! Intð½0; 1Þ such that

Rðx; yÞ ¼ ½Rðx; yÞ; Rðx; yÞ 2 Intð½0; 1Þ;

for all pairs (x, y) 2 (X  Y). The class of all interval-valued fuzzy relations between universes X, Y will be denoted by
IVFR(X  Y) or IVFR(X) for X = Y.
Interval-valued fuzzy relations reflect the idea that membership grades are often not precise and the intervals represent
such uncertainty.
Let I; R 2IVFR(X  Y). Then for every (x, y) 2 (X  Y) we can define

Iðx; yÞ 6 Rðx; yÞ () Iðx; yÞ 6 Rðx; yÞ; Iðx; yÞ 6 Rðx; yÞ;

ðI _ RÞðx; yÞ ¼ ½maxðIðx; yÞ; Rðx; yÞÞ; maxðIðx; yÞ; Rðx; yÞÞ;
ðI ^ RÞðx; yÞ ¼ ½minðIðx; yÞ; Rðx; yÞÞ; minðIðx; yÞ; Rðx; yÞÞ:

The isomorphism which proves the equivalence between Atanassov’s intuitionistic fuzzy relations and interval-valued
fuzzy relations is the following:

Theorem 1 (cf. [2]). The mapping w:AIFR(X  Y) ? IVFR(X  Y) is an isomorphism between the lattices (AIFR(X  Y), _, ^) and
(IVFR(X  Y), _, ^), where q(x, y) = (R(x, y), Rd(x, y)) 2 AIFR(X  Y) and w(q) = [R(x, y),1  Rd(x, y)], (x, y) 2 (X  Y).
The existence of the membership degree interval [R(x, y), 1  Rd(x, y)] is always possible thanks to the condition R(x,
y) + Rd(x, y) 6 1 which each Atanassov’s intuitionistic fuzzy relation q = (R, Rd) should fulfill. The presented above isomor-
phism is due to the appropriate choice of operations in both theories, however the initial ideas behind each theory are dif-
ferent (see [14]). Let us now turn to considerations connected strictly with Atanassov’s intuitionistic fuzzy relations. The
fuzzy relation pq: X  X ? [0, 1] is associated with each Atanassov’s intuitionistic fuzzy relation q = (R, Rd), where

pq ðx; yÞ ¼ 1  Rðx; yÞ  Rd ðx; yÞ; x; y 2 X: ð1Þ

The number pq(x, y) is called an index of an element (x, y) in an Atanassov’s intuitionistic fuzzy relation q. It is also described
as an index (a degree) of hesitation whether x and y are in the relation q or not (see [21]). This value is also regarded as a
measure of non-determinacy or uncertainty (see [19]) and is useful in applications. For example, if we consider the multi-
criteria decision making problems in the Atanassov’s intuitionistic fuzzy environment with the finite set of alternatives
A = {A1, . . . , Am} and a finite set of criteria C = {C1, . . . , Cn}, then rij ; rdij are the degree of membership and the degree of non-
membership of the alternative Ai 2 A satisfying the criterion Cj 2 C, respectively. The intuitionistic indices pij ¼ 1  rij  rdij
are such that the larger pij the higher is a hesitation margin of the decision maker of the alternative Ai with respect to
the criterion Cj which intensity is given by rij. Intuitionistic indices allow to calculate the best final result and the worst
one that may be expected in a process leading to a final optimal decision (see [19]).
Now we put the definition of some Atanassov’s operators.

Definition 3 (cf. [4, Definitions 1.63, 1.100, 1.118]). Let q 2 AIFR(X  Y), q = (R, Rd), a, b 2 [0, 1], a + b 6 1, we define the
operators Fa,b, Pa,b, Qa,b, a, a: AIFR(X  Y) ? AIFR(X  Y) such that

F a;b ðqÞðx; yÞ ¼ ðRðx; yÞ þ apq ðx; yÞ; Rd ðx; yÞ þ bpq ðx; yÞÞ; ð2Þ
d
Pa;b ðqÞðx; yÞ ¼ ðmaxða; Rðx; yÞÞ; minðb; R ðx; yÞÞÞ; ð3Þ
Q a;b ðqÞðx; yÞ ¼ ðminða; Rðx; yÞÞ; maxðb; Rd ðx; yÞÞÞ; ð4Þ
d
a ðqÞðx; yÞ ¼ ðaRðx; yÞ; 1 þ aR ðx; yÞ  aÞ; ð5Þ
d
a ðqÞðx; yÞ ¼ ð1 þ aRðx; yÞ  a; aR ðx; yÞÞ: ð6Þ

3. Properties of Atanassov’s intuitionistic fuzzy relations

For our further considerations we need the following properties:

Definition 4 (cf. [16]). Let ⁄:[0, 1]2 ? [0, 1]. Operation  is infinitely sup-distributive, if
 B. Peßkala / Information Sciences 213 (2012) 84–93 87

! !
_ _ _ _
ðxt  yÞ ¼ xt  y; ðy  xt Þ ¼ y  xt
t2T t2T t2T t2T

and  is infinitely inf-distributive, if

! !
^ ^ ^ ^
ðxt  yÞ ¼ xt  y; ðy  xt Þ ¼ y  xt :
t2T t2T t2T t2T

3.1. Properties of powers

Let us recall the notion of the composition and powers in AIFR.

Definition 5 (cf. [16], [6, Definition 3]). Let :[0, 1]2 ? [0, 1], r = (S, Sd) 2 AIFR(X  Y), q = (R, Rd) 2 AIFR(Y  Z). By the
composition of relations r and q we call the relation r  q 2 AIFR(X  Z),

ðr  qÞðx; zÞ ¼ ððS  RÞðx; zÞ; ðSd 0 Rd Þðx; zÞÞ;

where
_
ðS  RÞðx; zÞ ¼ ðSðx; yÞ  Rðy; zÞÞ;
y2Y
^
ðSd 0 Rd Þðx; zÞ ¼ ðSd ðx; yÞ0 Rd ðy; zÞÞ
y2Y

and a  0 b = 1  [(1  a)  (1  b)], a, b 2 [0, 1].

The dual operation 0 preserves associativity and isotonicity property. Moreover, the dual operation changes neutral ele-
ment 1–0 and zero element 0–1 or conversely.
As a consequence of results of [22,12] and [13, Theorem 2] we obtain:

Lemma 1. Let q, r 2 AIFR(X  Y) and : [0, 1]2 ? [0, 1].

If  is an isotonic operation, then
!
_ _
ð rt  q Þ 6 rt  q ; ð7Þ
t2T t2T
!
^ ^
ðrt  qÞ P rt  q ; ð8Þ
t2T t2T
!
_ _
ð rt  0 q Þ 6 rt 0 q; ð9Þ
t2T t2T
!
^ ^
ðrt 0 qÞ P rt 0 q: ð10Þ
t2T t2T
Moreover, similarly to [22, Lemma 3] and by [13, Theorem 1] we deduce:

Lemma 2. Let :[0, 1]2 ? [0, 1].

1. If an operation  is infinitely sup-distributive, then sup   composition is also infinitely sup-distributive.

2. If an operation  is infinitely sup-distributive, then 0 is infinitely inf-distributive.
3. If 0 is infinitely inf-distributive, then inf  0 is infinitely inf-distributive.

As a result by associativity and infinite sup-distributivity of operation  we obtain associativity of composition sup  
and inf  0 .

Proposition 1 (cf. [22]). If  is an associative, infinitely sup-distributive operation with a zero element z = 0 and a neutral element
e = 1, then
AIFR(X), ) is an ordered semigroup with the identity I = [I, Id], where
 
1; forx ¼ y 0; forx ¼ y
I¼ ; Id ¼ ; x; y 2 X:
0; forx – y 1 forx – y
In the sequel we will consider arbitrary binary operations.
88 B. Peßkala / Information Sciences 213 (2012) 84–93

:[0, 1]2 ? [0, 1] which are associative and infinitely sup-distributive (these conditions imply that  is isotonic). As a re-
sult a special case of  may be a left-continuous and 0 right-continuous triangular norm or conorm. If  is associative and
infinitely sup-distributive, then in a semigroup (AIFR(X), ) we can consider the powers of its elements, i.e. relations qn
for q 2 AIFRðXÞ; n 2 N. Analogously to [17] we define.

Definition 6. By the powers of a relation q = (R, Rd) 2 AIFR(X) we call Atanassov’s intuitionistic fuzzy relations

q1 ¼ q; qmþ1 ¼ qm  q; where m ¼ 1; 2; . . . :
By the upper closure q_ and the lower closure q^ of the relation q we call
_
1 ^
1
q_ ¼ qk ; q^ ¼ qk :
k¼1 k¼1

Theorem 2. If q 2 AIFR(X),  is associative and infinitely sup – distributive, then

qn  q_ ¼ q_  qn ; ð11Þ
ðq_ Þn P ðqn Þ_ ; ðq^ Þn 6 ðqn Þ^ ; n ¼ 1; 2; . . . : ð12Þ

Proof. By infinite sup-distributivity of the composition  and infinite inf-distributivity of the composition 0 we obtain
!
_
1 _
1 _
1
q  q_ ¼ q  qk ¼ qkþ1 ¼ qk ;
k¼1 k¼1 k¼2
!
_
1 _
1 _
1
k k
_
q q¼ q q¼ ðq  q Þ ¼ qk ;
k¼1 k¼1 k¼2

which proves that q and q_ are commuting. Then by mathematical induction we obtain commutativity of their powers and
we get (11). Inequalities (12) are obtained by mathematical induction using also infinite sup-distributivity, infinite inf-dis-
tributivity and sub-distributivity from (7)–(10). h
From the above conditions we obtain the following dependencies between upper and lower closures.

Corollary 1. If q 2 AIFR(X),  is associative and infinitely sup – distributive, then (q_)_ = q_, (q^)^ 6 q^, (q^)_ 6 (q_)^.
These inequalities may be strong.

Example 2. Let c 2 (0, 1], q = (A, Ad), r = (B, Bd) 2 AIFR(X),  = ^. We have for the memberships relations the following
dependencies
2 3 2 3
0 0 c 0 0 0
6 7 6 7
A¼4c c 0 5; ðA^ Þ_ ¼ 4 c c 0 5 < ðA_ Þ^ ¼ ½c:
0 c 0 c c 0
Similarly,
2 3 2 3 2 3
0 c 0 0 0 c c c 0
6 7 6 7 6 7
B ¼ 40 0 c 5; B2 ¼ 4 c c 0 5; B3 ¼ 4 0 c c 5;
c c 0 0 c c c c c
2 3 2 3
0 c c 0 0 0
6 7 6 7
B4 ¼ 4 c c c 5; B5 ¼ ½c; B^ ¼ 4 0 0 0 5:
c c c 0 c 0
^ ^ ^
Thus B > [0] = (B ) .

3.2. Reflexivity

Now we examine some standard properties. We apply the following ones which are modifications of the properties ap-
plied in [7,15]. We study preservation of these properties by powers, lattice operations and Atanassov operators. Preserva-
tion of some properties of Atanassov’s intuitionistic fuzzy relations by triangular norms and conorms were investigated by
Bustince in [6]. First, we give definition of reflexivity.
 B. Peßkala / Information Sciences 213 (2012) 84–93 89

Definition 7 (cf. [6, Definition 5]). Let q 2 AIFR(X). A relation q is called reflexive if

qðx; xÞ ¼ 1 for all x 2 X:

Proposition 2. Let q 2 AIFR(X) and an operation  be associative and infinitely sup-distributive having a neutral element 1. If q is
reflexive, then

qn ; q_ ; q^
are reflexive, where n 2 N.

Proof. Let q be reflexive, so (I, Id) 6 (R, Rd). By infinite sup-distributivity we have isotonicity of the composition, I 6 q 6 q2,
q 6 qn, as a result qn ; n 2 N are reflexive. Moreover, by the property of supremum and infimum we obtain I 6 q_ and I 6 q^,
so q_ and q^ are reflexive. h
Similarly, lattice operations and only some Atanassov’s operators preserve reflexivity.

Proposition 3. Let q, r 2 AIFR(X), a, b 2 [0, 1], a + b 6 1. If relations q,r are reflexive, then

q ^ r; q _ r
F a;b ðqÞ; Pa;b ðqÞ; a ðqÞ
are reflexive.

Proof. If relations q = (R, Rd), r = (S, Sd) are reflexive, then by definition of intersection we obtain

ðq ^ rÞðx; xÞ ¼ qðx; xÞ ^ rðx; xÞ ¼ ðRðx; xÞ ^ Sðx; xÞ; Rd ðx; xÞ _ Sd ðx; xÞÞ ¼ ð1 ^ 1; 0 _ 0Þ ¼ ð1; 0Þ;
so q ^ r is reflexive. Similarly, for reflexive relation q or r we obtain reflexivity of q _ r, so by reflexivity of q we have

ðq _ rÞðx; xÞ ¼ ðRðx; xÞ _ Sðx; xÞ; Rd ðx; xÞ ^ Sd ðx; xÞÞ ¼ ð1 _ Sðx; xÞ; 0 ^ Sd ðx; xÞÞ ¼ ð1; 0Þ:
Now we will prove reflexivity of Fa,b(q). By reflexivity of q we have pq(x, x) = 0 and we obtain

F a;b ðqÞðx; xÞ ¼ ðRðx; xÞ þ apq ðx; xÞ; Rd ðx; xÞ þ bpq ðx; xÞÞ ¼ ð1 þ a  0; 0 þ b  0Þ ¼ ð1; 0Þ:
Thus Fa,b(q) is reflexive. Other conditions we may prove in a similar way. Directly by the definition of Pa,b(q), a(q) and by
reflexivity of q we obtain

Pa;b ðqÞðx; xÞ ¼ ðmaxða; Rðx; xÞÞ; minðb; Rd ðx; xÞÞÞ ¼ ðmaxða; 1Þ; minðb; 0ÞÞ ¼ ð1; 0Þ:

a ðqÞðx; xÞ ¼ ð1 þ aRðx; xÞ  a; aRd ðx; xÞÞ ¼ ð1 þ a  1  a; a  0Þ ¼ ð1; 0Þ:

This finishes the proof. h

3.3. Irreflexivity

Now we consider irreflexivity.

Definition 8 (cf. [6, Definition 5]). Let q 2 AIFR(X). A relation q is called irreflexive if

qðx; xÞ ¼ 0 for all x 2 X:

Directly from definition of the lower closure we obtain preservation of irreflexivity.

Proposition 4. Let q 2 AIFR(X) and an operation  be associative and infinitely sup-distributive. If q is irreflexive, then q^ is
irreflexive.
Moreover, for lattice operations and some Atanassov operators we get.

Proposition 5. Let q, r 2 AIFR(X), a, b 2 [0, 1], a + b 6 1. If relations q,r are irreflexive, then

q ^ r; q _ r
F a;b ðqÞ; Q a;b ðqÞ; a ðqÞ
are irreflexive.
90 B. Peßkala / Information Sciences 213 (2012) 84–93

Proof. If relations q = (R, Rd), r = (S, Sd) are irreflexive, then by definition of sum we obtain

ðq _ rÞðx; xÞ ¼ qðx; xÞ _ rðx; xÞ ¼ ðRðx; xÞ _ Sðx; xÞ; Rd ðx; xÞ ^ Sd ðx; xÞÞ

¼ ð0 _ 0; 1 ^ 1Þ ¼ ð0; 1Þ;
so q _ r is irreflexive. Similarly, for irreflexive relation q or r we obtain irreflexivity of q ^ r, so by irreflexivity of q we have

ðq ^ rÞðx; xÞ ¼ ðRðx; xÞ ^ Sðx; xÞ; Rd ðx; xÞ _ Sd ðx; xÞÞ ¼ ð0 ^ Sðx; xÞ; 1 _ Sd ðx; xÞÞ ¼ ð0; 1Þ:
We will prove irreflexivity of Fa,b(q). If relation q is irreflexive, then

F a;b ðqÞðx; xÞ ¼ ðRðx; xÞ þ apðx; xÞ; Rd ðx; xÞ þ bpðx; xÞÞ ¼ ð0; 1Þ:
By the properties of minimum, maximum and the product operation we have irreflexivity of Qa,b(q) and a(q).

Q a;b ðqÞðx; xÞ ¼ ðminða; Rðx; xÞÞ; maxðb; Rd ðx; xÞÞÞ ¼ ðminða; 0Þ; maxðb; 1ÞÞ ¼ ð0; 1Þ;

a ðqÞðx; xÞ ¼ ðaRðx; xÞ; 1 þ aRd ðx; xÞ  aÞ ¼ ða  0; 1 þ a  1  aÞ ¼ ð0; 1Þ:

This finishes the proof. h

3.4. Symmetry

Definition 9 (cf. [6, Definition 7]). Let q 2 AIFR(X). A relation q is called symmetric if
qðx; yÞ ¼ qðy; xÞ for all x; y 2 X:

Proposition 6. Let q 2 AIFR(X), an operation  be associative, infinitely sup-distributive and commutative. If q is symmetric, then

qn ; q_ ; q^ ; ;n 2 N
are symmetric.

Proof. Let q be symmetric and  be associative, infinitely sup – distributive and commutative. Then for x, z 2 X we obtain
commutativity of 0 and we see that
!
_ ^
q2 ðx; zÞ ¼ ðRðx; yÞ  Rðy; zÞÞ; ðRd ðx; yÞ0 Rd ðy; zÞÞ ¼
y2X y2X
!
_ ^
ðRðy; zÞ  Rðx; yÞÞ; ðRd ðy; zÞ0 Rd ðx; yÞÞ ¼
y2X y2X
!
_ ^ d 0 d
ðRðz; yÞ  Rðy; xÞÞ; ðR ðz; yÞ R ðy; xÞÞ ¼ q2 ðz; xÞ
y2X y2X

and by mathematical induction we obtain symmetry of qn. Moreover, by the properties of infimum and supremum we com-
plete the proof. h
Directly by definition of Atanassov operators and lattice operations we see that.

Proposition 7. Let q, r 2 AIFR(X), a, b 2 [0, 1], a + b 6 1. If relations q, r are symmetric, then

q ^ r; q _ r
F a;b ðqÞ; Pa;b ðqÞ; Q a;b ðqÞ; a ðqÞ; a ðqÞ
are symmetric.

3.5. Pseudo-asymmetry

Definition 10 (cf. [6, Definition 7]). Let q 2 AIFR(X). A relation q is called pseudo-asymmetric if
qðx; yÞ – qðy; xÞ and pq ðx; yÞ ¼ pq ðy; xÞ; for all x; y 2 X and x – y:
 B. Peßkala / Information Sciences 213 (2012) 84–93 91

Proposition 8. Let q 2 AIFR(X), a, b 2 [0, 1], a + b 6 1. If relation q is pseudo-asymmetric, then

1. Fa,b(q) is pseudo-asymmetric.
2. a(q),  a(q) are pseudo-asymmetric for a, b 2 (0, 1], a + b 6 1.
3. If R P a, Rd 6 b, then Pa,b(q) is pseudo-asymmetric.
4. If R 6 a, Rd P b, then Qa,b(q) is pseudo-asymmetric.

Proof. For example we will prove the first and the third condition. Let x, y 2 X and q 2 AIFR(X) be pseudo-asymmetric, i.e.
pq(x, y) = pq(y, x) and R(x, y) – R(y, x) or Rd(x, y) – Rd(y, x),
then
apq(x, y) = apq(y, x) and R(x, y) + apq(x, y) – R(y, x) + apq(y, x) or.
Rd(x, y) + bpq(x, y) – Rd(y, x) + bpq(y, x),
so
F a;b ðqÞðx; yÞ – F a;b ðqÞðy; xÞ:
Moreover,

pF a;b ðqÞ ðx; yÞ ¼ 1  Rd ðx; yÞ  bpq ðx; yÞ  ðRðx; yÞ þ apq ðx; yÞÞ ¼ 1  Rd ðy; xÞ  bpq ðy; xÞ  ðRðy; xÞ þ apq ðy; xÞÞ
¼ pF a;b ðqÞ ðy; xÞ:

Thus Fa,b(q) is pseudo-asymmetric.

If R P a,Rd 6 b, then
ðmaxða; Rðx; yÞÞ ¼ Rðx; yÞ – Rðy; xÞ ¼ maxða; Rðy; xÞÞ;
minðb; Rd ðx; yÞÞ ¼ Rd ðx; yÞ – Rd ðy; xÞ ¼ minðb; Rd ðy; xÞÞ:
As a result we obtain Pa,b(q)(x, y) – Pa,b(q)(y, x) and

1  minðb; Rd ðx; yÞÞ  maxða; Rðx; yÞÞ ¼ 1  Rd ðx; yÞ  Rðx; yÞ ¼

1  Rd ðy; xÞ  Rðy; xÞ ¼ 1  minðb; Rd ðy; xÞÞ  maxða; Rðy; xÞÞ:
Thus Pa,b(q) is pseudo-asymmetric. Other conditions we may prove in a similar way. h

3.6. Transitivity

In this section we continue the study from [22] and we examine transitivity. Similarly to definitions of properties of fuzzy
relations considered by Kaufmann we have:

Definition 11. Let q 2 AIFR(X). A relation q is called subidempotent (transitive) if q2 6 q.

Similarly to [22,6] we observe that.

Theorem 3. Let  be an associative, infinitely sup-distributive operation and q2 AIFR(X). q_ is the least subidempotent relation
greater than or equal to q. Moreover, the relation q is subidempotent if and only if q = q_.

Corollary 2. Let q 2 AIFR(X), an operation  be associative and infinitely sup-distributive. If q is symmetric and reflexive, then q_
is symmetric, reflexive and transitive.
In [6], the authors introduced for Atanassov’s intuitionistic fuzzy relations the concept of a ‘‘partially included relation’’.
The justification for consideration of partially included relations is connected with the fact that Atanassov’s operators
Fa,1a(q), where a 2 [0, 1], do not generally keep the transitivity property of a Atanassov’s intuitionistic fuzzy relation. How-
ever, for Atanassov’s intuitionistic fuzzy relations which are partially included, transitivity of an Atanassov’s intuitionistic
fuzzy relation q is equivalent to transitivity of operator Fa,1a(q) for each a 2 [0, 1]. We consider a more general form of
the concept of a partially included Atanassov’s intuitionistic fuzzy relation in which the sgn : R ! R function occurs, where
8
< 1;
> for t > 0
sgnðtÞ ¼ 0; for t ¼ 0 :
>
:
1; for t < 0

Definition 12 (cf. [6, Definition 11]). A relation q = (R, Rd) 2 AIFR(X) is called partially included if

sgnðRðx; zÞ  Rðz; yÞÞ ¼ sgnðRd ðz; yÞ  Rd ðx; zÞÞ; x; y; z 2 X: ð13Þ

92 B. Peßkala / Information Sciences 213 (2012) 84–93

Proposition 9. Let relation q = (R, Rd) 2 AIFR(X),a, b 2 [0, 1], a + b 6 1 and  = ^. If q is partially included and transitive, then
Fa,b(q) is transitive.

Proof. Let q2 6 q and q be partially included, x, y 2 X. From (13) we obtain, for every z 2 X,

ðð1  aÞRðx; zÞ þ að1  Rd ðx; zÞÞÞ ^ ðð1  aÞRðz; yÞ þ að1  Rd ðz; yÞÞÞ
¼ ð1  aÞðRðx; zÞ ^ Rðz; yÞÞ þ aðð1  Rd ðx; zÞÞ ^ ð1  Rd ðz; yÞÞÞ and ðð1  bÞRd ðx; zÞ þ bð1  Rðx; zÞÞÞ _ ðð1  bÞRd ðz; yÞ
þ bð1  Rðz; yÞÞÞ
¼ ð1  bÞðRd ðx; zÞ _ Rd ðz; yÞÞ þ bðð1  Rðx; zÞÞ _ ð1  Rðz; yÞÞÞ:
For q = (R, Rd) 2 AIFR(X) we denote by q
 ; q the fuzzy relations describing it, i.e. q ¼ ðq
 ; qÞ. Then we get
!
_ ^
F 2a;b ðqÞðx; yÞ ¼ F a;b ðqÞðx; yÞ  F a;b ðqÞðx; yÞ ¼ ðF a;b ðx; zÞ ^ F a;b ðz; yÞÞ; ðF a;b ðx; zÞ _ F a;b ðz; yÞÞ
z2X z2X
!
_ ^
¼ ððRðx; zÞ þ apq ðx; zÞÞ ^ ðRðz; yÞ þ apq ðz; yÞÞÞ; ððRd ðx; zÞ þ bpq ðx; zÞÞ _ ðRd ðz; yÞ þ bpq ðz; yÞÞÞ
z2X z2X

_
¼ ððð1  aÞRðx; zÞ þ að1  Rd ðx; zÞÞÞ ^ ðð1  aÞRðz; yÞ þ að1  Rd ðz; yÞÞÞÞ;
z2X
!
^ d d
ððð1  bÞR ðx; zÞ þ bð1  Rðx; zÞÞÞ _ ðð1  bÞR ðz; yÞ þ bð1  Rðz; yÞÞÞÞ :
z2X

From the above considerations we have

_ 
ð1  aÞðRðx; zÞ ^ Rðz; yÞÞ þ aðð1  Rd ðx; zÞÞ ^ ð1  Rd ðz; yÞÞÞ ;
z2X
!!
^ d d

ð1  bÞðR ðx; zÞ _ R ðz; yÞÞ þ bðð1  Rðx; zÞÞ _ ð1  Rðz; yÞÞÞ
z2X

_ _ ^
6 ð1  aÞðRðx; zÞ ^ Rðz; yÞÞ þ aðð1  Rd ðx; zÞÞ ^ ð1  Rd ðz; yÞÞÞ; ð1  bÞðRd ðx; zÞ _ Rd ðz; yÞÞ
z2X z2X z2X
! !
^ _ _ ^ d d
þ bðð1  Rðx; zÞÞ _ ð1  Rðz; yÞÞÞ ¼ ðRðx; zÞ ^ Rðz; yÞÞ  a ðRðx; zÞ ^ Rðz; yÞÞ þ a 1  ðR ðx; zÞ _ R ðz; yÞÞ ;
z2X z2X z2X z2X
!!
^ ^ _
z2X
ðRd ðx; zÞ _ Rd ðz; yÞÞ  b ðRd ðx; zÞ _ Rd ðz; yÞÞ þ b 1  ðRðx; zÞ ^ Rðz; yÞÞ ¼ F a;b ðq2 Þðx; yÞ;
z2X z2X
W V
where q2(x, y) = ( z2X(R(x, z) ^ R(z, y)), z2X(Rd(x, z) _ Rd(z, y))), so we have F 2a;b ðqÞðx; yÞ 6 F a;b ðq2 Þðx; yÞ.
By the isotonicity of Fa,b we obtain Fa,b (q2)(x, y) 6 Fa,b(q)(x, y).
Thus F 2a;b ðqÞðx; yÞ 6 F a;b ðqÞðx; yÞ which finishes the proof. h
Moreover, the following Atanassov’s operators preserve transitivity, what we may prove from Definition 3 and lattice
properties.

Proposition 10. Let q 2 AIFR(X),  = ^, a, b 2 [0, 1], a + b 6 1.

If q is transitive, then
Pa;b ðqÞ; Q a;b ðqÞ; a ðqÞ; a ðqÞ
are transitive.

Proof. For example we will prove the first condition. Let x, y 2 X and q2 AIFR(X) be transitive.
!
2
_ ^ d d
Pa;b ðqÞðx; yÞ ¼ ðmaxða; Rðx; zÞÞ ^ ðmaxða; Rðz; yÞÞÞÞ; ðminðb; R ðx; zÞÞ _ ðminðb; R ðz; yÞÞÞÞ :
z2X z2X

By absorption law and distributivity of _ and ^ we obtain

 B. Peßkala / Information Sciences 213 (2012) 84–93 93

!
_ ^
P2a;b ðqÞðx; yÞ ¼ a_ ðRðx; zÞ ^ Rðz; yÞÞ; b ^ ðRd ðx; zÞ _ Rd ðz; yÞÞ ¼ ða _ R2 ðx; yÞ; b ^ ðRd Þ2 ðx; yÞÞ
z2X z2X

6 ða _ Rðx; yÞ; b ^ Rd ðx; yÞÞ ¼ Pa;b ðqÞðx; yÞ:

The other conditions we may prove in a similar way. h

4. Conclusion

In this work we considered preservation of properties of Atanassov’s intuitionistic fuzzy relations (reflexivity, irreflexiv-
ity, symmetry, pseudo – asymmetry and transitivity) by closures, powers, lattice operations and Atanassov’s operators. We
observed, that some properties are preserved by some Atanassov’s operators and we examined adequate assumptions which
guarantee preservation of other properties. The examined properties may be applied to consider relations connected with
preferences property (see [20]), entropy and similarity measure (see [24]). In our future work we can also consider other
Atanassov’s operators and other properties of Atanassov’s intuitionistic fuzzy relations. Moreover, we can examine more
general composition of Atanassov’s intuitionistic fuzzy relations by adequate additional assumptions on  and 0 .

Acknowledgements

The author thanks Prof. J. Drewniak and Dr. U. Dudziak for helpful suggestions.
Partially supported by the Ministry of Science and Higher Education Grant No. N N519 384936.

References

[1] P.P. Angelov, Optimization in an intuitionistic fuzzy environment, Fuzzy Sets and Systems 86 (1997) 299–306.
[2] K.T. Atanassov, G. Gargov, Interval valued intuitionistic fuzzy sets, Fuzzy Sets and Systems 31 (1989) 343–349.
[3] K.T. Atanassov, Intuitionistic fuzzy sets, Fuzzy Sets and Systems 20 (1986) 87–96.
[4] K. Atanassov, Intuitionistic Fuzzy Sets: Theory and Applications, Springer-Verlag, Heidelberg, 1999.
[5] G. Birkhoff, Lattice Theory, vol. 25, AMS Coll. Publ., Providence, 1967.
[6] P. Burillo, H. Bustince, Intuitionistic fuzzy relations: (Part I), Mathware and Soft Computing 2 (1995) 5–38.
[7] H. Bustince, P. Burillo, Structures on intuitionistic fuzzy relations, Fuzzy Sets and Systems 78 (1996) 293–303.
[8] J.T. Cacioppo, W.L. Gardner, G.G. Berntson, Beyond bipolar conceptualization and measures: the case of attitudes and evaluative spaces, Personality and
Social Psychological Review 1 (1997) 3–25.
[9] G. Deschrijver, E.E. Kerre, On the composition of intuitionistic fuzzy relations, Fuzzy Sets and Systems 136 (2003) 333–361.
[10] G. Deschrijver, E.E. Kerre, On the relationship between some extensions of fuzzy set thory, Fuzzy Sets and Systems 133 (2003) 227–235.
[11] G. Deschrijver, E.E. Kerre, On the position of intuitionistic fuzzy set theory in the framework of theories modelling imprecision, Information Sciences
177 (2007) 1860–1866.
[12] J. Drewniak, K. Kula, Generalized compositions of fuzzy relations, International Journal of Uncertainty. Fuzziness Knowledge-Based Systems 10 (2002)
149–163.
[13] J. Drewniak, B. Peßkala, Properties of fuzzy relations powers, Kybernetika 43 (2007) 133–142.
[14] D. Dubois, S. Gottwald, P. Hajek, J. Kacprzyk, H. Prade, Terminological difficulties in fuzzy set theory – the case of Intuitionistic Fuzzy Sets, Fuzzy Sets
and Systems 156 (2005) 485–491.
[15] T. Gerstenkorn, A. Tepavčević, Lattice valued intuitionistic fuzzy relations and applications, in: K.T. Atanassov et al. (Eds.), Soft Computing Foundations
and Theoretical Aspects, EXIT, Warszawa, 2004, pp. 221–234.
[16] A. Goguen, L-fuzzy sets, Journal of Mathematical Analysis and Applications 18 (1967) 145–174.
[17] A. Kaufmann, Introduction to the Theory of Fuzzy Subsets, Acad. Press, New York, 1975.
[18] D.-F. Li, Multiattribute decision making models and method using intuitionistic fuzzy sets, Journal of Computer and System Sciences 70 (2005) 73–85.
[19] L. Lin, X-H. Yuan, Z-Q. Xia, Multicriteria fuzzy decision-making methods based on intuitionistic fuzzy sets, Journal of Computer and System Sciences 73
(2007) 84–88.
[20] F. Liu, W.-G. Zhang, J.-H. Fu, A new method of obtaining the priority weights from an interval fuzzy preference relation, Information Sciences 185
(2012) 32–42.
[21] A. Pankowska, M. Wygralak, A general concept of IF-Sets with triangular norms, in: K. Atanassov et al. (Eds.), Soft Computing Foundations and
Theoretical Aspects, EXIT, Warszawa, 2004, pp. 319–335.
[22] B. Peß kala, Preservation of properties of interval-valued fuzzy relations, in: J.P. Carvalho et al. (Eds.), Proceedings of IFSA/EUSFLAT 2009, 2009, pp. 1206–
1210.
[23] R. Sambuc, Fonctions /-floues. Application á l’aide au diagnostic en pathologie thyroidienne, Ph.D. Thesis, Universit é de Marseille, France, 1975.
[24] C.-P. Wei, P. Wangb, Y.-Z. Zhang, Entropy, similarity measure of interval-valued intuitionistic fuzzy sets and their applications, Information Sciences
181 (2011) 4273–4286.
[25] L.A. Zadeh, The concept of a linguistic variable and its application to approximate reasoning-I, Information Sciences 8 (1975) 199–249.
[26] L.A. Zadeh, Fuzzy sets, Information and Control 8 (1965) 338–353.