ao

DOI: 10.1515/ms-2021-0005
Math. Slovaca 71 (2021), No. 3, 573–594

L-FUZZY COSETS IN UNIVERSAL ALGEBRAS

Gezahagne Mulat Addis

(Communicated by Anatolij Dvurečenskij )

PY
ABSTRACT. In this paper, we introduce the notion of fuzzy costs in a more general context, in
universal algebra by the use of coset terms. We study the structure of fuzzy cosets by applying the
general theory of algebraic fuzzy systems. Fuzzy cosets generated by a fuzzy set are characterized in
different ways. It is also proved that the class of fuzzy cosets determined by an element forms an
algebraic closure fuzzy set system. Finally, we give a set of necessary and sufficient conditions for a
given variety of algebras to be congruence permutable by applying the theory of fuzzy cosets.

O
2021
c
Mathematical Institute
Slovak Academy of Sciences

C
1. Introduction

Cosets in universal algebra were first introduced by P. Agliano in the paper [Agl1] and later
R
studied by R. Bělohlávek [Bel] under a name ‘convex sets’. Agliano, in his paper [Agl1], has defined
cosets using coset terms. He gives a natural structure to the set of congruence classes containing
a given element of the algebra, and relates the properties of this structure to general features of
O

the variety generated by the algebra. It comes out that such kind of results are better understood
if we consider the set of congruence classes containing a given element as a subset of a generally
richer family of subset of the algebra; called ‘cosets’ of the algebra which in fact is endowed with
TH

the very natural structure of an algebraic lattice. It is observed that in many classical cases cosets
are very well-known structures: in the case of groups left-cosets (or right-cosets) determined by
normal subgroups, in rings cosets determined by ideals.
In 1965, L. A. Zadeh [Zad] introduced the notion of a fuzzy subset of a set X as a function from X
into the unit interval [0, 1]. This idea marked a new direction and stirred the interest of researchers
worldwide. It provided tools and an approach to model imprecision and uncertainty present in
U

phenomena that do not have sharp boundaries. A lot of work on fuzzy sets has come into being
with many applications to various fields such as computer science, artificial intelligence, expert
systems, control systems, decision making, medical diagnosis, management science, operations
A

research, pattern recognition, neural network and others (see [Gup, NeR, Wang, Zim]).
In 1971, Rosenfeld [Ros] applied fuzzy sets in group theory and has formulated the concept of a
fuzzy subgroup of a group. Since then, many researchers have been studying fuzzy subalgebras of
several algebraic structures; fuzzy subgroups and fuzzy normal subgroups of a group (see [AnS1,
AnS2, Das1, DKA1, KuSY, Kum5]), fuzzy action of groups (see [Add3, BoiJ]), fuzzy ideals of a
semigroup (see [Kuro1,Kuro2,Kuro3]), fuzzy subrings and fuzzy ideals of a ring see (see [DKA2,Liu,
MuS,Kum3]), on prime and maximal fuzzy ideals of a ring (see [Kum1,Kum2,KuSY,MaM2,MaM3,
MaM4, SwS]), on fuzzy homomorphisms (see [Add1, ChaK, ChoCK, MaM1]), fuzzy sublattices and

2020 M a t h e m a t i c s S u b j e c t C l a s s i f i c a t i o n: 08A30, 08A72, 08B99.

K e y w o r d s: universal algebras, coset terms, cosets, fuzzy cosets.

573
 GEZAHAGNE MULAT ADDIS

fuzzy ideals of a lattice (see [BoW, SwR3, AlN1, AlN2]), fuzzy ideals of a pseudo-complemented
semi-lattice (see [AlN3]), fuzzy ideals of a poset (see [AlTE1, AlTE2, AlE]), fuzzy ideals and fuzzy
filters of MS-algebras (see [AlAl1, AlAl2, AlAl3, AlAl4]), fuzzy ideals of BCC-algebras (see [DuJ1,
DuJ2]), fuzzy ideals of BCI/BCK algebras ([Meng, JuX, JuLP]), fuzzy submodules of a module
(see [Mak, Zah]), fuzzy subspaces of a vector space (see [Bis, Das2, KatL]), fuzzy subalgebras of
universal algebras (see [FoT, Kura1, Kura2]), etc.
In a series of papers [AlAd1,AlAd2,AlAd3], Alaba and the author of this paper have studied L-
fuzzy ideals in a more general context in universal algebras. On the other hand, in [Add2] we study
fuzzy congruence classes in universal algebras. We gave several Mal’cev type characterizations for
a fuzzy subset µ of an algebra A to be a class of some fuzzy congruence on A. In the present

PY
paper, we continue our study and define L-fuzzy cosets in universal algebras by the use of coset
terms. Fuzzy cosets are natural generalizations of fuzzy ideals and fuzzy congruence classes. We
give several characterizing theorems for fuzzy cosets generated by a fuzzy set. Moreover, we obtain
a representation for fuzzy cosets using finitely generated crisp cosets. For an algebra A of a fixed
type and each a ∈ A, it is observed that the class of all its fuzzy cosets determined by the element

O
a forms an algebraic closure fuzzy set system. Finally, we give a set of necessary and sufficient
conditions for a class of algebras to be congruence permutable, by applying the structural theory
of fuzzy cosets.

C
It was Gougen [Gog] who first realize that the unit interval [0, 1] is insufficient to take truth
values for fuzzy statements. K. L. N. Swamy and U. M. Swamy [SwS] have suggested that complete
lattices satisfying the infinite meet distributive property are the most suitable candidates to take
truth values of the general fuzzy statements. In this paper, we consider complete lattices satisfying
the infinite meet distributive law as the structure of truth degrees for fuzzy statements.
R
O

2. Preliminary
TH

This section contains some notations and preliminary results which will be used in this paper.
For the standard concepts on universal algebras we refer the readers to [BurS,Gra]. Z + (resp. Z0 )
denotes the set of all positive integers (resp. non-negative integers). An algebra A of type F is an
ordered pair (A, F ) where A is a nonempty set and F is a specified set of finitary operations on
A indexed by the language F such that corresponding to each n-ary function symbol f in F there
is an n-ary operation f A on A. The set A is called the universe (or underlying set) of A = (A, F )
U

and the f A ’s are called the fundamental operations of A. For n ∈ Z0 , we denote by Fn , the set of
all n-ary fundamental operations of A. The n-ary terms of type F are formal expressions obtained
in finitely many steps by the following process:
A

(1) The variables x1 , . . . , xn are n-ary terms of type F.

(2) If m ∈ Z0 , t1 , . . . , tm are n-ary terms of type F and f ∈ Fm , then f (t1 , . . . , tm ) is also a term
of type F.
For n ∈ Z + a mapping p : An → A is called an n-ary polynomial over A, if there exist an m ∈ Z0 ,
an (n + m)-ary term t of type F and a1 , . . . , am ∈ A such that

p(x1 , . . . , xn ) = t(x1 , . . . , xn , a1 , . . . , am )

for all x1 , . . . , xn ∈ A. Pn (A) denotes the set of all n-ary polynomials over A. For n ∈ Z + , we
write →−
x to denote the n-tuple hx1 , x2 , . . . , xn i ∈ An .

574
 L-FUZZY COSETS IN UNIVERSAL ALGEBRAS

Definition 2.1 ([Gum]). Let K be a class of algebras having an equationaly definable constant
0. An (n + m)-ary term p(→ −
x,→−
y ) is said to be a K-ideal term (or simply an ideal term) in → −
y if for
each A ∈ K
p(a1 , . . . , an , 0, 0, . . . , 0) = 0
for all a1 , . . . , an ∈ A. A nonempty subset I of A will be called a K-ideal (or simply an ideal) of A
if p(a1 , . . . , an , b1 , . . . , bm ) ∈ I for all a1 , . . . , an ∈ A, b1 , . . . , bm ∈ I and each ideal term p(→
−x,→
−y)
→
−
in y .
The class of all ideals of A, denoted by I(A) is an algebraic lattice. For further detail on ideals
of universal algebras we refer to [Agl1, Agl2, Gum, Urs].

PY
Definition 2.2 ([Agl1]). Let K be a class of algebras of the fixed type F. An (n + m)-ary term
t(→
−
x,→
−
y ) is said to be a K-coset term in →
−
y if the identity
t(x1 , . . . , xn , y, y, . . . , y) ≈ y
holds in K. We denote by CTK the set of all K-coset terms.

O
Note that the fact t(x1 , . . . , xn , y, y, . . . , y) ≈ y holds in K means that for each A ∈ K and every
a1 , . . . , an , b ∈ A it holds that t(a1 , . . . , an , b, b, . . . , b) = b.
Definition
nonempty C ⊆ A it holds that if C is →
terms of CTK .
−
C
2.3 ([Bel]). A subset B ⊆ CTK is called a base of CTK if for each A ∈ K and each
y -closed under all terms of B, then C is →

Example 2.4. For any variety K, the term p(y) = y is a coset term in y.
−
y -closed under all
R
Example 2.5. Let K be a class of groups. Then
(1) p1 (y1 , y2 , y3 ) = y1 y2−1 y3 is a coset term in y1 , y2 , y3 .
O

(2) p2 (x, y1 , y2 , y3 ) = y1 xy2−1 y3 x−1 is a coset term in y1 , y2 , y3 .

The set {p1 , p2 } forms a base for CTK , i.e., all other coset terms in K are the compositions of these
terms only.
TH

Example 2.6. If K is a class of rings, then

(1) p1 (y1 , y2 , y3 ) = y1 + (y2 − y3 ) is a coset term in y1 , y2 , y3 .
(2) p2 (x, y1 , y2 , y3 ) = y1 + x(y2 − y3 ) is a coset term in y1 , y2 , y3 .
The set {p1 , p2 } forms a base for CTK , i.e., all other coset terms in K are the compositions of these
U

terms only.
Example 2.7. If K is a class of lattices, then
A

(1) p1 (y1 , y2 ) = y1 ∨ y2 is a coset term in y1 , y2 .

(2) p2 (y1 , y2 ) = y1 ∧ y2 is a coset term in y1 , y2 .
(3) p2 (x, y1 , y2 ) = y1 ∧ (y2 ∨ x) is a coset term in y1 , y2 .
The set {p1 , p2 , p3 } forms a base for CTK , i.e., all other coset terms in K are the compositions of
these terms only.
Definition 2.8 ([Agl1]). Let A ∈ K. A nonempty subset H of A is called a K-coset of A if
t(a1 , . . . , an , b1 , . . . , bm ) ∈ H
for all a1 , . . . , an ∈ A, b1 , . . . , bm ∈ H and each coset term t(→ −x,→
−
y ) in →
−
y . For a ∈ A, we say H is
a coset of A determined by a, to mean H is a coset of A to which a belongs.

575
 GEZAHAGNE MULAT ADDIS

Remark. If K is obvious from context (or is not important), we drop the prefix “K” and we use
only “coset terms” and “cosets” instead of “K-coset terms” and “K-cosets” respectively.
Example 2.9. (1) For any K and A ∈ K, every singleton {a} in A is a coset of A.
(2) Cosets in a group G are the usual left cosets as well as right cosets in G formed by a normal
subgroup. Similarly, cosets of a ring R are those left cosets of R formed by an ideal.
(3) Cosets in a lattice L are those convex sublattices of L.
For each a ∈ A, the set of all cosets of A determined by a is denoted by Ca (A). It is observed
that Ca (A) is closed under arbitrary intersection. So that, for any subset X of A, always there

PY
a
exists the smallest coset of A determined by a containing X. It is denoted by X .
By an L-fuzzy subset of A, we mean a mapping µ : A → L, where L is a non-trivial complete
Brouwerian lattice (L, ∧, ∨, 0, 1); i.e., a complete lattice satisfying the infinite meet distributive
law:  _  _
α∧ β = (α ∧ β)

O
β∈M β∈M
for all α ∈ L and any M ⊆ L. For each α ∈ L, the α-level set of µ denoted by µα is a subset of A
given by:
µα = {x ∈ A : α ≤ µ(x)}.

C
By a normal fuzzy subset of A, we mean a fuzzy subset µ of A such that µ(a) = 1 for some a ∈ A.
For more detail on fuzzy set theory, we refer to [Gog, MoBR, MoM, Zad].
R
3. Fuzzy cosets

In this section, we define fuzzy cosets in universal algebras and investigate some of their prop-
O

erties. The following concept of a → − y -closed subset of an algebra was introduced by R. Bělohlávek
[Bel].
→
−
TH

Definition 3.1. A subset H of A is said to be y closed under the (n + m)-ary term operation
→
− →
−
p( x , y ) if for all a , . . . , a , b , . . . , b ∈ A,
1 n 1 m

b1 , . . . , bm ∈ H ⇒ p(a1 , . . . , an , b1 , . . . , bm ) ∈ H.
Analogous to this concept, we define the following in a fuzzy setting.
→
−
Definition 3.2. A fuzzy subset µ of A is said to be y closed under the (n+m)-ary term operation
U

→
− →
−
p( x , y ) if
µ(p(a1 , . . . , an , b1 , . . . , bm )) ≥ µ(b1 ) ∧ · · · ∧ µ(bm )
for all a1 , . . . , an , b1 , . . . , bm ∈ A.
A

We use this concept to define fuzzy cosets.

Definition 3.3. An L-fuzzy subset µ of A is said to be an L-fuzzy coset of A (or shortly a fuzzy
coset of A) if and only if the following conditions are satisfied:
(1) µ(a) = 1 for some a ∈ A, and
(2) µ is →
−
y -closed under each coset term t(→
−
x,→
−
y ) in →
−
y . In the sense that, if t(→
−
x,→
−
y ) is a coset
→
−
term in y , then
µ(t(a1 , a2 , . . . , an , b1 , b2 , . . . , bm )) ≥ µ(b1 ) ∧ · · · ∧ µ(bm )
for all a1 , a2 , . . . , an , b1 , b2 , . . . , bm ∈ A.

576
 L-FUZZY COSETS IN UNIVERSAL ALGEBRAS

In this case, we say that µ is a fuzzy coset of A determined by a. For each a ∈ A, we denote by
F Ca (A) the set of all fuzzy cosets of A determined by a.
Example 3.4. (1) Fuzzy cosets of a group G determined by an element a ∈ G are exactly of
the form ηa , where ηa is a fuzzy subset of G defined by:
ηa (x) = η(xa−1 ) for all x ∈ G
where η is a fuzzy normal subgroup of G.
(2) Fuzzy cosets of a ring R determined by an element a ∈ R are exactly of the form µa , where
µa is a fuzzy subset of R defined by:

PY
µa (x) = µ(x − a) for all x ∈ R
where µ is a fuzzy ideal of R.
For a given group hG, ·i, Rosenfield [Ros] has defined a binary operation (called the product)
on the class of all fuzzy subsets of G in the following way:

O
(µ ◦ ν)(x) = sup{µ(x1 ) ∧ ν(x2 ) : x1 x2 = x}.
Rosenfield use this product to characterize fuzzy subgroups. Analogous to this concept, we define
the following and give a Rosenfield type characterization for fuzzy subsets to be a fuzzy coset.

A define
→
−
C
Definition 3.5. Let p( x ) be an n-ary term operation on A. For each fuzzy subsets η , . . . , η of

p : LA × · · · × LA → LA
1 n
R
p(η1 , . . . , ηn ) 7→ η
where, _
η(x) = {η1 (x1 ) ∧ · · · ∧ ηn (xn ) : p(x1 , . . . , xn ) = x}.
O

Supremum being taken over all n-tuples (x1 , . . . , xn ) ∈ An for which p(x1 , . . . , xn ) = x. If there
exists no such n-tuples, η(x) is defined to be 0.
3.6. Let a ∈ A and µ a fuzzy subset of A. Then, µ ∈ F Ca (A) if and only if
TH

Theorem

(1) µ(a) = 1, and

(2) For each fuzzy subsets η1 , . . . , ηn of A, and each (n + m)-ary coset term p(→ −x,→
−y ) in → −y it
holds that
p(η1 , . . . , ηn , µ, µ, . . . , µ) ≤ µ.
U

P r o o f. Suppose that µ ∈ F Ca (A). Let p(→ −x,→ −y ) be an (n+m)-ary coset term in → −

y and η1 , . . . , ηn
be arbitrary fuzzy subsets of A. For simplicity, let us put η = p(η1 , . . . , ηn , µ, µ, . . . , µ). For x ∈ A,
if there is no (n+m)-tuple (x1 , . . . , xn , y1 , . . . , ym ) ∈ An+m such that x = p(x1 , . . . , xn , y1 , . . . , ym ),
A

then η(x) = 0 and hence η(x) ≤ µ(x). Assume that there are such (n + m)-tuples. Let
(a1 , . . . , an , b1 , . . . , bm ) ∈ An+m such that x = p(a1 , . . . , an , b1 , . . . , bm ). Since p(→
−
x,→−
y ) is a coset
→
−
term in y and µ is a fuzzy coset, we get
µ(x) ≥ µ(b1 ) ∧ · · · ∧ µ(bm )
≥ η1 (a1 ) ∧ · · · ∧ ηn (an ) ∧ µ(b1 ) ∧ · · · ∧ µ(bm ).
This implies that
_
µ(x) ≥ {η1 (x1 ) ∧ · · · ∧ ηn (xn ) ∧ µ(y1 ) ∧ · · · ∧ µ(ym ) : p(x1 , . . . , xn , y1 , . . . , ym ) = x} = η(x).
Thus, η ≤ µ. Conversely suppose that the above two conditions are satisfied. We show that µ
is →
−
y closed under each (n + m)-ary coset term p(→
−
x,→
−
y ) in →
−
y . Let p(→
−
x,→−y ) be an (n + m)-ary

577
 GEZAHAGNE MULAT ADDIS

coset term in →−y , and let a1 , . . . , an , b1 , . . . , bm ∈ A. Put x = p(a1 , . . . , an , b1 , . . . , bm ). For each

i, 1 ≤ i ≤ n define fuzzy subsets ηi of A by:
(
1 if x = ai
ηi (z) =
0 otherwise
for all z ∈ A. Again put η = p(η1 , . . . , ηn , µ, µ, . . . , µ). By condition (2), we have η(x) ≤ µ(x),
which implies that µ(x) ≥ µ(b1 ) ∧ · · · ∧ µ(bm ). Therefore µ ∈ F Ca (A). Hence proved. 

The following theorem gives an equivalent condition for fuzzy subsets to be a fuzzy coset in

PY
terms of their level sets.
Theorem 3.7. Let a ∈ A. A fuzzy subset µ of A is a fuzzy coset of A determined by a if and only
if µα is a coset of A containing a for all α ∈ L.

P r o o f. Suppose that µ is a fuzzy coset of A determined by a. Then µ(a) = 1. So a ∈ µα for all

α ∈ L. Also, for any α ∈ L, let a1 , . . . , an ∈ A, b1 , . . . , bm ∈ µα and t(→
−
x,→
−
y ) be a coset term in →
−
y.

O
Since µ is a fuzzy coset, it follows that
µ(t(a1 , . . . , an , b1 , . . . , bm )) ≥ µ(b1 ) ∧ · · · ∧ µ(bm ) ≥ α

C
Then t(a1 , . . . , an , b1 , . . . , bm ) ∈ µα and hence each µα is a coset of A containing a. Conversely,
suppose that the level subset µα is a coset of A containing a for all α ∈ L. In particular µα is
a coset of A containing a for α = 1. So that µ(a) = 1. Let t(→ −x,→ −y ) be a coset term in → −
y and
a1 , . . . , an , b1 , . . . , bm ∈ A. Put α = µ(b1 ) ∧ · · · ∧ µ(bm ). Then, b1 , . . . , bm ∈ µα . Since each µα is a
R
coset, we get t(a1 , . . . , an , b1 , . . . , bm ) ∈ µα . So that
µ(t(a1 , . . . , an , b1 , . . . , bm )) ≥ α = µ(b1 ) ∧ · · · ∧ µ(bm ).
O

Therefore µ is a fuzzy coset of A determined by a. 

This theorem confirms that a fuzzy coset of A determined by a is precisely a fuzzy L-subset of
A (in the sense of [SwR1]), where L is the set of all cosets of A containing a.
TH

Definition 3.8. For a subset H ⊆ A and each α ∈ L − {1}, define a fuzzy subset αH of A by
(
1 if x ∈ H
αH (x) =
α otherwise
for all x ∈ A.
U

Lemma 3.9. Let a ∈ A and H ⊆ A. Then, H is a coset of A containing a if and only if for any
α ∈ L − {1}, αH is a fuzzy coset of A determined by a.
A

Corollary 3.10. Let a ∈ A. A subset H of A is a coset of A containing a if and only if its

characteristic function χH is a fuzzy coset of A determined by a.
Lemma 3.11. If K has a constant 0 and A ∈ K, then every fuzzy ideal of A is a fuzzy coset
determined by 0.

P r o o f. It is enough to show that every coset term is an ideal term. 

a
Lemma 3.12. Let a ∈ A and µ ∈ F Ca (A). Then, for any a1 , . . . , am ∈ A, if x ∈ {a1 , . . . , am } ,
a
then µ(x) ≥ µ(a1 ) ∧ · · · ∧ µ(am ). More generally, for any nonempty subset S of A, if x ∈ S , then
there exist a1 , . . . , am ∈ S such that µ(x) ≥ µ(a1 ) ∧ · · · ∧ µ(am ).

578
 L-FUZZY COSETS IN UNIVERSAL ALGEBRAS

a
P r o o f. Suppose that x ∈ {a1 , . . . , am } . Then, x = p(b1 , . . . , bn , a1 , . . . , am ) for some b1 , . . . , bn ∈
A and some coset term p(→ −
x,→−y ) in → −y . So we have the following:
µ(x) = µ(p(b1 , . . . , bn , a1 , . . . , am )) ≥ µ(a1 ) ∧ · · · ∧ µ(am ).
Hence proved. 
Theorem 3.13. Let a ∈ A. Then, µ ∈ F Ca (A) if and only if for each m ≥ 0 and each
a
b1 , b2 , . . . , bm ∈ A, if x ∈ {a1 , . . . , am } , then µ(x) ≥ µ(b1 ) ∧ · · · ∧ µ(bm ).
P r o o f. One part of this theorem is proved in the above Lemma. So we proceed to the converse
part. Assume the given condition is satisfied for µ. Let us put Sm = {b1 , . . . , bm }. If we take

PY
a
m = 0, then Sm = ∅ and it is known that ∅ = {a}. So by our assumption, we have
^
µ(a) ≥ µ(b) = 1.
b∈∅

Thus µ(a) = 1. Let a1 , . . . , an , b1 , . . . , bm ∈ A and p(→

−
x,→
−
y ) be a coset term in →
−
y . If we consider

O
the set Sm = {b1 , . . . , bm }, then one can observe that
a
p(a1 , . . . , an , b1 , . . . , bm ) ∈ {b1 , . . . , bm } .
It follows from our assumption that

Therefore µ ∈ F Ca (A). Hence proved.

C
µ(p(a1 , . . . , an , b1 , . . . , bm )) ≥ µ(b1 ) ∧ · · · ∧ µ(bm ).

In the following theorem, we give a more general setting to characterize fuzzy cosets.

R
Theorem 3.14. Let a ∈ A. Then, µ ∈ F Ca (A) if and only if for any subset S of A
^ a
µ(b) ≥ µ(x) for all b ∈ S .
O

x∈S
a
P r o o f. Suppose that µ ∈ F Ca (A). If S = ∅, then S = {a} and the condition holds trivially.
a
TH

Assume that S is nonempty and let b ∈ S . Then b = t(a1 , . . . , an , b1 , . . . , bm ) for some b1 , . . . , bm ∈

S, a1 , . . . , an ∈ A and some coset term t(→−x,→ −y ) in →
−y . Since µ is fuzzy a fuzzy coset, it follows
that ^
µ(b) ≥ µ(b1 ) ∧ · · · ∧ µ(bm ) ≥ µ(x).
x∈S
The converse part follows from the above theorem by assuming the condition for finite sets. 
U

4. Fuzzy cosets generated by a fuzzy set

A

In this section, we characterize fuzzy cosets generated by a fuzzy set from the fuzzy point of
view (using level sets) as well as from the algebraic view point (using coset terms).
Theorem 4.1. Let a ∈ A. If {µi }i∈∆ is a family of fuzzy cosets of A determined by a, then
∩i∈∆ µi is a fuzzy coset of A determined by a.
This theorem confirms that, for each a ∈ A and any fuzzy subset λ of A, always there exists a
smallest fuzzy coset determined by a containing λ which we call the fuzzy coset of A determined
a
by a generated by λ and is denoted by λ . Note also that for a subset X of A and a ∈ X, we
a
denote by X the coset of A containing a generated by X.

579
 GEZAHAGNE MULAT ADDIS

Lemma 4.2. Let a ∈ A, S ⊆ A and a ∈ S. Then χS a = χS a .

P r o o f. We show that χS a is the smallest fuzzy coset of A determined by a such that χS ≤ χS a .

a
Since S is a coset of A containing a, it follows from Corollary 3.10 that χS a is a fuzzy coset of A
determined by a. It is also clear that χS ≤ χS a . Let λ be any fuzzy coset of A determined by a
such that χS ≤ λ. Then λ(s) = 1 for all s ∈ S and hence λ(a) = 1. Let z be any element in A. If
/ S , then χS a (z) = 0 ≤ λ(z). I z ∈ S , then z = t(→
−a ,→
−s ) for some →
−
a ∈ An , →
−
a a
z∈ s ∈ S m and some
→
− →
− →
−
coset term t( x , y ) in y . Now consider:
λ(z) = λ(t(a1 , . . . , an , s1 , . . . , sm )) ≥ λ(s1 ) ∧ · · · ∧ λ(sm ). = 1

PY
So that χS a ≤ λ. Therefore χS a = χS a . 

For any fuzzy subset λ of A, recall from [KoNL] that

_
λ(x) = {α ∈ L : x ∈ λα }
for all x ∈ A. In the following theorem, we characterize fuzzy cosets generated by a fuzzy set in

O
terms of their level sets.
Theorem 4.3. Let a ∈ A. For a fuzzy subset λ of A, let Λa1 be a fuzzy subset of A defined by
a

Then Λa1 = λ .
a
_

C
Λa1 (x) = {α ∈ L : x ∈ (λα ) } for all x ∈ A.

P r o o f. We show that Λa1 is the smallest fuzzy coset of A determined by a containing λ. Let us
R
first show that Λa1 is a fuzzy coset. It is clear that Λa1 (a) = 1. Let a1 , . . . , an , b1 , . . . , bm ∈ A and
t(→
−x,→−y ) be a coset term in → −y . Then consider
_ a _ a
Λa1 (b1 ) ∧ · · · ∧ Λa1 (bm ) = {α1 ∈ L : b1 ∈ (λα1 ) } ∧ · · · ∧ {αm ∈ L : bm ∈ (λαm ) }
O

m
_n ^ a
o
= αi ∈ L : bi ∈ (λαi ) .
i=1
TH

a m
V
Let α1 , . . . , αm ∈ L such that bi ∈ (λαi ) for all i. If we put β = αi , then β ∈ L such that
i=1
a a
b1 , . . . , bm ∈ (λβ ) . Moreover, (λβ ) being a coset of A, we get that t(a1 , . . . , an , b1 , . . . , bm ) ∈
a
(λβ ) . Now it follows that
m
U

_n ^ a
o
a a
Λ1 (b1 ) ∧ · · · ∧ Λ1 (bm ) = αi ∈ L : bi ∈ (λαi )
i=1
_ a
≤ {β ∈ L : b1 , . . . , bm ∈ (λβ ) }
A

_ a
≤ {β ∈ L : t(a1 , . . . , an , b1 , . . . , bm ) ∈ (λα ) }
= Λa1 (t(a1 , . . . , an , b1 , . . . , bm )).
Therefore Λa1 is a fuzzy coset of A determined by a. It is also clear to see that λ ≤ Λa1 . Suppose
a
that µ is any other fuzzy coset of A determined by a such that λ ≤ µ. Then (λα ) ⊆ µα for all
α ∈ L. Now for any x ∈ A consider
_ a
Λa1 (x) = {α ∈ L : x ∈ (λα ) }
_
≤ {α ∈ L : x ∈ µα } = µ(x).

580
 L-FUZZY COSETS IN UNIVERSAL ALGEBRAS

a
Therefore Λa1 is the smallest fuzzy coset of A determined by a containing λ. Thus Λa1 = λ . 

Corollary 4.4. Let a ∈ A and µ be a fuzzy subset of A. Then

a
(µα ) ⊆ (µa )α for all α ∈ L.
Moreover, if L is a chain and µ is finite valued or equivalently if µ has sup property, then the
equality holds.
Theorem 4.5. Let µ be a fuzzy subset of A and α ∈ L:
[n \ a
o
(µa )α = (µγ ) : M ⊆ L and α ≤ sup M .

PY
γ∈M

P r o o f. Let us put
[n \ a
o
H= (µγ ) : M ⊆ L and α ≤ sup M .
γ∈M

O
T a a
If x ∈ H, then x ∈ (µγ ) for some M ⊆ L with α ≤ sup M ; i.e., x ∈ (µγ ) for all γ ∈ M and
γ∈M
α ≤ sup M . By Theorem 4.3, we have the following
_
Ca
µa (x) = {β ∈ L : x ∈ (µβ ) }.

So that µa (x) ≥ γ for all γ ∈ M . This gives µa (x) ≥ α. Thus x ∈ (µa )α and hence H ⊆ (µa )α . To
prove the other inequality, let us take x ∈ (µa )α . Then
R
_ a
{β ∈ L : x ∈ (µβ ) } ≥ α.
a a
If we put M = {β ∈ L : x ∈ (µβ ) }, then M ⊆ L such that α ≤ sup M and x ∈ (µγ ) for all γ ∈ M .
O

T a
This means that x ∈ (µγ ) and α ≤ sup M . Thus x ∈ H, and hence the proof ends. 
γ∈M
TH

In the following we give an algebraic characterization of fuzzy cosets generated by fuzzy sets.
Definition 4.6. Let a ∈ A. For a fuzzy subset λ of A, let us define a fuzzy subset Λa2 of A as
follows: Λa2 (a) = 1 and for a 6= x ∈ A
_n
Λa2 (x) = λ(b1 ) ∧ · · · ∧ λ(bm ) : b1 , . . . , bm ∈ A, t(a1 , . . . an , b1 , . . . , bm ) = x
U

where a1 , . . . an ∈ A, and t(→ −x,→ −y ) is a coset term in → −

o
y .
a
Theorem 4.7. Let a ∈ A. For a fuzzy subset λ of A, we have Λa2 = λ .
A

P r o o f. By Theorem 4.3, it is enough to show that Λa2 = Λa1 . For each x 6= a in A, let us define
two sets Hx and Gx as follows:

Hx = λ(b1 ) ∧ · · · ∧ λ(bm ) : t(a1 , . . . an , b1 , . . . , bm ) = x for some a1 , . . . an ∈ A
and some coset term t(→ −x,→ −y ) in → −
y ,
 a
Gx = α ∈ L : x ∈ (λα ) .
Clearly both Hx and Gx are subsets of L. Our claim is to show that
_ _
{α : α ∈ Hx } = {α : α ∈ Gx }.

581
 GEZAHAGNE MULAT ADDIS

We first show that Hx ⊆ Gx . If α ∈ Hx , then α = λ(b1 ) ∧ · · · ∧ λ(bm ), for some b1 , . . . , bm ∈ A,

such that t(a1 , . . . , an , b1 , . . . , bm ) = x for some a1 , . . . , an ∈ A where t(→
−
x,→
−
y ) is a coset term in
→
− a
y . That is, b1 , . . . , bm ∈ λα and so that x ∈ λα . Then α ∈ Gx and hence Hx ⊆ Gx . Thus
_ _
{α : α ∈ Hx } ≤ {α : α ∈ Gx }.
To prove the inequality, we show that for each α ∈ Gx , there exists β ∈ Hx such that α ≤ β.
a
Let α ∈ Gx . Then x ∈ λα ; that is, x = t(a1 , . . . , an , b1 , . . . , bm ) for some b1 , . . . , bm ∈ λα and
a1 , . . . , an ∈ A where t(→
−
x,→
−
y ) is a coset term in →
−
y . If we put β = λ(b1 ) ∧ · · · ∧ λ(bm ), then we get
β ∈ Hx and α ≤ β. This completes the proof. 

PY
In the following theorem, we give a finite characterization for fuzzy cosets.
Theorem 4.8. Let a ∈ A. For a fuzzy subset λ of A, let us define a fuzzy subset Λa3 of A by
_n ^ a
o
Λa3 (x) = λ(y) : x ∈ F , F ⊂⊂ A
y∈F
a

O
for all x ∈ A. Then Λa3 = λ .
P r o o f. It is enough if we show that Λa3 = Λa1 . Clearly Λa3 (a) = 1. For each a 6= x ∈ A, let us take
the set Gx as in Theorem 4.7 and define a set Hx as follows:

Our claim is to show that

Hx =

_
n ^

a∈F

{α : α ∈ Hx } =
C _
a
λ(a) : x ∈ x ∈ F , F ⊂⊂ A .

{α : α ∈ Gx }.
o
R
V a
We first show that Hx ⊆ Gx . α ∈ Hx , implies that α = λ(a) and x ∈ F , for some finite
a∈F
a a
O

subset F of A. That is, a ∈ λα for all a ∈ F and x ∈ F . So that x ∈ λα . Then α ∈ Gx and

hence Hx ⊆ Gx . Next we show that, for each α ∈ Gx , there exists β ∈ Hx such that α ≤ β. Let
a
α ∈ Gx . Then x ∈ λα , which implies that x = t(a1 , . . . , an , b1 , . . . , bm ) for some b1 , . . . , bm ∈ λα
and a1 , . . . an ∈ A where t(→
−
x,→
−
y ) is a coset term in → −y . If we put β = λ(b1 ) ∧ · · · ∧ λ(bm ), then
TH

a
α ≤ β. Moreover, if we put F = {b1 , b2 , . . . , bm }, then F is a finite subset of A such that x ∈ F .
Thus β ∈ Hx such that α ≤ β. This completes the proof. 
Theorem 4.9. Let a ∈ A. Suppose that {Hα }α∈L is a subfamily of Ca (A) such that
\
Hα = Hsup M
U

α∈M
for all M ⊆ L. Then, there is a unique fuzzy coset µ of A determined by a such that µα = Hα for
all α ∈ L.
A

P r o o f. We first show that the map α 7→ Hα is antitone; in the sense that, for each α, β ∈ L,
α ≤ β ⇒ Hβ ⊆ Hα . Let α, β ∈ L such that α ≤ β. Put M = {α, β}. Then sup M = β. By our
hypothesis Hα ∩ Hβ = Hsup M = Hβ . So that Hβ ⊆ Hα and hence the map α 7→ Hα is antitone.
Define a fuzzy subset µ of A by
_
µ(x) = {α ∈ L : x ∈ Hα }
for all x ∈ A. Clearly µ is well defined. Our first aim is to show that µα = Hα for all α ∈ L. The
inclusion Hα ⊆ µα follows easily from the definition of µ. To prove the other inclusion, let x ∈ µα .
Then µ(x) ≥ α, i.e., _
{γ ∈ L : x ∈ Hγ } ≥ α.

582
 L-FUZZY COSETS IN UNIVERSAL ALGEBRAS

If we put M = {γ ∈ L : x ∈ Hγ }, then M ⊆ L such that α ≤ sup M and x ∈ Hγ for all γ ∈ M ,

i.e.,
\
x∈ Hγ .
γ∈M

By our assumption it follows that x ∈ Hsup M . Since α ≤ sup M and the map α 7→ Hα is antitone
we get x ∈ Hα . Thus Hα = µα . This means that µ is a fuzzy subset of A for which its α-level sets
are Hα ’s. Moreover, each Hα being an coset of A containing a, it follows from Theorem 3.7 that
µ ∈ F Ca (A). The uniqueness of µ follows from the fact µα = Hα for all α ∈ L. 

PY
5. The lattice of fuzzy cosets

Let us first recall some important results from [SwR1] and [Mur1]. Let C be a nonempty
collection of fuzzy subsets of a nonempty set X.

O
Definition 5.1. C is said to be a closure system in LX if it is closed under arbitrary intersection
of fuzzy sets, i.e., if for any subcollection D of fuzzy subsets of X in C it holds that

µ∈D
^

Cµ ∈ C.

A closure system of fuzzy sets is also known as the “Moor family” of fuzzy sets.
R
Remark. If C is a closure system in LX , then it contains the fuzzy subset 1X of X. This is because
the fuzzy set 1X can be expressed as the infimum of an empty collection of fuzzy subsets of X.
Theorem 5.2. If C is a closure system in LX , then (C, ≤) forms a complete lattice, where ≤ is
O

the inclusion ordering of fuzzy sets.

Definition 5.3. A nonempty collection C of fuzzy subsets of X is called inductive if every
TH

nonempty chain in C has a supremum in C.

Definition 5.4. A nonempty collection C of fuzzy subsets of X is said to be an algebraic closure
fuzzy set system if it is an inductive closure fuzzy set system in LX .

As observed in Theorem 4.1, for each a ∈ A the class F Ca (A) is closed under arbitrary intersec-
tion of fuzzy sets. So that (F Ca (A), ≤) forms a closure fuzzy set system and hence by Theorem 5.2
U

it is a complete lattice, where ≤ is a pointwise ordering of fuzzy sets. The following theorem
summarizes this.
A

Theorem 5.5. Let a ∈ A. Then the set F Ca (A) of all fuzzy cosets of A determined by a forms
a complete lattice where the infimum and supremum of any sub-family {µi : i ∈ ∆} of F Ca (A) is
given by
^ _ a
µi = ∩µi and µi = (∪µi ) .

Theorem 5.6. (F Ca (A), ≤) forms an algebraic closure fuzzy set system in LA .

P r o o f. By Definition 5.4, it is enough to show that F Ca (A) is inductive in LA . Let {µi }i∈∆ be
a chain in F Ca (A). Let us put
[
η= µi .
i∈∆

583
 GEZAHAGNE MULAT ADDIS

We show that η is a fuzzy coset of A determined by a. Clearly η(a) = 1. Let a1 , . . . , an , b1 , . . . , bm ∈

A and p(→ −
x,→−
y ) be a coset term in → −
y . First observe that, for each m-tuples i1 , . . . , im ∈ ∆, there
exists k ∈ {1, 2, . . . , m} such that µij ≤ µik for all j ∈ {1, 2, . . . , m}. Now consider
 _   _ 
η(b1 ) ∧ · · · ∧ η(bm ) = µi1 (b1 ) ∧ · · · ∧ µim (bm )
i1 ∈∆ im ∈∆
_
= (µi1 (b1 ) ∧ · · · ∧ µim (bm ))
i1 ,...,im ∈∆
_
≤ (µik (b1 ) ∧ · · · ∧ µik (bm ))

PY
ik ∈∆
_
≤ µik (p(a1 , . . . , an , b1 , . . . , bm ))
ik ∈∆
= η(p(a1 , . . . , an , b1 , . . . , bm )).
Therefore η is a fuzzy coset of A determined by a and this completes the proof. 

O
Remark. If L is an algebraic lattice, then F Ca (A) forms an algebraic lattice.

C
6. Fuzzy cosets and fuzzy congruences
R
Fuzzy congruence relations in universal algebras have been studied by different scholars (see
[Chi, Mur1, Mur2, SwR1]). The following definition is taken from [SwR1].
Definition 6.1. A fuzzy relation Θ on A is said to be:
O

(1) reflexive if: Θ(x, x) = 1 for all x ∈ A

(2) symmetric if: Θ(x, y) = Θ(y, x) for all x, y ∈ A
TH

(3) transitive if for each x, z ∈ A: Θ(x, z) ≥ Θ(x, y) ∧ Θ(y, z) for all y ∈ A.

A reflexive, symmetric and transitive fuzzy relation on A is called a fuzzy equivalence relation
on A.
Definition 6.2. A fuzzy relation Θ on A is said to be compatible, if
Θ(f A (x1 , x2 , . . . , xn ), f A (y1 , y2 , . . . , yn )) ≥ Θ(x1 , y1 ) ∧ · · · ∧ Θ(xn , yn )
U

for every n ∈ Z + , f ∈ Fn and all x1 , x2 , . . . , xn , y1 , y2 , . . . , yn ∈ A.

A

This fact is often expressed in the way that Θ is said to have the substitution property with
respect to each fundamental operations f ∈ F. According to Werner [Wer], compatible fuzzy
relations may also be refereed as admissible fuzzy relations.
Definition 6.3. A fuzzy congruence on A is a fuzzy equivalence relation on A which is compatible
with all fundamental operations of A. We denote by F Con(A) the class of all fuzzy congruence
relations on A.

If t(x1 , . . . , xm ) is an m-ary term operation on A and Θ ∈ F Con(A), then it holds that

Θ(t(a1 , . . . , am ), t(b1 , . . . , bm )) ≥ Θ(a1 , b1 ) ∧ · · · ∧ Θ(am , bm )
for all a1 , . . . , am , b1 , . . . , bm ∈ A.

584
 L-FUZZY COSETS IN UNIVERSAL ALGEBRAS

For a fuzzy relation Θ on A and each α ∈ L, remember that

Θα = {(x, y) ∈ A × A : Θ(x, y) ≥ α}
is the α-level relation of Θ on A. The following lemma is simple but a useful characterization of
fuzzy congruences using their level relations.
Lemma 6.4. Let Θ be a fuzzy relation on A. Then the following hold:
(1) Θ is a fuzzy equivalence relation on A if and only if Θα is an equivalence relation on A for
all α ∈ L.
(2) Θ is a fuzzy congruence relation on A if and only if Θα is a congruence relation on A for all

PY
α ∈ L.
From the above theorem, one can conclude that fuzzy congruence relations are fuzzy L-subsets
of A × A (in the sense of [SwR1]), where L- is a the class of congruence relations on A.
Definition 6.5 ([Mur1, Mur2]). For any two fuzzy relations Θ and Φ on A, their composition
Θ ◦ Φ is a fuzzy relation on A given by

O
_
Θ ◦ Φ(x, y) = {Θ(x, z) ∧ Φ(y, z) : z ∈ A}
for all x, y ∈ A. For a positive integer n, by Θn , we mean Θ ◦ Θ ◦ · · · ◦ Θ (n copies).

by C
Remember also that for a binary relation R on A and S ⊆ A, the set R[S] is a subset of A given

R[S] = {y ∈ A : (x, y) ∈ R for some x ∈ S}.

In particular, if S = {a}, then we write R[a] instead of R[S]. If θ is a congruence on A (in the
R
usual sense) and a ∈ A, then the set
θ[a] = {b ∈ A : (a, b) ∈ θ}
O

is called the congruence class of θ determined by a. Analogous to this classical concept, we define
the following.
Definition 6.6. For a fuzzy subset µ of A and a fuzzy relation Θ on A we define [µ]Θ to be a
TH

fuzzy subset of A as follows:

_
[µ]Θ (x) = {µ(y) ∧ Θ(x, y) : y ∈ A}
for all x ∈ A. Also for each a ∈ A, we define [a]Θ to be [χ{a} ]Θ .
It is observed that
U

[a]Θ (x) = Θ(a, x) for all x ∈ A.

If Θ is a fuzzy congruence on A and a ∈ A, then we call [a]Θ the fuzzy congruence class of A
determined by Θ and a. For each a ∈ A, we denote by F La (A) the following set:
A

F La (A) = {[a]Θ : Θ ∈ F Con(A)}.

It was proved by A. I. Mal’cev [Mal] that a nonempty subset H of A is a class of some θ ∈ Con(A)
if and only if for each a, b ∈ H
p(a) ∈ H ⇒ p(b) ∈ H
for each unary polynomial p of A. Analogous to this well known characterization, we state and
prove the following lemma in the fuzzy sense in the paper [Add2].
Lemma 6.7. A normal fuzzy subset µ of A is a class of some fuzzy congruence on A if and only if
µ(a) ∧ µ(b) ∧ µ(p(a)) = µ(p(b)) ∧ µ(a) ∧ µ(b)
for each a, b ∈ A and each p ∈ P1 (A).

585
 GEZAHAGNE MULAT ADDIS

The following lemma shows that fuzzy congruence classes are fuzzy cosets.
Lemma 6.8. For each a ∈ A,
F La (A) ⊆ F Ca (A).

P r o o f. Let Θ ∈ F Con(A). We show that [a]Θ ∈ F Ca (A). Clearly [a]Θ (a) = 1.

Let a1 , . . . , an , b1 , . . . , bm ∈ A and t(→
−
x,→
−
y ) be a coset term in →
−
y . Consider
[a]Θ (b1 ) ∧ · · · ∧ [a]Θ (bm ) = Θ(a, b1 ) ∧ · · · ∧ Θ(a, bm )
≤ Θ(t(a1 , . . . , an , a, a, . . . , a), t(a1 , . . . , an , b1 , . . . , bm ))

PY
= Θ(a, t(a1 , . . . , an , b1 , . . . , bm ))
= [a]Θ (t(a1 , . . . , an , b1 , . . . , bm )).
Therefore [a]Θ ∈ F Ca (A). Hence proved. 

Theorem 6.9. Let a ∈ A. If Θ ∈ F Con(A) and µ ∈ F Ca (A), then [µ]Θ ∈ F Ca (A).

O
P r o o f. Since µ ∈ F Ca (A), we have µ(a) = 1. Now consider
_
[µ]Θ (a) = {µ(b) ∧ Θ(a, b) : b ∈ A}

Thus [µ]Θ (a) = 1. Let t(→ −x,→

−
= 1.
y ) be a coset term in →
−
C
≥ µ(a) ∧ Θ(a, a)

y and a1 , . . . , an , b1 , . . . , bm ∈ A. Then, the

R
proof is similar to that of Lemma 6.8 to show that
[µ]Θ (t(a1 , . . . , an , b1 , . . . , bm )) ≥ [µ]Θ (b1 ) ∧ · · · ∧ [µ]Θ (bm ). 
O

Lemma 6.10. Let a ∈ A. For each Θ, Φ ∈ F Con(A), we have

[[a]Θ ]Φ = [a]Θ◦Φ .
TH

P r o o f. For any x ∈ A, consider the following:

_
[[a]Θ ]Φ (x) = {[a]Θ (y) ∧ Φ(x, y) : y ∈ A}
_
= {Θ(a, y) ∧ Φ(x, y) : y ∈ A} = Θ ◦ Φ(a, x)
U

= [a]Θ◦Φ (x).
Thus, [[a]Θ ]Φ = [a]Θ◦Φ . 
A

Lemma 6.11. Let a ∈ A. For each Θ, Φ ∈ F Con(A), we have

_
[a]Θ [a]Φ ≤ [[a]Θ ]Φ ≤ [a]Θ∨Φ
a
W
where denotes the supremum of fuzzy cosets in F Ca (A).
a

P r o o f. To prove the first inequality let us put λ = [a]Θ ∪ [a]Φ . Then λ is a fuzzy subset of A (not
necessarily a fuzzy coset) such that
_ a
[a]Θ [a]Φ = λ .
a

586
 L-FUZZY COSETS IN UNIVERSAL ALGEBRAS

Let x ∈ A. Then by Theorem 4.7

a _n
λ (x) = λ(b1 ) ∧ · · · ∧ λ(bm ) : t(a1 , . . . an , b1 , . . . , bm ) = x for some a1 , . . . an ∈ A,

and some coset term t(→ −x,→ −

y ) in → −
o
y .
Again from Lemma 6.10, we have
_
[[a]Θ ]Φ (x) = {Θ(a, y) ∧ Φ(x, y) : y ∈ A}.
Let us define two sets Hx and Gx as follows:

Hx = λ(b1 ) ∧ · · · ∧ λ(bm ) : t(a1 , . . . an , b1 , . . . , bm ) = x for some a1 , . . . an ∈ A

PY
and some coset term t(→ −x,→ −y ) in → −
y ,

Gx = Θ(a, y) ∧ Φ(x, y) : y ∈ A .
W W
Then both Hx and Gx are nonempty subsets of L. Our aim is to show that Hx ≤ Gx for all
→
−
x ∈ A. Let α ∈ Hx . Then α = λ(b1 ) ∧ · · · ∧ λ(bm ), for some b1 , . . . , bm ∈ A, such that t(→ −
a, b)=x

O
for some →−a ∈ An where t(→ −x,→ −y ) is a coset term in → −y . That is, b1 , . . . , bm ∈ λα = ([a]Θ )α ∪ ([a]Φ )α .
Without loss of generality, we can assume that b1 , . . . , bk ∈ ([a]Θ )α and bk+1 , . . . , bm ∈ ([a]Φ )α . So
that

Since x = t(→
→
−
−

−
→
−
C
Θ(a, b1 ) ∧ · · · ∧ Θ(a, bk ) ≥ α and Φ(a, bk+1 ) ∧ · · · ∧ Φ(a, bm ) ≥ α.
a , b1 , . . . , bk , bk+1 , . . . , bm ) and t(→ −x,→ −y ) is a coset term in →
a = t( a , a, . . . , a, a, . . . , a). Put y = t( a , b1 , . . . , bk , a, a, . . . , a) and consider the following:
Θ(a, y) = Θ(t(→ a , a, . . . , a, a, . . . , a), t(→
−
1a , b , . . . , b , a, a, . . . , a))
k
−
y , we get
R
≥ Θ(a, b1 ) ∧ · · · ∧ Θ(a, bk )
≥α
O

and
Φ(x, y) = Φ(t(→
−
a , b1 , . . . , bk , bk+1 , . . . , bm ), t(→
−
a , b1 , . . . , bk , a, a, . . . , a))
≥ Φ(bk+1 , a) ∧ · · · ∧ Φ(bm , a)
TH

≥ α.
W W
If we put β = Θ(a, y) ∧ Φ(x, y), then β ∈ Gx such that α ≤ β, which gives that Hx ≤ Gx .
Therefore _
[a]Θ [a]Φ ≤ [[a]Θ ]Φ .
a
U

The last inequality follows from the fact that Θ ◦ Φ ≤ Θ ∨ Φ and this completes the proof. 
A

7. Characterizing congruence permutable varieties

The main purpose of this section is to give several characterizations for a class of algebras to be
congruence permutable by applying fuzzy cosets.
Definition 7.1 ([ChEL]). A class K of algebras is called congruence permutable if the following
holds for each A ∈ K and each θ, φ ∈ Con(A)
θ ◦ φ = φ ◦ θ.
In the following theorem we give an equivalent condition for a variety K of algebras to be
congruence permutable.

587
 GEZAHAGNE MULAT ADDIS

Theorem 7.2. A class K of algebras is congruence permutable if and only if

Θ◦Φ=Φ◦Θ
for each A ∈ K and all Θ, Φ ∈ F Con(A).
P r o o f. Suppose that K is congruence permutable. Let Θ, Φ ∈ F Con(A). For any x, y ∈ A, let
us define two sets Hx,y and Gx,y as follows:
Hx,y = {Θ(x, z) ∧ Φ(y, z) : z ∈ A},
Gx,y = {Θ(y, z) ∧ Φ(x, z) : z ∈ A}.

PY
Then both Hx,y and Gx,y are nonempty subsets of L. Our claim is to show that
_ _
Hx,y = Gx,y for all x, y ∈ A.
Let α ∈ Hx,y . Then α = Θ(x, z) ∧ Φ(y, z) for some z ∈ A. So that Θ(x, z) ≥ α and Φ(y, z) ≥ α;
i.e., (x, z) ∈ Θα , (y, z) ∈ Φα and both Θα and Φα are congruence relations on A, which gives that
(x, y) ∈ Φα ◦ Θα . Since K is congruence permutable, we get (x, y) ∈ Θα ◦ φα . So there exists some

O
u ∈ A such that (x, u) ∈ φα and (y, u) ∈ Θα ; that is,
Θ(y, u) ∧ Φ(x, u) ≥ α.
Thus
W
Gx,y
_
C
Gx,y ≥
Similarly, by interchanging Θ and Φ we can show that
_
W
≥ α. Since α is arbitrary in Hx,y , we get Gx,y ≥ α for all α ∈ Hx,y and hence
Hx,y .
R
_ _
Gx,y ≤ Hx,y .
Therefore the equality holds. The converse part of this theorem follows from the fact that every
O

congruence relation on A can be identified by as a fuzzy congruence on A by its characteristic

mapping. 
Corollary 7.3. A class K of algebras is congruence permutable if and only if
TH

Θ◦Φ=Θ∨Φ
for each A ∈ K and all Θ, Φ ∈ F Con(A).
Theorem 7.4. A class K of algebras is congruence permutable if and only if for each A ∈ K and
each a ∈ A, the map fa : F Con(A) → F Ca (A) defined by:
U

fa (Θ) = [a]Θ
is a lattice homomorphism.
A

P r o o f. Suppose that K is congruence permutable. For any Θ, Φ ∈ F Con(A), it is clear that

[a](Θ∧Φ) = [a]Θ ∧ [a]Φ .
W
So, it is enough to show that [a](Θ∨Φ) = [a]Θ [a]Φ . One inequality is given in Lemma 6.11.
a
To prove the other inequality, let x ∈ A. Since K is congruence permutable, it follows from
Corollary 7.3 that Θ ◦ Φ = Θ ∨ Φ. Then we have
[a](Θ∨Φ) (x) = Θ ∨ Φ(a, x)
= Θ ◦ Φ(a, x)
_
= {Θ(a, y) ∧ Φ(x, y) : y ∈ A}.

588
 L-FUZZY COSETS IN UNIVERSAL ALGEBRAS

For an arbitrary y in A, let us put

α = Θ(a, y) ∧ Φ(x, y).
So, we have [a]Θ (y) ≥ α and Φ(x, y) ≥ α. Since, by our assumption, K is congruence permutable,
it has a Mal’cev term p(u, v, w) such that the following equations are valid in K
p(u, u, w) ≈ w ≈ p(w, u, u).
Let us put b1 = a, b2 = y and b3 = p(a, y, x). Since we can write a as a = p(a, x, x) we have the
following:
Φ(a, b3 ) = Φ(p(a, x, x), p(a, y, x))

PY
≥ Φ(x, y)
≥ α.
W
That is [a]Φ (b3 ) ≥ α. If we put, for simplicity, µ = [a]Θ [a]Φ , then µ is a fuzzy coset of A
a
determined by a such that

O
µ(b1 ) ∧ µ(b2 ) ∧ µ(b3 ) ≥ α.
Let us define a (2 + 3)-ary term t( x , →
→
− −
y ) by
t(x1 , x2 , y1 , y2 , y3 ) = p(p(x1 , y1 , x2 ), p(x1 , y2 , x2 ), y3 ).

C
Then t(x1 , x2 , y1 , y2 , y3 ) is a coset term in y1 , y2 , y3 . To verify this consider
t(x1 , x2 , y, y, y) = p(p(x1 , y, x2 ), p(x1 , y, x2 ), y) = y.
Moreover, consider the following
R
t(a, x, b1 , b2 , b3 ) = t(a, x, a, y, p(a, y, x))
= p(p(a, a, x), p(a, y, x), p(a, y, x))
O

= p(a, a, x)
= x.
Since µ is a fuzzy coset, we get
TH

µ(x) ≥ µ(b1 ) ∧ µ(b2 ) ∧ µ(b3 ) ≥ α.

Since α = Θ(a, y) ∧ Φ(x, y) and y is arbitrary in A, we can conclude that
_
µ(x) ≥ {Θ(a, y) ∧ Φ(x, y) : y ∈ A} = Θ ◦ Φ(a, x) = Θ ∨ Φ(a, x).
W
Therefore [a](Θ∨Φ) ≤ [a]Θ [a]Φ and hence the equality holds. Conversely suppose that the map
U

a
fa : F Con(A) → F Ca (A) defined by
fa (Θ) = [a]Θ
A

is a lattice homomorphism. Then it follows from Lemma 6.10 and 6.11 that
_
[a](Θ∨Φ) = [a](Θ◦Φ) = [a]Θ [a]Φ
a
for all a ∈ A, Θ, Φ ∈ F Con(A). Now for any a, b ∈ A, consider
Θ ∨ Φ(a, b) = [a](Θ∨Φ) (b) = [a](Θ◦Φ) (b) = Θ ◦ Φ(a, b).
Therefore by Corollary 7.3, K is congruence permutable. 
Theorem 7.5. The following are equivalent:
(1) F La (A) = F Ca (A) for all a ∈ A.
(2) F La (A) is a dual ideal of F Ca (A).

589
 GEZAHAGNE MULAT ADDIS

P r o o f. It is known that every lattice is a dual ideal of it self. This makes trivial the proof of
(1) ⇒ (2). To prove (2) ⇒ (1), assume that F La (A) is a dual ideal of F Ca (A). The inclusion
F La (A) ⊆ F Ca (A) is proved in Lemma 6.8. To prove the other inclusion let µ ∈ F Ca (A). Consider
the the zero fuzzy congruence 0A on A defined by
(
1 if x = y
0A (x, y) =
0 otherwise
for all x, y ∈ A. Since µ(a) = 1, we have [a]0A ≤ µ; that is, [a]0A ∈ F La (A) and µ ∈ F Ca (A) such
that [a]0A ≤ µ. Since F La (A) is a dual ideal of F Ca (A), we get µ ∈ F La (A). So F Ca (A) ⊆ F La (A)

PY
and hence the equality holds. 

Theorem 7.6. If K is congruence permutable, then

F La (A) = F Ca (A)

P r o o f. It is enough to show that F Ca (A) ⊆ F La (A). Let µ ∈ F Ca (A).

O
Claim. For each unary algebraic polynomial f (x) on A, we show that
µ(x) ∧ µ(y) ∧ µ(f (x)) = µ(x) ∧ µ(y) ∧ µ(f (y))
for all x, y ∈ A. Put
C
α = µ(x) ∧ µ(y) ∧ µ(f (x)).
Then α ≤ µ(x) ∧ µ(y) and α ≤ µ(f (x)). Let a1 , a2 , . . . , an ∈ A and t(x1 , x2 , . . . , xn , y) be a term
R
on A such that f (z) = t(a1 , a2 , . . . , an , z) for all z ∈ A. Since K is congruence permutable, it has
a Mal’cev term p as given in Theorem 7.4. Now let us define a term r(x1 , x2 , . . . , xn , y1 , y2 , y3 ) by
r(x1 , x2 , . . . , xn , y1 , y2 , y3 ) = p(y1 , t(x1 , x2 , . . . , xn , y2 ), t(x1 , x2 , . . . , xn , y3 )).
O

Then it can be verified that r(→ −x , y1 , y2 , y3 ) is a coset term in y1 , y2 , y3 . Moreover,

r(a1 , a2 , . . . , an , f (x), x, y) = p(f (x), p(t(a1 , a2 , . . . , an , x)), p(t(a1 , a2 , . . . , an , y)))
TH

= p(f (x), f (x), f (y))

= f (y).
Being µ a fuzzy coset and r(→
−
x , y1 , y2 , y3 ) a coset term in y1 , y2 , y3 , we have the following:
µ(f (y)) = µ(r(a1 , a2 , . . . , an , f (x), x, y))
U

≥ µ(f (x)) ∧ µ(x) ∧ µ(y)

= α.
A

So that µ(f (y)) ≥ α. Since µ(x) ∧ µ(y) ≥ α, we get µ(f (y)) ∧ µ(x) ∧ µ(y) ≥ α; that is,
µ(f (y)) ∧ µ(x) ∧ µ(y) ≥ µ(f (x)) ∧ µ(x) ∧ µ(y).
The other inequality can be proved by interchanging x and y, and hence the equality holds. By
Lemma 6.7, µ ∈ F La (A) and this completes the proof. 

Theorem 7.7. If for each Θ, Φ ∈ F Con(A) there exists Ψ ∈ F Con(A) such that
_
[a]Θ [a]Φ = [a]Ψ
a

for all a ∈ A, then K is congruence permutable.

590
 L-FUZZY COSETS IN UNIVERSAL ALGEBRAS

P r o o f. Let Θ, Φ ∈ F Con(A). Then there exists Ψ ∈ F Con(A) such that

_
[a]Θ [a]Φ = [a]Ψ
a
for all a ∈ A. Our aim is to show that Ψ = Θ ◦ Φ. For any a, b ∈ A consider
_
Θ ◦ Φ(x, y) = {Θ(x, a) ∧ Φ(y, a) : a ∈ A}
_
= {[a]Θ (x) ∧ [a]Φ (y) : a ∈ A}
_
≤ {[a]Ψ (x) ∧ [a]Ψ (y) : a ∈ A}

PY
_
= {Ψ(x, a) ∧ Ψ(y, a) : a ∈ A}
≤ Ψ(x, y).
Moreover,
Ψ(x, y) = [y]Ψ (x)

O
_
= [y]Θ [y]Φ (x)
y
≤ [y](Θ◦Φ) (x) (By Lemma 6.10 and 6.11)
= Θ ◦ Φ(x, y).
W W C
Therefore Ψ = Θ ◦ Φ. Since [a]Θ [a]Φ = [a]Φ [a]Θ , we get Θ ◦ Φ = Φ ◦ Θ. Thus, by Theorem 7.2,
K is congruence permutable.
a a

R
A Mal’cev condition on admissible relations was applied by Werner in [Wer]. In the following
theorem we adopt his theorem and give equivalent conditions for the Mal’cev characterization of
O

congruence prermutable varieties.

Theorem 7.8. For any variety K, the following are equivalent:
(1) K is congruence permutable.
TH

(2) For any A ∈ K each admissible reflexive fuzzy relation on A is symmetric.

(3) For any A ∈ K each admissible reflexive fuzzy relation on A is transitive.
(4) For any A ∈ K each admissible reflexive fuzzy relation on A is a fuzzy congruence on A.
P r o o f. (1) ⇒ (2). Suppose that K is permutable. Then, it has the Mal’cev term p(x, y, z). Let
U

Θ be an admissible reflexive fuzzy relation on A, and x, y ∈ A. Then we have

x = p(x, y, y) and y = p(x, x, y).
A

Now consider
Θ(x, y) = Θ(p(x, y, y), p(x, x, y))
≥ Θ(x, x) ∧ Θ(y, x) ∧ Θ(y, y)
= Θ(y, x).
Similarly we can verify that Θ(y, x) ≥ Θ(x, y). Thus Θ(x, y) = Θ(y, x) and hence Θ is symmetric.
(2) ⇒ (1). Assume (2). Let θ be an admissible reflexive relation on A. Then it is routine to
verify that the fuzzy relation θ on A defined by:
(
1 if (x, y) ∈ θ
θ(x, y) =
0 otherwise

591
 GEZAHAGNE MULAT ADDIS

for all x, y ∈ A, is an admissible reflexive fuzzy relation on A. So by (2), θ is symmetric and hence
θ is symmetric. i.e., each admissible reflexive relation on A is symmetric. By the equivalency in
[Wer] (1) holds.
(1) ⇒ (3). Suppose that K is permutable. Then, it has the Mal’cev term p(x, y, z). Let Θ be
an admissible reflexive fuzzy relation on A, and x, y, z ∈ A. Then we have
x = p(x, y, y) and y = p(z, z, y).
Then consider the following:
Θ(x, y) = Θ(p(x, y, y), p(z, z, y))

PY
≥ Θ(x, z) ∧ Θ(y, z) ∧ Θ(y, y)
= Θ(x, z) ∧ Θ(y, z)
= Θ(x, z) ∧ Θ(z, y).
Since z is arbitrary in A, we get

O
_
Θ(x, y) ≥ {Θ(x, z) ∧ Θ(z, y) : z ∈ A}.
Therefore Θ is transitive. Hence proved.
(3) ⇒ (1). Assume (3) and let θ be an admissible reflexive relation on A. Then it is routine to
C
verify that the fuzzy relation θ on A as defined above is an admissible reflexive fuzzy relation on
A. So by our assumption, θ is transitive which implies that θ is transitive. i.e., each admissible
reflexive relation on A is transitive. By the equivalency in [Wer] (1) holds.
One can easily observe that (2) and (3) are equivalent with (4). Thus, the equivalency (1) ⇔
R
(2) ⇔ (3) ⇔ (4) holds. 

Acknowledgement. The author would like to thank the anonymous referee whose comments
O

improved the original version of this manuscript.

TH

REFERENCES

[Add1] ADDIS, G. M.: Fuzzy homomorphism theorems on groups, Korean J. Math. 26 (2018), 373–385.
[Add2] ADDIS, G. M.: L-Fuzzy congruence classes in universal algebras, to appear.
[Add3] ADDIS, G. M.—ENGIDAW, D. A.—GETACHEW, T. G.: Fuzzy action of fuzzy groups on a set, Quasi-
groups Related Systems 26 (2018), 155–164.
U

[Agl1] AGLIANO, P.—URSINI, A.: Cosets in universal algebra, J. Algebra 107 (1987), 376–384.
[Agl2] AGLIANO, P.—URSINI, A.: Ideals and other generalizations of congruence classes, J. Austral. Math.
Soc. 53 (1992), 103–115.
[Agl3] AGLIANO, P.—URSINI, A.: On Some Ideal Basis Theorems. Rapporto Matematico n. 182, Universita
A

di Siena, 1988.
[AlAd1] ALABA, B. A.—ADDIS, G. M.: L-Fuzzy ideals in universal algebras, Ann. Fuzzy Math. Inform. 17(1)
(2019), 31-39.
[AlAd2] ALABA, B. A.—ADDIS, G. M.: L-Fuzzy prime ideals in universal algebras, Adv. Fuzzy Syst. 2019, Art.
ID 5925036.
[AlAd3] ALABA, B. A.—ADDIS, G. M.: L-Fuzzy semi-prime ideals in universal algebras, Korean J. Math. 28(1)
(2020), 49–64.
[AlAl1] ALABA, B. A.—ALEMAYEHU, T. G.: Closure Fuzzy Ideals of MS-algebras, Ann. Fuzzy Math. Inform.
16 (2018), 247–260.
[AlAl2] ALABA, B. A.—ALEMAYEHU, T. G.: Fuzzy Ideals in Demipseudocomplemented MS-algebras, Ann.
Fuzzy Math. Inform. 18 (2019), 123–143.
[AlAl3] ALABA, B. A.—ALEMAYEHU, T. G.: β-Fuzzy filters of MS-algebras, Korean J. Math. 27 (2019),
595–612.

592
 L-FUZZY COSETS IN UNIVERSAL ALGEBRAS

[AlAl4] ALABA, B. A.—ALEMAYEHU, T. G.: e-Fuzzy filters of MS-algebras, Korean J. Math. 27 (2019),
1159–1180.
[AlN1] ALABA, B. A.—NORAHUN, W. Z.: Fuzzy annihilator ideals in distributive lattices, Ann. Fuzzy Math.
Inform. 16 (2018), 191–200.
[AlN2] ALABA, B. A.—NORAHUN, W. Z.: α-fuzzy ideals and space of prime a-fuzzy ideals in distributive
lattices, Ann. Fuzzy Math. Inform. 17 (2019), 147–163.
[AlN3] ALABA, B. A.—NORAHUN, W. Z.: Fuzzy ideals and fuzzy filters of pseudo-complemented semilattices,
Adv. Fuzzy Syst. 2019, Art. ID 4263923.
[AlTE1] ALABA, B. A.—TAYE, M. A.—ENGIDAW, D. A.: L-fuzzy ideals of a poset, Ann. Fuzzy Math. Inform.
16(3) (2018), 285–299.
[AlTE2] ALABA, B. A.—TAYE, M. A.—ENGIDAW, D. A.: L-fuzzy prime ideals and maximal L-fuzzy ideals of

PY
a poset, Ann. Fuzzy Math. Inform. 18(1) (2019), 1–13.
[AlE] ALABA, B. A.—ENGIDAW, D. A.: L-fuzzy semiprime ideals of a poset, Adv. Fuzzy Syst. 2020, Art. ID
1834970.
[AnS1] ANTHONY, J. M.—SHERWOOD, H.: Fuzzy groups redefined, J. Math. Anal. Appl. 69 (1979), 124–130.
[AnS2] ANTHONY, J. M.—SHERWOOD, H.: A characterization of fuzzy subgroups, Fuzzy Sets and Systems 7
(1982), 297–305.
[Bel] BĚLOHLÁVEK, R.: Convex sets in algebras, Acta Univ. Palack. Olomuc. Fac. Rerum Natur. Math. 41

O
(2002), 21–33.
[Bis] BISWAS, R.: Fuzzy fields and fuzzy linear spaces redefined, Fuzzy Sets and Systems 33 (1989), 257–259.
[BoW] BO, Y.—WU, W.: Fuzzy ideals on a distributive lattice, Fuzzy Sets and Systems 35 (1990), 231–240.
[BoiJ] BOIXADER, D.—JACAS, J.: Fuzzy actions, Fuzzy Sets and Systems 339 (2018), 17–30.
[BurS]
1981.
C
BURRIS, S.—SANKAPPANAVOR, H. P.: A Course in Universal Algebra, Springer-Verlag, GTM78,

[ChEL] CHAJDA, I.—EIGENTHALER, G.—LÄNGER, H.: Congruence Classes in Universal Algebras. Res. Exp.
Math. 26, 2004.
[ChaK] CHAKRABORTY, A. B.—KHARE, S. S.: Fuzzy homomorphism and algebraic structures, Fuzzy Sets
R
and Systems 59 (1993), 211–221
[ChoCK] CHOUDHURY, F. P.—CHAKRABORTY, A. B.—KHARE, S. S.: A note on fuzzy subgroups and fuzzy
homomorphism, J. Math. Anal. Appl. 131 (1988), 537–553.
[Chi] CHITIPOLU, P. R.: Fuzzy Subalgebras and Congruences, Doctoral Thesis, Andhra University, 2013.
O

[Das1] DAS, P. S.: Fuzzy groups and level subgroups, J. Math. Anal. Appl. 84 (1981), 264–269.
[Das2] DAS, P.: Fuzzy vector spaces under triangular norms, Fuzzy Sets and Systems 25 (1988), 73–85.
[DKA1] DIXIT, V. N.—KUMAR, R.—AJMAL, N.: Level subgroups and union of fuzzy subgroups, Fuzzy Sets and
Systems 37 (1990), 359–371.
TH

[DKA2] DIXIT, V. N.—KUMAR, R.—AJMAL, N.: Fuzzy ideals and fuzzy prime ideals of a ring, Fuzzy Sets and
Systems 44 (1991), 127–138.
[DuJ1] DUDEK, W. A.—JUN, Y. B.: Fuzzy BCC-ideals in BCC-algebras, Math. Montisnigri 10 (1999), 21–30.
[DuJ2] DUDEK, W. A.—JUN, Y. B.—STOJAKOVIČ, Z.: On fuzzy ideals in BCC-algebras, Fuzzy Sets and
Systems 123 (2001), 251–258.
[FoT] FOKA, S. V. T.—TONGA, M.: A note on the algebraicity of L-fuzzy subalgebras in universal algebra,
Soft Comput. 24 (2020), 895–899.
U

[Gog] GOGUEN, J. A.: L-fuzzy sets, J. Math. Anal. Appl. 18 (1967), 145–174.
[Gra] GRÄTZER, G.: Universal Algebras, 2nd ed., Springer, New York 1979.
[Gum] GUMM, H. P.—URSINI, A.: Ideals in universal algebra, Algebra Universalis 19 (1984), 45–54.
A

[Gup] GUPTA, M. M.—RAGADE, R. K.: Fuzzy set theory and its applications: A survey, IFAC Proceedings
Volumes 10 (1977), 247–259.
[JaMU] JANELIDZE, G.—MÁRKI, L.—URSINI, A.: Ideals and clots in universal algebra and in semi-abelian
categories, J. Algebra 307 (2007), 191–208.
[JuX] JUN, Y. B.—XIN, X. L.: Fuzzy prime ideals and invertible fuzzy ideals in BCK-algebras, Fuzzy Sets and
Systems 117 (2001), 471–476.
[JuLP] JUN, Y. B.—LEE, K. J.—PARK, C. H.: Fuzzy soft set theory applied to BCK/BCI-algebras, Comput.
Math. Appl. 59 (2010), 3180–3192.
[KatL] KATSARAS, A. K.—LIU, D. B.: Fuzzy vector spaces and fuzzy topological vector spaces, J. Math. Anal.
Appl. 58 (1977), 135-146.
[KoNL] KOGUEP, B. B. N.—NKUIMI, C.—LELE, C.: On fuzzy prime ideals of lattices, SJPAM 3 (2008), 1–11.
[KuSY] KUMAR, I. J.—SAXTENA, P. K.—YADAV, P.: Fuzzy normal subgroups and fuzzy quotients, Fuzzy Sets
and Systems 46 (1992), 121–132.

593
 GEZAHAGNE MULAT ADDIS

[Kum1] KUMAR, R.: Fuzzy nil radical and fuzzy primary ideals, Fuzzy Sets and Systems 43 (1991), 81–93.
[Kum2] KUMAR, R.: Fuzzy semi-primary ideals of rings, Fuzzy Sets and Systems 42 (1991), 263–273.
[Kum3] KUMAR, R.: Certain fuzzy ideals of rings redefined, Fuzzy Sets and Systems 46 (1992), 257–260.
[Kum4] KUMAR, R.: Fuzzy prime spectrum of a ring, Fuzzy Sets and Systems 46 (1992), 147–154.
[Kum5] KUMAR, R.: Fuzzy subgroups, fuzzy ideals, and fuzzy cosets: Some properties, Fuzzy Sets and Systems
48 (1992), 267–274.
[Kura1] KURAOKA, T.—SUZUKI, N. Y.: Lattice of fuzzy subalgebras in universal algebra, Algebra Universalis
47 (2002), 223–237.
[Kura2] KURAOKA, T.: Formulas on the lattice of fuzzy subalgebras in universal algebra, Fuzzy Sets and Systems
158 (2007), 1767–1781.
[Kuro1] KUROKI, N.: Fuzzy bi-ideals in semigroups, Comment. Math. Univ. St. Paul. 28 (1980), 17–21.

PY
[Kuro2] KUROKI, N.: Fuzzy semiprime ideals in semigroups, Fuzzy Sets and Systems 8 (1982), 71–79.
[Kuro3] KUROKI, N.: On fuzzy ideals and fuzzy bi-ideals in semigroups, Fuzzy Sets and Systems 5 (1981),
203–215.
[Liu] LIU, W. J.: Fuzzy invariant subgroups and fuzzy ideals, Fuzzy Sets and Systems 8 (1982), 133–139.
[Mak] MAKAMBA, B. B.—MURALI, B. B.: On prime fuzzy submodules and radicals, J. Fuzzy Math. 8(4),
(2000), 831–843.
[Mal] MAL’CEV, A. I.: On the general theory of algebraic systems, Matem. Sbornik 35 (1954), 3–20 (in Russian).

O
[MaM1] MALIK, D. S.—MORDESON, J. N.: Fuzzy homomorphism of rings, Fuzzy Sets and Systems 46 (1992),
139–146
[MaM2] MALIK, D. S.—MORDESON, J. N.: Fuzzy prime ideals of a ring, Fuzzy Sets and Systems 37 (1990),
93–98.

C
[MaM3] MALIK, D. S.—MORDESON, J. N.: Fuzzy maximal, radical and primary ideals of a ring, Inform. Sci.
53 (1991), 237–250.
[MaM4] MALIK, D. S.—MORDESON, J. N.: Radicals of fuzzy ideals, Inform. Sci. 65 (1992), 239–252.
[Meng] MENG, J.—GUO, X.: On fuzzy ideals in BCK/BCI-algebras, Fuzzy Sets and Systems 149 (2005),
509–525.
R
[MoBR] MORDESON, N.—BHUNTANI, K. R.—ROSENFELD, A.: Fuzzy Group Theory, Springer, 2005.
[MoM] MORDESON, N.—MALIK, D. S.: Fuzzy Commutative Algebra, World Scientific, 1998.
[MuS] MUKHERJEE, T. K.—SEN, M. K.: On fuzzy ideals of a ring (I), Fuzzy Sets and Systems 21 (1987),
99-104.
O

[Mur1] MURALI, V.: A Study of Universal Algebras in Fuzzy Set Theory, Doctoral thesis, Rhodes University,
1987.
[Mur2] MURALI, V.: Fuzzy congruence relation, Fuzzy Sets and Systems 41 (1991), 359–369.
[NeR] NEGOITA, C. V.—RALESCU, D. A.: Application of Fuzzy Systems Analysis, Birkhauser, Basel, 1975.
TH

[Ros] ROSENFELD, A.: Fuzzy groups, J. Math. Anal. Appl. 35 (1971), 512–517.
[SwS] SWAMY, U. M.—SWAMY, K. L. N.: Fuzzy prime ideals of rings, J. Math. Anal. Appl. 134 (1988),
94–103.
[SwR1] SWAMY, U. M.—RAJU, D. V.: Algebraic fuzzy systems, Fuzzy Sets and Systems 41 (1991), 187–194.
[SwR2] SWAMY, U. M.—RAJU, D. V.: Irreducibility in algebraic fuzzy systems, Fuzzy Sets and Systems 41
(1991), 233–241.
[SwR3] SWAMY, U. M.—RAJU, D. V.: Fuzzy ideals and congruences of lattices, Fuzzy Sets and Systems 95
U

(1998), 249-253.
[Urs] URSINI, A.: Prime ideals in universal algebras, Acta Univ. Carolin. Math. Phys. 25 (1984), 75–87.
[Wang] WANG, L. X.: A First Course in Fuzzy Systems and Control, Prentice-Hall International Inc., 1997.
[Wer] WERNER, H.: A Mal’cev condition on admissible relations, Algebra Universalis 3 (1973), 263.
A

[Zad] ZADEH, L. A.: Fuzzy Sets, Information and Control 8 (1965), 338–353.
[Zah] ZAHEDI, M. M.: Some results on L-fuzzy modules, Fuzzy Sets and Systems 55 (1993), 355–361.
[Zim] ZIMMERMANN, H. J.: Fuzzy Set Theory and Its Applications, 4th ed., Springer Science+Business Media,
New York, 2001.

Received 9. 4. 2020 Department of Mathematics

Accepted 3. 7. 2020 University of Gondar
P.O. BOX: 196, GONDAR
ETHIOPIA
E-mail: buttu412@yahoo.com, gezahagne412@gmail.com

594