=Paper=
{{Paper
|id=Vol-2925/short1
|storemode=property
|title=Decomposition schemes for symmetric n-ary bands
|pdfUrl=https://ceur-ws.org/Vol-2925/short1.pdf
|volume=Vol-2925
|authors=Jimmy Devillet,Pierre Mathonet
|dblpUrl=https://dblp.org/rec/conf/algos/DevilletM20
}}
==Decomposition schemes for symmetric n-ary bands==
https://ceur-ws.org/Vol-2925/short1.pdf
D ECOMPOSITION SCHEMES FOR SYMMETRIC n- ARY BANDS
Jimmy Devillet Pierre Mathonet
Department of Mathematics Department of Mathematics
University of Luxembourg University of Liege
Luxembourg Liege, Belgium
jimmy.devillet@uni.lu p.mathonet@uliege.be
A BSTRACT
We extend the classical (strong) semilattice decomposition scheme of certain classes of semigroups
to the class of idempotent symmetric n-ary semigroups (i.e. symmetric n-ary bands) where n ≥ 2 is
an integer. More precisely, we show that these semigroups are exactly the strong n-ary semilattices
of n-ary extensions of Abelian groups whose exponents divide n − 1. We then use this main result to
obtain necessary and sufficient conditions for a symmetric n-ary band to be reducible to a semigroup.
Keywords Semigroup ⋅ idempotency ⋅ semilattice decomposition ⋅ reducibility
1 Introduction
Semigroups are ubiquitous and have numerous applications both in theoretical and applied mathematics. An extensive
study of these structures began in the second half of the 20th century (see the pioneering works [2] and [19], or the
textbooks [3, 10, 12, 20, 21] and references therein). In the algebraic analysis of semigroups, it soon became clear that it
was useful to obtain a decomposition scheme of the semigroup under consideration into subsemigroups that are easier to
describe or have additional properties (e.g. being groups), but also to be able to build a semigroup by combining given
subsemigroups in a suitable way, that is, to use a composition scheme for semigroups.
Several classes of semigroups have the remarkable property to admit such composition/decomposition schemes; see,
e.g., Krohn-Rhodes theorem for finite semigroups and finite automata [14]. A noteworthy example of such a scheme
is given by strong semilattice decompositions of certain classes of bands 1 . In this paper we generalize these strong
semilattice decompositions to structures with higher arities, defined as follows.
An n-ary operation F ∶ X n → X (where n ≥ 2 is an integer and X is a non-empty set) is associative if
F (x1 , . . . , xi−1 , F (xi , . . . , xi+n−1 ), xi+n , . . . , x2n−1 ) = F (x1 , . . . , xi , F (xi+1 , . . . , xi+n ), xi+n+1 , . . . , x2n−1 ), (1)
for all x1 , . . . , x2n−1 ∈ X and all 1 ≤ i ≤ n − 1. If F is an n-ary associative operation on X, then (X, F ) is an n-ary
semigroup. These n-ary structures, first studied in [9] and [22], have applications in different fields such as automata
theory (see, e.g., [11]), coding theory, and cryptology (see, e.g., [16, 17]).
The classical definitions of symmetry and idempotency can also be extended to n-ary operations as follows: F is
idempotent if F (x, . . . , x) = x for every x ∈ X and F is symmetric (or commutative) if F is invariant under the action
of permutations.
Many examples of n-ary semigroups are obtained by extending binary semigroups: if G∶ X 2 → X is an associative
operation, then we can define a sequence of operations inductively by setting G1 = G, and
Gm (x1 , . . . , xm+1 ) = Gm−1 (x1 , . . . , xm−1 , G(xm , xm+1 )), m ≥ 2.
Setting F = G , it is straightforward to see that the pair (X, F ) is indeed an n-ary semigroup. It is said to be the
n−1
n-ary extension of (X, G) and we say that (X, F ) is reducible to (X, G)2 . However, not every n-ary semigroup
1
A band is a semigroup (X, G) where X is a nonempty set and the binary associative operation G∶ X 2 → X satisfies G(x, x) = x
for every x ∈ X; see, e.g., [12] for more details.
2
We also say that F is the n-ary extension of G or that F is reducible to G or even that G is a binary reduction of F .
Copyright © 2020 for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0).
is the n-ary extension of a binary semigroup. For instance, the ternary associative operation F defined on R3 by
F (x1 , x2 , x3 ) = x1 − x2 + x3 is not reducible to any binary associative operation. The problem of reducibility was
considered recently in [13, 18] for n-ary semigroups endowed with additional structures and in [1, 4, 6] for the class
of quasitrivial n-ary semigroups3 . These are n-ary semigroups (X, F ) that preserve all unary relations, i.e., such
that F (x1 , . . . , xn ) ∈ {x1 , . . . , xn } for all x1 , . . . , xn ∈ X. It was shown [4] that all quasitrivial n-ary semigroups are
reducible. Then in [5], the authors relaxed the quasitriviality condition by considering operations whose restrictions on
certain subsets of the domain are quasitrivial. It turns out that these operations are also reducible.
In this work we study the class of symmetric (or commutative) n-ary bands, that is, symmetric idempotent n-ary
semigroups. Typical examples of symmetric n-ary bands are given by n-ary extensions of semilattices and n-ary
extensions of Abelian groups whose exponents divide n − 1. Both classes of examples will play a central role in our
constructions. However, as shown in the following examples, not every symmetric n-ary band is obtained in this way.
Example 1.1. (a) We consider the set X = {1, 2, 3, 4} and we define the symmetric ternary operation
F1 ∶ X 3 → X by its level sets given (up to permutations) by F1−1 ({1}) = {(1, 1, 1)}, F1−1 ({2}) = {(2, 2, 2)},
F1−1 ({3}) = {(1, 1, 2), (1, 1, 3), (1, 2, 4), (1, 3, 4), (2, 2, 3), (2, 3, 3), (2, 4, 4), (3, 3, 3), (3, 4, 4)}. Then
F1−1 ({4}) is made up of all the remaining elements of X 3 . This operation defines a symmetric ternary
band and is not reducible to any binary operation.
(b) We consider the set X = {1, 2, 3} and we define the symmetric ternary operation F2 ∶ X 3 →
X again by its level sets given (up to permutations) by F2−1 ({1}) = {(1, 1, 1)}, F2−1 ({2}) =
{(1, 1, 2), (1, 2, 2), (1, 3, 3), (2, 2, 2), (2, 3, 3)}, and F2−1 ({3}) = {(1, 1, 3), (1, 2, 3), (2, 2, 3), (3, 3, 3)}.
This operation defines a symmetric ternary band. It turns out that it is reducible to a binary operation
on X.
In the next section we define the n-ary counterpart of the classical strong semilattice (de)composition for semigroups
(namely the strong n-ary semilattice decomposition). We show that it enables us to compose n-ary semigroups: every
strong n-ary semilattice of n-ary semigroups is an n-ary semigroup (see Proposition 2.2). Then in Section 3, we provide
a constructive description of the class of symmetric n-ary bands, that is, we show that the symmetric n-ary bands
are exactly the strong n-ary semilattices of n-ary extensions of Abelian groups whose exponents divide n − 1 (see
Theorem 3.12). In the final section, we give a reducibility criterion for symmetric n-ary bands based on their strong
n-ary semilattice decomposition (see Proposition 4.3). Also, Example 1.1 shows how these constructions enable us to
build and analyze examples of symmetric n-ary bands. Almost all the definitions and results in this work stem from [7],
where the reader may find their proofs and alternative developments as well.
2 Strong n-ary semilattices of n-ary semigroups
Throughout this work, we consider a nonempty set X and an integer n ≥ 2. Recall that (X, F ) is said to be an n-ary
groupoid whenever F ∶ X n → X is an n-ary operation. Moreover, if F is associative (i.e., satisfies (1)), then (X, F )
is said to be an n-ary semigroup. The concepts of homomorphims and isomorphisms of n-ary groupoids and n-ary
semigroups are defined as usual.
Recall that e ∈ X is said to be a neutral element for F ∶ X n → X if
F ((k − 1) ⋅ e, x, (n − k) ⋅ e) = x, x ∈ X, k ∈ {1, . . . , n},
where, for any k ∈ {0, . . . , n} and any x ∈ X, the notation k ⋅ x stands for the k-tuple x, . . . , x (for instance F (3 ⋅ x, 0 ⋅
y, 2 ⋅ z) = F (x, x, x, z, z)).
In [8, Lemma 1], it was proved that any associative operation F ∶ X n → X having a neutral element e is reducible to an
associative binary operation Ge ∶ X 2 → X defined by
Ge (x, y) = F (x, (n − 2) ⋅ e, y), x, y ∈ X. (2)
Finally, recall that an equivalence relation ∼ on X is said to be a congruence for F ∶ X n → X (or on (X, F )) if it is
compatible with F , that is, if F (x1 , . . . , xn ) ∼ F (y1 , . . . , yn ) for any x1 , . . . , xn , y1 , . . . , yn ∈ X such that xi ∼ yi for
all i ∈ {1, . . . , n}. We denote by [x]∼ (or [x] when there is no risk of confusion) the equivalence class of x for ∼ and by
F̃ the map induced by F on X�∼ defined by
F̃ ([x1 ]∼ , . . . , [xn ]∼ ) = [F (x1 , . . . , xn )]∼ , ∀x1 , . . . , xn ∈ X.
3
For n = 2, the quasitrivial semigroups were described by Länger [15].
We say that a congruence ∼ on an n-ary groupoid (X, F ) is an n-ary semilattice congruence if (X�∼, F̃ ) is an n-ary
semilattice4 .
Now, let us extend the well-known concept of semilattice of semigroups to n-ary semigroups. Let (Y, ⋏) be a semilattice
and let {(X↵ , F↵ )∶ ↵ ∈ Y } be a set of n-ary semigroups such that X↵ ∩ X = � for any ↵ ≠ . We say that an n-ary
groupoid (X, F ) is an n-ary semilattice (Y, ⋏n−1 ) of n-ary semigroups (X↵ , F↵ ) if X = � X↵ , F �X↵n = F↵ for every
↵ ∈ Y , and
↵∈Y
F (X↵1 × � × X↵n ) ⊆ X↵1 ⋏�⋏↵n , ↵1 , . . . , ↵n ∈ Y. (3)
In this case we write (X, F ) = ((Y, ⋏n−1 ); (X↵ , F↵ )) and we simply say that (X, F ) is an n-ary semilattice of n-ary
semigroups.
Actually, any decomposition of an n-ary semigroup (X, F ) as an n-ary semilattice of n-ary semigroups is associated
with an n-ary semilattice congruence on (X, F ); see, e.g., [12] for the binary counterpart of this result.
The fact that an n-ary groupoid is an n-ary semilattice of n-ary semigroups is not sufficient to ensure that it is an n-ary
semigroup. We need to introduce a generalization of the strong semilattice decomposition. This is done in the following
definition.
Definition 2.1. Let (X, F ) = ((Y, ⋏n−1 ); (X↵ , F↵ )) be an n-ary semilattice of n-ary semigroups. Suppose that for
any ↵, ∈ Y such that ↵ � there is a homomorphism '↵, ∶ X↵ → X such that the following conditions hold.
(a) The map '↵,↵ is the identity on X↵ .
(b) For any ↵, , ∈ Y such that ↵ � � we have ' , ○ '↵, = '↵, .
(c) For any (x1 , . . . , xn ) ∈ X↵1 × � × X↵n we have
F (x1 , . . . , xn ) = F↵1 ⋏�⋏↵n ('↵1 ,↵1 ⋏�⋏↵n (x1 ), . . . , '↵n ,↵1 ⋏�⋏↵n (xn )).
Then (X, F ) is said to be a strong n-ary semilattice (Y, ⋏n−1 ) of n-ary semigroups (X↵ , F↵ ). In this case we write
(X, F ) = ((Y, ⋏n−1 ); (X↵ , F↵ ); '↵, ) and we also say that (X, F ) is a strong n-ary semilattice of n-ary semigroups.
This definition enables us to obtain the main result concerning the composition of n-ary semigroups, which is important
on its own, but also in the next sections.
Proposition 2.2. If (X, F ) is a strong n-ary semilattice of n-ary semigroups, then it is an n-ary semigroup.
3 The structure theorem
Throughout this section, we consider a symmetric n-ary band (X, F ). We associate with it a family of unary operations
and study their most important properties.
Definition 3.1. For every x ∈ X, we define the operation `F
x ∶ X → X by
x (y) = F ((n − 1) ⋅ x, y),
`F y ∈ X.
When there is no risk of confusion, we also denote this operation by `x . We now study elementary properties of this
operation.
Example 3.2. For the structures presented in Example 1.1, these maps are given in the following tables.
y 1 2 3 4
`1 (y)
y 1 2 3
`1 (y)
1 3 3 4
`2 (y)
1 2 3
`2 (y)
4 2 3 4
`3 (y)
2 2 3
`3 (y)
4 3 3 4
`4 (y)
2 2 3
4 3 3 4
Proposition 3.3. The pair ({`x ∶ x ∈ X}, ○) is a semilattice.
We also observe that the pair (X, B) where B is defined by B(x, y) = `x (y) for all x, y ∈ X is a band. For instance,
the tables in Example 3.2 are the operation tables of the corresponding binary operations B. We will not elaborate on
this in the present work but refer the reader to [7] for more details.
4
We say that the n-ary extension of a semilattice is an n-ary semilattice.
The semilattice defined in Proposition 3.3 can be extended to define a symmetric n-ary band. The following result
establishes a tight relation between (X, F ) and this n-ary band.
Proposition 3.4. For every x1 , . . . , xn ∈ X we have
`F (x1 ,...,xn ) = `x1 ○ � ○ `xn ,
that is, the map `∶ (X, F ) → ({`x ∶ x ∈ X}, ○n−1 ) defined by `(x) = `x is a homomorphism.
The map ` defined in the previous proposition enables us to characterize the reducibility of a symmetric n-ary band to a
symmetric (binary) band, i.e., a semilattice.
Proposition 3.5. Let (X, F ) be a symmetric n-ary band. The following assertions are equivalent.
(i) The map ` is injective.
(ii) The n-ary band (X, F ) is isomorphic to ({`x ∶ x ∈ X}, ○n−1 ).
(iii) The n-ary band (X, F ) is an n-ary semilattice.
Also, the map ` enables us to characterize those symmetric n-ary bands that are reducible to Abelian groups. Recall
that a group (X, ∗) with neutral element e has bounded exponent if there exists an integer m ≥ 1 such that the m-fold
product x ∗ � ∗ x is equal to e for any x ∈ X. In that case, the exponent of the group is the smallest integer m ≥ 1
having this property. Using the characterization of Abelian groups having bounded exponent given by Prüfer and Baer
(see [23]), it is straighforward to see that the exponent of an Abelian group divides n − 1 if and only if the group is a
direct sum of cyclic groups whose orders divide n − 1.
Proposition 3.6. Let (X, F ) be a symmetric n-ary band. The following conditions are equivalent.
(i) The map ` is constant (i.e., `x is the identity map on X for any x ∈ X).
(ii) The n-ary band (X, F ) is the n-ary extension of a group (X, ∗) (and in particular (X, ∗) is Abelian and its
exponent divides n − 1).
Now, when the map ` associated with F is not injective, it is natural to consider a quotient, and identify the elements of
X that have the same image by `. In this context, we have the following result.
Proposition 3.7. The binary relation ∼ on X defined by
x∼y ⇔ `x = `y , x, y ∈ X,
is an n-ary semilattice congruence on (X, F ).
Example 3.8. For the structure presented in Example 1.1(a), we have `3 = `4 so [1]∼ = {1}, [2]∼ = {2}, and
[3]∼ = {3, 4}. The binary reduction of (X�∼, F˜1 ) is a semilattice whose Hasse diagram is given in Figure 1 (left). For
instance, we have
[1]∼ ⋏ [2]∼ = F˜1 ([1]∼ , [1]∼ , [2]∼ ) = [F1 (1, 1, 2)]∼ = [3]∼ .
Now, for the structure presented in Example 1.1(b), we only have `2 = `3 so [1]∼ = {1} and [2]∼ = {2, 3}. We see that
the binary reduction of (X�∼, F˜2 ) is a semilattice whose Hasse diagram is given in Figure 1 (right).
[1] [2] [1]
[3] [2]
Figure 1: Hasse diagrams of the binary reductions of (X�∼, F˜1 ) (left) and (X�∼, F˜2 ) (right)
Since ∼ is a congruence for F , this operation restricts to each equivalence class. It is then natural to study the most
important properties of this restriction. Using Proposition 3.6, we directly obtain the following result.
Proposition 3.9. For any x ∈ X, ([x]∼ , F �[x]n∼ ) is the n-ary extension of an Abelian group whose exponent divides
n − 1.
Example 3.10. For both structures presented in Example 1.1, the restrictions of F1 and F2 to [3]3∼ and [2]3∼ , respectively,
are isomorphic to the ternary extension of (Z2 , +).
The congruence ∼ enabled us to decompose X as a n-ary semilattice of n-ary semigroups. In order to obtain a strong
n-ary semilattice decomposition, we still need to define a suitable family of homomorphisms.
Proposition 3.11. For every x, y ∈ X such that [x]∼ � [y]∼ , the map '[x]∼ ,[y]∼ = `y �[x]∼ is a homomorphism from
([x]∼ , F �[x]n∼ ) to ([y]∼ , F �[y]n∼ ).
We can now state our main structure theorem for symmetric n-ary bands.
Theorem 3.12. An n-ary groupoid (X, F ) is a symmetric n-ary band if and only if it is a strong n-ary semilattice of
n-ary extensions of Abelian groups whose exponents divide n − 1.
As a direct application of this theorem, we obtain that a symmetric n-ary band is an n-ary group5 if and only if it is the
n-ary extension of an Abelian group whose exponent divides n − 1.
In view of the main result, in order to build symmetric n-ary bands, we have to consider Abelian groups whose
exponents divide n − 1, and build homomorphisms between the n-ary extensions of such groups. These homomorphisms
are described in the next result.
Proposition 3.13. Let (X1 , ∗1 ) and (X2 , ∗2 ) be two Abelian groups whose exponents divide n − 1 and denote by F1
and F2 the n-ary extensions of ∗1 and ∗2 , respectively. For every group homomorphism ∶ X1 → X2 and every g2 ∈ X2 ,
the map h∶ X1 → X2 defined by
h(x) = g2 ∗2 (x), x ∈ X1 ,
is a homomorphism of n-ary semigroups.
Conversely, every homomorphism from (X1 , F1 ) to (X2 , F2 ) is obtained in this way.
4 Reducibility of symmetric n-ary bands
In this section, we use Theorem 3.12 in order to analyze the reducibility problem for symmetric n-ary bands. We thus
consider a symmetric n-ary band (X, F ).
Proposition 4.1. If F is reducible to an associative operation G∶ X 2 → X, then the following assertions hold.
(i) G is surjective and symmetric;
(ii) The n-ary semilattice congruence ∼ associated with F is a binary semilattice congruence for G.
It follows from Proposition 4.1 that if F is reducible to G, then G induces an operation G�[x]2∼ on every equivalence
class [x]∼ of X�∼. This operation is a binary reduction of F �[x]n∼ . Therefore, it is natural to study the properties of the
reductions of such operations. This is performed in the following result.
Proposition 4.2. If (X, F ) is the n-ary extension of an Abelian group (X, G1 ) whose exponent divides n − 1, then
every reduction (X, G2 ) of F is a group that is isomorphic to (X, G1 ). Moreover, all the reductions of (X, F ) are
obtained by using (2) with any element e of X.
We are now able to analyze the reducibility of symmetric n-ary bands.
Proposition 4.3. A symmetric n-ary band (X, F ) = ((Y, ⋏n−1 ); (X↵ , F↵ ); '↵, ) is reducible to a semigroup if and
only if there exists a map e∶ Y → X such that
(i) For every ↵ ∈ Y , e(↵) = e↵ belongs to X↵ ;
(ii) For every ↵, ∈ Y such that ↵ � , we have '↵, (e↵ ) = e .
Moreover, when (X, F ) is reducible to a semigroup, a reduction is given by the semigroup decomposed as
((Y, ⋏n−1 ); (X↵ , G↵ ); '↵, ), where G↵ is the reduction of F↵ with respect to e↵ .
Example 4.4. For the structures of Example 1.1(a) and (b), respectively, the only non obvious homomorphisms are
given by '[1]∼ ,[3]∼ = `3 �[1]∼ , '[2]∼ ,[3]∼ = `3 �[2]∼ , and '[1]∼ ,[2]∼ = `2 �[1]∼ , respectively.
• For (X, F1 ), we have '[1]∼ ,[3]∼ (1) = `3 (1) = 4 and '[2]∼ ,[3]∼ (2) = `3 (2) = 3.
• For (X, F2 ), we have '[1]∼ ,[2]∼ (1) = `2 (1) = 2.
Recall that an n-ary group is an n-ary semigroup (X, F ) such that for any i ∈ {1, . . . , n} and any x1 , . . . , xi−1 , xi+1 , . . . , xn , y ∈
5
X there exists a unique z ∈ X such that F (x1 , . . . , xi−1 , z, xi+1 , . . . , xn ) = y.
By Proposition 3.13 these are homomorphisms from the ternary extension of the trivial group to the ternary extension of
(Z2 , +). It is easy to see that (X, F1 ) and (X, F2 ), respectively, are the strong ternary semilattices associated with the
semilattices whose Hasse diagrams are depicted in Figure 1 (left) and (right), respectively, and the ternary extensions of
groups and homomorphisms given here. It follows from Theorem 3.12 that (X, F1 ) and (X, F2 ) are symmetric ternary
bands. Finally, we can use Proposition 4.3 to analyze the reducibility problem for (X, F1 ) and (X, F2 ).
1. For (X, F1 ) we must have e([1]∼ ) = 1 and e([2]∼ ) = 2. Then we must have e([3]∼ ) = '[1]∼ ,[3]∼ (1) = 4 but
also e([3]∼ ) = '[2]∼ ,[3]∼ (2) = 3, a contradiction. So (X, F1 ) is not reducible to a semigroup.
2. For (X, F2 ) the map e defined by e([1]∼ ) = 1 and e([2]∼ ) = 2 satisfies the conditions of Proposition 4.3 and
so (X, F2 ) is reducible to a semigroup (X, G). The operation table of G is given below.
G 1 2 3
1 1 2 3
2 2 2 3
3 3 3 2
Acknowledgments
The first author is supported by the Luxembourg National Research Fund under the project PRIDE 15/10949314/GSM.
References
[1] Nathanael L. Ackerman. A characterization of quasitrivial n-semigroups. To appear in Algebra Universalis.
[2] Alfred H. Clifford. Bands of semigroups, Proc. Amer. Math. Soc., 5, 499–504, 1954.
[3] Alfred H. Clifford and Gordon B. Preston. The algebraic theory of semigroups. American Mathematical Society,
Vol. 1, 1961.
[4] Miguel Couceiro and Jimmy Devillet. Every quasitrivial n-ary semigroup is reducible to a semigroup. Algebra
Universalis, 80(4), 2019.
[5] Miguel Couceiro, Jimmy Devillet, Jean-Luc Marichal, and Pierre Mathonet. Reducibility of n-ary semigroups:
from quasitriviality towards idempotency. Submitted for publication. arXiv preprint arXiv:1909.10412, 2019.
[6] Jimmy Devillet, Gergely Kiss, and Jean-Luc Marichal. Characterizations of quasitrivial symmetric nondecreasing
associative operations. Semigroup Forum, 98(1), 154–171, 2019.
[7] Jimmy Devillet and Pierre Mathonet. On the structure of symmetric n-ary bands. Submitted for publication. arXiv
preprint arXiv:2004.12423 2020.
[8] Wisław A. Dudek and Vladimir V. Mukhin. On n-ary semigroups with adjoint neutral element. Quasigroups and
Related Systems, 14:163–168, 2006.
[9] Wilhelm Dörnte. Untersuchungen über einen verallgemeinerten Gruppenbegriff. Math. Z., 29:1–19, 1928.
[10] Pierre A. Grillet. Commutative semigroups. Springer, Boston, MA, 2001.
[11] Jerzy W. Grzymala-Busse. Automorphisms of polyadic automata. J. Assoc. Comput. Mach., 16:208–219, 1969.
[12] John M. Howie. Fundamentals of semigroup theory. Oxford University Press, 1995.
[13] Gergely Kiss and Gabor Somlai. Associative idempotent nondecreasing functions are reducible. Semigroup
Forum, 98(1), 140–153, 2019.
[14] Kenneth Krohn and John Rhodes. Algebraic theory of machines. I. Prime decomposition theorem for finite
semigroups and machines. Trans. Amer. Math. Soc., 116, 450–464, 1965.
[15] Helmut Länger. The free algebra in the variety generated by quasi-trivial semigroups. Semigroup Forum,
20:151–156, 1980.
[16] Charles F. Laywine and Gary L. Mullen. Discrete mathematics using latin squares. Wiley, New York, 1998.
[17] Charles F. Laywine, Gary L. Mullen, and Geoff Whittle. D-dimensional hypercubes and the Euler and MacNeish
conjectures. Montsh. Math., 111:223–238, 1995.
[18] Erkko Lehtonen and Florian Starke. On associative operations on commutative integral domains. Semigroup
Forum, 100(3), 910–915, 2020.
[19] David McLean. Idempotent semigroups. Amer. Math. Monthly, 61:110–113, 1954.
[20] Mario Petrich. Introduction to semigroups. Charles E. Merrill, 1973.
[21] Mario Petrich. Lectures in semigroups. John Wiley, London, 1977.
[22] Emil L. Post. Polyadic groups. Trans. Amer. Math. Soc., 48:208–350, 1940.
[23] Joseph J. Rotman, An Introduction to the Theory of Groups. 4th Ed. Springer-Verlag, New York, USA, 1995.