1 Introduction

The concept of an Abel-Grassmann’s groupoid (\({\mathcal{AG }}\)-groupoid) was first given by Kazim and Naseeruddin in 1972 [3] and they have called it a left almost semigroup (LA-semigroup). In [2], the same structure is called a left invertive groupoid. Several examples and interesting properties of LA-semigroups can be found in [4, 7].

The most generalized algebraic structure is a groupoid which is a set together with a binary operation. A groupoid satisfying the following left invertive law is known as an \({\mathcal{AG }}\)-groupoid [9].

$$\begin{aligned} (ab)c=(cb)a,\quad \text{ for} \text{ all} \;a,b,c\in \mathcal S . \end{aligned}$$

This left invertive law has been obtained by introducing braces on the left of ternary commutative law \(abc=cba\).

In an \({\mathcal{AG }}\)-groupoid \(\mathcal S \), the medial law holds [3]

$$\begin{aligned} (ab)(cd)=(ac)(bd),\quad \text{ for} \text{ all} \;a,b,c,d\in \mathcal S . \end{aligned}$$

Since \({\mathcal{AG }}\)-groupoids satisfy medial law, they belong to the class of entropic groupoids which are also called abelian quasigroups [10].

An \({\mathcal{AG }}\)-groupoid may or may not contains a left identity. The left identity of an \({\mathcal{AG }}\)-groupoid allow us to introduce the inverses of elements in an \({\mathcal{AG }}\)-groupoid. If an \({\mathcal{AG }}\)-groupoid contains a left identity, then it is unique [7].

In an \({\mathcal{AG }}\)-groupoid \(\mathcal S \) with left identity, the paramedial law holds [7]

$$\begin{aligned} (ab)(cd)=(dc)(ba),\quad \text{ for} \text{ all} \;a,b,c,d\in \mathcal S . \end{aligned}$$

Further if an \({\mathcal{AG }}\)-groupoid \(\mathcal S \) contains a left identity, the following law holds [7]

$$\begin{aligned} a(bc)=b(ac) ,\quad \text{ for} \text{ all} \;a,b,c\in \mathcal S . \end{aligned}$$

If an \({\mathcal{AG }}\)-groupoid \(\mathcal S \) satisfies the above law without left identity, then it is called an \({\mathcal{AG }}\)**-groupoid. An \(\mathcal AG \)**-groupoid also satisfies a paramedial law without left identity\(.\) An \( {\mathcal{AG }}\)**-groupoid is the generalization of an \({\mathcal{AG }}\)-groupoid with left identity. Every \({\mathcal{AG }}\)-groupoid with left identity is an \( {\mathcal{AG }}\)**-groupoid but the converse is not true in general [1].

An \({\mathcal{AG }}\)-groupoid is a useful algebraic structure, midway between a groupoid and a commutative semigroup with wide applications in theory of flocks [8]. An \({\mathcal{AG }}\)-groupoid is non-associative and non-commutative in general, however, there is a close relation with semigroup as well as with commutative structures. It has been investigated in [7] that if an \({\mathcal{AG }}\)-groupoid contains a right identity, then it becomes a commutative monoid. An \({\mathcal{AG }}\)-groupoid is the generalization of a semigroup theory [7] and has vast applications in collaboration with semigroup like other branches of mathematics. The connections of an \({\mathcal{AG }}\)-groupoid with the vector spaces over finite fields have been investigated in [4].

From the above discussion, we see that \({\mathcal{AG }}\)-groupoids have very closed links with semigroups and vector spaces which make an \({\mathcal{AG }}\) -groupoid to be among the most interesting non-associative algebraic structure.

2 \({\mathcal{AG }}\)-groupoids generated by other algebraic structures

Example 1

Define a binary operation “\(\diamond \)” on a commutative inverse semigroup \(( \mathcal S \), .) as follows:

$$\begin{aligned} a\diamond b=ba^{-1} \prod \limits _{i=1}^{n}r_{i}^{-1},\quad \text{ where} \; a,b,r_{i}\in \mathcal S = \mathbb R \backslash \{0\}, \text{ for} \;i=1,2,\ldots ,n. \end{aligned}$$

Then \((\mathcal S ,\diamond )\) becomes an \({\mathcal{AG }}\)-groupoid. Indeed

$$\begin{aligned} (a\diamond b)\diamond c&= ba^{-1}\prod \limits _{i=1}^{n}r_{i}^{-1}\diamond c \\&= c\left( ba^{-1}\prod \limits _{i=1}^{n}r_{i}^{-1}\right) ^{-1}\prod \limits _{i=1}^{n}r_{i}^{-1} \\&= c\left( \prod \limits _{i=1}^{n}r_{i}^{-1}\right) ^{-1}ab^{-1}\prod \limits _{i=1}^{n}r_{i}^{-1}\\&= a\left(\prod \limits _{i=1}^{n}r_{i}^{-1}\right) ^{-1}cb^{-1}\prod \limits _{i=1}^{n}r_{i}^{-1}, \quad \text{ as} \;(\mathcal S ,.) \text{ is} \text{ commutative} \\&= a\left( bc^{-1}\prod \limits _{i=1}^{n}r_{i}^{-1}\right) ^{-1}\prod \limits _{i=1}^{n}r_{i}^{-1} \\&= bc^{-1}\prod \limits _{i=1}^{n}r_{i}^{-1}\diamond a \\&= (c\diamond b)\diamond a. \end{aligned}$$

 

It is not hard to see that \((\mathcal S ,\diamond )\) is non-commutative and non-associative.

Example 2

Let us consider an abelian group \(( \mathbb R ,+)\) of all real numbers under the binary operation of addition. If we define

$$\begin{aligned} a\odot b=b-a-\left(\sum \limits _{i=1}^{n}r_{i}\right),\quad \text{ where} \;a,b,r_{i}\in \mathbb R , \text{ for} \;i=1,2,\ldots ,n. \end{aligned}$$

Then \((\mathbb R ,\odot )\) becomes an \({\mathcal{AG }}\)-groupoid. Indeed

$$\begin{aligned} (a\odot b)\odot c&= c-(a\odot b)-\left( \sum \limits _{i=1}^{n}r_{i}\right) \\&= c-\left\{ b-a-\left( \sum \limits _{i=1}^{n}r_{i}\right) \right\} -\left( \sum \limits _{i=1}^{n}r_{i}\right) \\&= c-b+a+\left( \sum \limits _{i=1}^{n}r_{i}\right) -\left( \sum \limits _{i=1}^{n}r_{i}\right) \\&= a-b+c+\left( \sum \limits _{i=1}^{n}r_{i}\right) -\left( \sum \limits _{i=1}^{n}r_{i}\right), \quad \text{ as} \;( \mathbb R ,+) \text{ is} \text{ an} \text{ abelian} \text{ group} \\&= a-\left\{ b-c-\left( \sum \limits _{i=1}^{n}r_{i}\right) \right\} -\left( \sum \limits _{i=1}^{n}r_{i}\right) \\&= a-(c\odot b)-\left( \sum \limits _{i=1}^{n}r_{i}\right) \\&= (c\odot b)\odot a. \end{aligned}$$

One can easily verify that \(( \mathbb R ,\odot )\) is non-commutative and non-associative.

3 Some algebraic structures generated by \({\mathcal{AG }}\)-groupoids

In [10], it has been shown that an \(\mathcal{AG }\)*-groupoid \(( \mathcal S ,.)\) under the binary operation “\(\circ \)” (sandwich operation) defined as follows:

$$\begin{aligned} x\circ y=(xa)y, \quad \text{ for} \text{ all} \; x,y\in \mathcal S \text{ and} \;a\in \mathcal S \text{ is} \text{ fixed,} \end{aligned}$$
(1)

becomes a commutative semigroup \((\mathcal S ,\circ )\).

Example 3

An \(\mathcal{AG }\)**-groupoid \((\mathcal S ,.)\) under the sandwich operation “\(\circ \)” [defined in (1)] also becomes a commutative semigroup \(( \mathcal S ,\circ )\).

Example 4

Let \((\mathcal S \), .) be an \(\mathcal{AG }\)-groupoid with left identity e . Define a binary operation “\(\circ _{e}\)” (e-sandwich operation) as follows:

$$\begin{aligned} a\circ _{e}b=(ae)b,\quad \text{ for} \text{ all} \;a,b\in \mathcal S . \end{aligned}$$

Then \((\mathcal S \), \(\circ _{e})\) becomes a commutative monoid. Indeed

$$\begin{aligned} a\circ _{e}b=(ae)b=(be)a=b\circ _{e}a, \end{aligned}$$

and

$$\begin{aligned} (a\circ _{e}b)\circ _{e}c&= ((ae)b)\circ _{e}c=(((ae)b)e)c \\&= (((be)a)e)c=(ce)((be)a) \\&= (a(be))(ec)=(ae)((be)c) \\&= a\circ _{e}((be)c)=a\circ _{e}(b\circ _{e}c), \end{aligned}$$

also

$$\begin{aligned} a\circ _{e}e=(ae)e=(ee)a=a=ea=(ee)a=e\circ _{e}a. \end{aligned}$$

4 Some studies in fully regular\(\mathcal{AG }\)-groupoids

Definition 1

[5] An element \(a\) of an \(\mathcal{AG }\)-groupoid \(\mathcal S \) is called a regular element of \(\mathcal S \) if there exists some \(x\in \mathcal S \) such that \(a=(ax)a\) and \(\mathcal S \) is called regular \( \mathcal{AG }\)-groupoid if all elements of \(\mathcal S \) are regular.

Definition 2

[5] An element \(a\) of an \(\mathcal{AG }\)-groupoid \(\mathcal S \) is called an intra-regular element of \(\mathcal S \) if there exist some \( u,v,x,y\in \mathcal S \) such that \(a=(ua)(av)=(xa^{2})y\) and \(\mathcal S \) is called intra-regular if all elements of \(\mathcal S \) are intra-regular.

Definition 3

[5] An element a of an \(\mathcal{AG }\)-groupoid \(\mathcal S \) is called a weakly regular element of \(\mathcal S \) if there exist some \(x,y\in \mathcal S \) such that \(a=(ax)(ay)\) and \(\mathcal S \) is called weakly regular if all elements of \(\mathcal S \) are weakly regular.

Definition 4

[5] An element \(a\) of an \(\mathcal{AG }\)-groupoid \(\mathcal S \) is called a right (left) regular element of \(\mathcal S \) if there exists some \(x\in \mathcal S \) such that \(a=a^{2}x=(aa)x\) \(( a=xa^{2}=x(aa)) \) and \(\mathcal S \) is called right (left) regular if all elements of \(\mathcal S \) are right (left) regular.

Definition 5

[5] An element \(a\) of an \(\mathcal{AG }\)-groupoid \(\mathcal S \) is called a left quasi regular element of \(\mathcal S \) if there exist some \( x,y\in \mathcal S \) such that \(a=(xa)(ya)\) and \(\mathcal S \) is called left quasi regular if all elements of \(\mathcal S \) are left quasi regular.

Definition 6

[5] An element \(a\) of an \(\mathcal{AG }\)-groupoid \(\mathcal S \) is called a \((2,2)\)-regular element of \(\mathcal S \) if there exists some \(x\in \mathcal S \) such that \(a=(a^{2}x)a^{2}\) and \(\mathcal S \) is called \((2,2)\)-regular \(\mathcal{AG }\)-groupoid if all elements of \(\mathcal S \) are \((2,2)\)-regular.

Definition 7

[5] An element \(a\) of an \(\mathcal{AG }\)-groupoid \(\mathcal S \) is called a completely regular element of \(\mathcal S \) if \(a\) is regular, right regular and left regular. An \(\mathcal{AG }\)-groupoid \(\mathcal S \) is called completely regular if it is regular, right and left regular.

In [5], it has been shown that all of these classes coincide in an \(\mathcal{AG }\)-groupoid with left identity except a regular class. It is important to note that none of any two mention classes coincide in a semigroup.

Here we introduce a new class of an \(\mathcal{AG }\)-groupoid as follows:

An element \(a\) of an \(\mathcal{AG }\)-groupoid \(\mathcal S \) is called a fully regular element of \(\mathcal S \) if there exist some \(p,q,r,s,t,u,v,w,x,y,z \in \mathcal S \) \((p,q,\ldots ,z\) may be repeated) such that

$$\begin{aligned} a=(pa^{2})q=(ra)(as)=(at)(au)=(aa)v=w(aa)=(xa)(ya)=(a^{2}z)a^{2}. \end{aligned}$$

An \(\mathcal{AG }\)-groupoid \(\mathcal S \) is called fully regular if all elements of \(\mathcal S \) are fully regular.

Example 5

Let \(\mathcal S =\{1,2,3,4,5,6,7\}\) be an \(\mathcal{AG }\)-groupoid with the following multiplication table.

 

figure a1

It is easy to verify that \(\mathcal S \) is fully regular \(\mathcal{AG }\)-groupoid.

Remark 1

An \(\mathcal{AG }\)-groupoid \(\mathcal S \) with left identity \( (\mathcal{AG }\)**-groupoid) is fully regular if and only if \(\mathcal S \) is intra-regular [weakly regular, right regular, left regular, left quasi regular, (2,2)-regular and completely regular].

Every intra-regular [weakly regular, right regular, left regular, left quasi regular, (2,2)-regular and completely regular] class of an \(\mathcal{AG }\)-groupoid \(\mathcal S \) with left identity \((\mathcal{AG }\)**-groupoid) is regular but the converse is not true in general [5].

Example 6

If we consider an \(\mathcal{AG }\)-groupoid \(\mathcal S =\{ 1,2,3,4\}\) with left identity 3 in the following Cayley’s table.

 

figure a2

Then by routine calculation, it is easy to see that \(\mathcal S \) is regular. Note that \(\mathcal S \) is not fully regular, because \(1\in \mathcal S \) is not a fully regular element of \(\mathcal S \).

Remark 2

An \(\mathcal{AG }\)-groupoid \(\mathcal S \) with left identity \((\mathcal{AG }\) **-groupoid) is regular if \(\mathcal S \) is fully regular but the converse is not valid in general which can be followed from Example 6.

If \(\mathcal S \) is an intra-regular [weakly regular, right regular, left regular, left quasi regular, (2,2)-regular or completely regular] \(\mathcal{AG }\)-groupoid\(, \)then \(\mathcal{S=S }^{2}\) holds but the converse is not true in general [5].

Remark 3

If an \(\mathcal{AG }\)-groupoid \(\mathcal S \) is fully regular, then \(\mathcal{S=S }^{2}\) holds but the converse is not valid in general which can be followed from Example 6.

Definition 8

A non-empty subset \(\mathcal A \) of an \(\mathcal{AG }\)-groupoid \(\mathcal S \) called left (right) ideal of \(\mathcal S \) if and only if \(\mathcal{SA } \subseteq \mathcal A \) \((\mathcal{AS }\subseteq \mathcal A )\) and is called two-sided ideal or ideal of \(\mathcal S \) if and only if it is both left and right ideal of \(\mathcal S \).

Definition 9

A non-empty subset \(\mathcal A \) of an \(\mathcal{AG }\)-groupoid \(\mathcal S \) called semiprime if and only if \(a^{2}\in \mathcal A \Longrightarrow a\in \mathcal A \).

Lemma 1

For an \(\mathcal{AG }\)-groupoid \(\mathcal S \) with left identity, the following conditions are equivalent.

  1. (i)

    \(\mathcal S \) is fully regular.

  2. (ii)

    \(\mathcal R \cap \mathcal L =\mathcal{RL }\), where \(\mathcal R \) and \(\mathcal L \) are any right and left ideals of \(\mathcal S \) respectively such that \(\mathcal R \) is semiprime.

Proof

\(\mathrm{(i)}\Longrightarrow \mathrm{(ii)}\): Suppose that an \(\mathcal{AG }\)-groupoid \(\mathcal S \) with left identity is fully regular. Now by using Remark 1, \(\mathcal S \) is an intra-regular. Let \(\mathcal R \) and \(\mathcal L \) be any right and left ideals of \(\mathcal S \) respectively and let \(a\in \mathcal S \), then there exist \(x,y\in \mathcal S \) such that \(a=(xa^{2})y\). Now let \(a^{2}\in \mathcal R \), then

$$\begin{aligned} a&= (xa^{2})y=((ex)(aa))y=((aa)(xe))y \\&= (y(xe))(aa)=(aa)((xe)y)\in \mathcal{RS\subseteq R }. \end{aligned}$$

Thus \(\mathcal R \) is semiprime. It is easy to see that \(\mathcal{ RL\subseteq R }\cap \mathcal L \). Clearly \(a^{2}\in \mathcal S a^{2}\) and also \(\mathcal S a^{2}\) right ideal of \(\mathcal S \), therefore \(a\in a^{2} \mathcal S \). Let \(a\in \mathcal R \cap \mathcal L \), then \(a\in \mathcal R \) and \(a\in \mathcal L \). Thus

$$\begin{aligned} a \in a^{2}\mathcal S&= (aa)(\mathcal{SS })=(a\mathcal S )(a\mathcal S )=( \mathcal S a)(\mathcal S a) \\&= ((\mathcal{SS })a)(\mathcal S a)=((a\mathcal S )\mathcal S )(\mathcal S a) \\&\subseteq ((\mathcal{RS })\mathcal S )(\mathcal{SL })\subseteq \mathcal{RL } , \end{aligned}$$

which shows that \(\mathcal R \cap \mathcal L =\mathcal{RL }\), where \(\mathcal R \) and \(\mathcal L \) are any right and left ideals of \(\mathcal S \) respectively such that \(\mathcal R \) is semiprime. \((ii)\Longrightarrow (i)\): Let every right ideal \(\mathcal R \) be semiprime and \(\mathcal L \) be any left ideal of an \(\mathcal{AG }\)-groupoid \(\mathcal S \) with left identity such that \(\mathcal R \cap \mathcal L =\mathcal RL \). Since \(\mathcal S a^{2}\) and \(\mathcal S a\) are any right and left ideals of \(\mathcal S \) respectively, then it is easy to see that \(a\in a^{2}\mathcal S \) and \(a\in \mathcal S a\), therefore

$$\begin{aligned} a \in (a^{2}\mathcal S )\cap (\mathcal S a)&= (a^{2}\mathcal S )(\mathcal S a)=((aa)(\mathcal{SS }))(\mathcal S a) \\&= ((\mathcal {SS} )(aa))(\mathcal S a)\subseteq (\mathcal S a^{2})\mathcal S , \end{aligned}$$

that is \(a=(xa^{2})y\) for some \(x,y\in \mathcal S \). Thus by using Remark 1, \(\mathcal S \) is fully regular. \(\square \)

Lemma 2

A non-empty subset \(\mathcal A \) of a fully regular \(\mathcal{AG } \)-groupoid \(\mathcal S \) with left identity is a left ideal of \(\mathcal S \) if and only if it is a right ideal of \(\mathcal S \).

Proof

It is simple. \(\square \)

Lemma 3

For an \(\mathcal{AG }\)-groupoid \(\mathcal S \) with left identity, the following conditions are equivalent.

  1. (i)

    \(\mathcal S \) is fully regular.

  2. (ii)

    \(\mathcal L \cap \mathcal R =\mathcal LR ,\) where \(\mathcal L \) and \(\mathcal R \) are any left and right ideals of \( \mathcal S \) respectively such that \(\mathcal R \) is semiprime.

Proof

\((i)\Longrightarrow (ii)\) can be followed by using Lemmas 1 and 2.

\((ii)\Longrightarrow (i)\) is straightforward. \(\square \)

Definition 10

A non-empty subset \(\mathcal A \) of an \(\mathcal{AG }\)-groupoid \(\mathcal S \) called idempotent if and only if \(\mathcal{A=A }^{2}\).

Corollary 1

For an \(\mathcal{AG }\)-groupoid \(\mathcal S \) with left identity, the following conditions are equivalent.

  1. (i)

    \(\mathcal S \) is fully regular.

  2. (ii)

    Every left ideal of \(\mathcal S \) is idempotent.

Corollary 2

For an \(\mathcal{AG }\)-groupoid \(\mathcal S \) with left identity, the following conditions are equivalent.

  1. (i)

    \(\mathcal S \) is fully regular.

  2. (ii)

    Every right ideal of \(\mathcal S \) is idempotent and semiprime.

Theorem 1

For an \(\mathcal{AG }\)-groupoid \(\mathcal S \) with left identity, the following conditions are equivalent.

  1. (i)

    \(\mathcal S \) is fully regular.

  2. (ii)

    \(\mathcal L =\mathcal L ^{3},\) where \(\mathcal L \) is any left ideal of \(\mathcal S \).

Proof

\((i)\Longrightarrow (ii):\) Let \(\mathcal S \) be a fully regular \(\mathcal{AG } \)-groupoid with left identity and \(\mathcal L \) be any left ideal of \(\mathcal S \). Then by using Corollory 1, we have

$$\begin{aligned} \mathcal L ^{3}=\mathcal L ^{2}\mathcal{L=LL\subseteq SL\subseteq L }. \end{aligned}$$

Now let \(a\in \mathcal L \), then by using Lemmas 1 and 2, \( \mathcal S a^{2}\) is left ideal of \(\mathcal S \) such that \(a\in a^{2} \mathcal S \), therefore

$$\begin{aligned} a \in a^{2}\mathcal S&= (aa)\mathcal S =(\mathcal S a)a\subseteq (\mathcal S (a^{2}\mathcal S ))a \\&= (a^{2}(\mathcal{SS }))a=((aa)\mathcal S )a=((\mathcal S a)a)a \\&\subseteq ((\mathcal{SL })\mathcal L )\mathcal L \subseteq (\mathcal{LL }) \mathcal L\subseteq L ^{3}, \end{aligned}$$

which is what we set out to prove. \((ii)\Longrightarrow (i):\) Let \(\mathcal L \) be any left ideal of an \(\mathcal{AG }\)-groupoid \(\mathcal S \) with left identity such that \(\mathcal L =\mathcal L ^{3}\). Since \(\mathcal S a\) is left ideals of \(\mathcal S \) and \(a\in \mathcal S a\), therefore

$$\begin{aligned} a\in \mathcal S a=((\mathcal S a)(\mathcal S a))(\mathcal S a)=((\mathcal{SS } )(aa))(\mathcal S a)\subseteq (\mathcal S a^{2})\mathcal S , \end{aligned}$$

that is \(a=(xa^{2})y\) for some \(x,y\in \mathcal S \). Thus by using Remark 1, \(\mathcal S \) is fully regular. \(\square \)

Theorem 2

For an \(\mathcal{AG }\)-groupoid \(\mathcal S \) with left identity, the following conditions are equivalent.

  1. (i)

    \(\mathcal S \) is fully regular.

  2. (ii)

    \(\mathcal{L=L }^{i+1},\) for any left ideal \(\mathcal L \) of \( \mathcal S \), where \(i=1,\ldots ,n\).

Proof

It can be easily followed by generalizing the proof of Theorem 1.

From the left–right dual of Theorem 2, we have the following theorem.

\(\square \)

Theorem 3

For an \(\mathcal{AG }\)-groupoid \(\mathcal S \) with left identity, the following conditions are equivalent.

  1. (i)

    \(\mathcal S \) is fully regular.

  2. (ii)

    \(\mathcal{R=R }^{i+1},\) for any right ideal \(\mathcal R \) of \(\mathcal S \) such that \(\mathcal R \) is semiprime, where \(i=1,\ldots ,n\).

Definition 11

An \(\mathcal{AG }\)-groupoid is called left (right) simple if and only if it has no proper left (right) ideal and is called simple if and only if it has no proper two-sided ideal.

Note that if an \(\mathcal{AG }\)-groupoid \(\mathcal S \) contains a left identity, then \(\mathcal S =\mathcal S ^{2}.\)

Theorem 4

The following conditions are equivalent for an \(\mathcal{AG }\) -groupoid \(\mathcal S \) with left identity.

  1. (i)

    \(a\mathcal S =\mathcal S \), some \(a\in \mathcal S \).

  2. (ii)

    \(\mathcal S a=\mathcal S \), for some \(a\in \mathcal S \).

  3. (iii)

    \(\mathcal S \) is simple.

  4. (iv)

    \(\mathcal{AS=S=SA }\), where \(\mathcal A \) two-sided ideal of \(\mathcal S \).

  5. (v)

    \(\mathcal S \) is fully regular.

Proof

\((i)\Longrightarrow (ii)\): Let \(\mathcal S \) be an \(\mathcal{AG }\)-groupoid with left identity and assume that \(a\mathcal S =\mathcal S \) holds for some \(a\in \mathcal S \), then

$$\begin{aligned} \mathcal S =\mathcal SS =(a\mathcal S )\mathcal S =(\mathcal SS )a=\mathcal S a. \end{aligned}$$

\((ii)\Longrightarrow (iii)\): Let \(\mathcal S \) be an \(\mathcal{AG }\)-groupoid with left identity such that \(a\mathcal S =\mathcal S \) holds for some \(a\in \mathcal S \). Suppose that \(\mathcal S \) is not left simple and let \( \mathcal L \) be a proper left ideal of \(\mathcal S \), then

$$\begin{aligned} \mathcal SL&\subseteq \mathcal L\subseteq S =\mathcal SS =(\mathcal S a)\mathcal S =((\mathcal SS )(ea))\mathcal S \\&= ((ae)(\mathcal SS ))\mathcal S =((ae)\mathcal S )(\mathcal SS ) \\&= ((\mathcal S e)a)(\mathcal SS )=(\mathcal SS )(a(\mathcal S e)) \\&= a((\mathcal SS )(\mathcal S e))\subseteq a\mathcal S , \end{aligned}$$

implies that \(sl=at\), for some \(a,s,t\in \mathcal S \) and \(l\in \mathcal L \). Since \(sl\in \mathcal L \), therefore \(at\in \mathcal L \), but \(at\in a \mathcal S \). Thus \(a\mathcal S \subseteq \mathcal L \) and therefore, we haves

$$\begin{aligned} \mathcal S =a\mathcal S \subseteq \mathcal L , \end{aligned}$$

implies that \(\mathcal S =\mathcal L \), which contradicts the given assumption. Thus \(\mathcal S \) is left simple and similarly we can show that \(\mathcal S \) is right simple, which shows that \(\mathcal S \) is simple.

\((iii)\Longrightarrow (iv)\): Let \(\mathcal S \) be a simple \(\mathcal{AG }\)-groupoid with left identity and let \(\mathcal A \) be any two-sided ideal of \(\mathcal S \), then \(\mathcal A=S \). Therefore, we have

$$\begin{aligned} \mathcal AS=SS=SA . \end{aligned}$$

\((iv)\Longrightarrow (v)\): Let \(\mathcal S \) be an \(\mathcal{AG }\)-groupoid with left identity such that \(\mathcal AS=S=SA \) holds for any two-sided ideal \(\mathcal A \) of \(\mathcal S \). Since \(\mathcal S a^{2}\) is a right ideal of \(\mathcal S \) and we know that every right ideal of \(\mathcal S \) with left identity is two-sided ideal of \(\mathcal S \). Thus \(\mathcal S a^{2}\) is two-sided ideal of \(\mathcal S \) such that \((a^{2}\mathcal S ) \mathcal S=S=S (a^{2}\mathcal S )\). Let \(a\in \mathcal S \), then

$$\begin{aligned} a \in \mathcal S&= (a^{2}\mathcal S )\mathcal S =((aa)(\mathcal SS ))\mathcal S \\&= ((\mathcal SS )(aa))\mathcal S =(\mathcal S a^{2})\mathcal S , \end{aligned}$$

that is \(a=(xa^{2})y\) for some \(x,y\in \mathcal S \). Thus by using Remark 1, \(\mathcal S \) is fully regular. \((v)\Longrightarrow (i)\): Let \(\mathcal S \) be a fully regular \(\mathcal{AG }\) -groupoid with left identity, then by using remark 1, \(\mathcal S \) is an intra-regular. Let \(a\in \mathcal S \), then there exist \(x,y\in \mathcal S \) such that \(a=(xa^{2})y\). Thus

$$\begin{aligned} a&= (xa^{2})y=((ex)(aa))y=((aa)(ex))y \\&= (y(ex))(aa)=a((y(ex))a)\in a\mathcal S , \end{aligned}$$

which shows that \(\mathcal S \subseteq \mathcal S a\) and \(\mathcal S a\subseteq \mathcal S \) is obvious. Thus \(\mathcal S a=\mathcal S \) holds for some \(a\in \mathcal S \). \(\square \)

Corollary 3

The following conditions are equivalent for an \(\mathcal{AG }\)-groupoid \(\mathcal S \) with left identity.

  1. (i)

    \(a\mathcal S =\mathcal S \), for some \(a\in \mathcal S \).

  2. (ii)

    \(\mathcal S a=\mathcal S \), for some \(a\in \mathcal S \).

  3. (iii)

    \(\mathcal S \) is right simple.

  4. (iv)

    \(\mathcal{AS=S=SA }\), where \(\mathcal A \) is any right ideal of \(\mathcal S \).

  5. (v)

    \(\mathcal S \) is fully regular.

Corollary 4

If \(\mathcal S \) is an \(\mathcal{AG }\)-groupoid with left identity, then the following conditions are equivalent.

  1. (i)

    \(\mathcal S a=\mathcal S \), for some \(a\in \mathcal S \).

  2. (ii)

    \(a\mathcal S =\mathcal S \), for some \(a\in \mathcal S \).

Example 7

Let \(\mathcal S =\{a,b,c,d,e\}\) be an \(\mathcal{AG }\)-groupoid with left identity \(b\) in the following multiplication table.

figure a3

It is clear to see that \(\mathcal S a=\mathcal S \) or \(a\mathcal S =\mathcal S \) does not holds for all \(a\in \mathcal S \).

Corollary 5

If \(\mathcal S \) is an \(\mathcal{AG }\)-groupoid, then \(e\mathcal S =\mathcal S=S e\) holds for \(e\in \mathcal S ,\) where \(e\) is a left identity of \(\mathcal S \). Note that \(e\) does not act as a right identity of \(\mathcal S \). For example, an element \(b\in \mathcal S \) considered in Example 7 is a left identity of \(\mathcal S \) such that \(\mathcal S b=\mathcal S \) holds for \(b\in \mathcal S \) but \(cb=e\ne c=eb\ne e\) for \(c,e\in \mathcal S \).

Theorem 5

The following conditions are equivalent for an \(\mathcal{AG }\)-groupoid \( \mathcal S \) with left identity.

  1. (i)

    \(\mathcal S \) is fully regular.

  2. (ii)

    \(\mathcal S a=\mathcal S =a\mathcal S \), for some a \(\in \mathcal S \).

Proof

It can be easily followed by using Remark 1 and Theorem 4. \(\square \)

5 On some algebraic structures in terms of \(\mathcal{AG }\)-groupoids

Lemma 4

A groupoid \(\mathcal S \) with left identity is an \(\mathcal{AG }\)-groupoid if and only if \(\mathcal S \) satisfy medial and paramedial laws.

Proof

Let a groupoid \(\mathcal S \) with left identity be medial and paramedial and \(a,\) \(b,\) \(c\in \mathcal S \), then

$$\begin{aligned} (ab)c=(ab)(ec)=(ae)(bc)=(cb)(ea)=(cb)a. \end{aligned}$$

The converse is obvious. \(\square \)

Lemma 5

An \(\mathcal{AG }\)-groupoid \(\mathcal S \) with left identity \((\mathcal{AG }\)**-groupoid) is a semigroup if and only if \(a(bc)=(ca)b=(cb)a\) holds for all a, b, \(c\in \mathcal S \).

Proof

Let \(\mathcal S \) be an \(\mathcal{AG }\)-groupoid with left identity such that \(a(bc)=(ca)b=(cb)a\) holds for all \(a,\) \(b,\) \(c\in \mathcal S \). Then

$$\begin{aligned} a(bc)=(ca)b=(cb)a=(ab)c. \end{aligned}$$

Conversely suppose that an \(\mathcal{AG }\)-groupoid \(\mathcal S \) with left identity \((\mathcal{AG }\)**-groupoid) is semigroup, then

$$\begin{aligned} a(bc)&= b(ac)=(ba)c=(ca)b \\&= (ba)c=b(ac)=a(bc) \\&= (ab)c=(cb)a. \end{aligned}$$

Which is what we set out to prove. \(\square \)

Definition 12

An element \(a\) of an \(\mathcal{AG }\)-groupoid \(\mathcal S \) with left identity \(e\) is called left (right) invertive if and only if there exists \(a^{{\prime }}\in S\) such that \(a^{{\prime }}a=e\) \((aa^{{\prime }}=e)\) and \(a\) is called invertive if and only if it is both left and right invertive. An \(\mathcal{AG }\)-groupoid \(\mathcal S \) is called left (right) invertible if and only if every element of \(\mathcal S \) is left (right) invertive and \(\mathcal S \) is called invertible if and only if it is both left and right invertible.

Theorem 6

A left \((\)right\()\) invertible \(\mathcal{AG }\)-groupoid \(\mathcal S \) with left identity is an abelian group if and only if \(a(bc)=c(ba)\), for all a, b, \(c\in \mathcal S \).

Proof

Assume that \(\mathcal S \) be a left invertible \(\mathcal{AG }\)-groupoid with left identity \(e\), then

$$\begin{aligned} ae=a(ee)=e(ea)=ea=a, \end{aligned}$$

implies that \(e\) is an identity of \(\mathcal S \). Since \(\mathcal S \) is left invertible, then there exists \(a^{{\prime }}\in \mathcal S \) such that \(a^{{\prime }}a=e,\) then

$$\begin{aligned} aa^{{\prime }}=(ea)a^{{\prime }}=(a^{{\prime }}a)e=ee=e. \end{aligned}$$

Thus \(\mathcal S \) have inverses for each \(a\in \mathcal S \). Now

$$\begin{aligned} (ab)c&= (ab)(ec)=(ce)(ba)=a(b(ce))=a(e(cb)) \\&= a(cb)=c(ab)=(ec)(ab)=(ba)(ce) \\&= e(c(ba))=c(ba)=a(bc), \end{aligned}$$

implies that \(\mathcal S \) is associative, also

$$\begin{aligned} ab=a(eb)=b(ea)=ba, \end{aligned}$$

which shows that \(\mathcal S \) is commutative and therefore \(\mathcal S \) is an abelian group.

The converse is simple. \(\square \)

Corollary 6

An invertible \(\mathcal{AG }\)-groupoid \(\mathcal S \) with left identity is an abelian group if and only if \(a(bc)=c(ba)\), for all a, b, \(c\in \mathcal S \).

Definition 13

[6] An \(\mathcal{AG }\)-groupoid \(\mathcal S \) is called an \(\mathcal{AG }\)*-groupoid if the following holds:

$$\begin{aligned} (ab)c=b(ac),\quad \text{ for} \text{ all} \;a,b,c\in \mathcal S . \end{aligned}$$

In an \(\mathcal{AG }\)*-groupoid \(\mathcal S \), the following law holds [9]:

$$\begin{aligned} (x_{1}x_{2})(x_{3}x_{4})=(x_{\pi (1)}x_{\pi (2)})(x_{\pi (3)}x_{\pi (4)}), \end{aligned}$$
(2)

where \(x_{1},x_{2},x_{3},x_{4}\in \mathcal S \) and \(\{ \pi (1),\pi (2),\pi (3),\pi (4)\}\) means any permutation on the set \(\{1,2,3,4\}\). It is an easy consequence that if \(\mathcal S =\mathcal S ^{2}\), then \(\mathcal S \) becomes a commutative semigroup.

Many characteristics of a non-associative AG*-groupoid are similar to a commutative semigroup.

Theorem 7

A fully regular \(\mathcal{AG }\)*-groupoid becomes a semigroup.

Proof

It can be followed from Remark 3 and Eq. (2). \(\square \)