On Maximal Subsemigroups of Partial Baer-Levi Semigroups
Abstract
Suppose that X is an infinite set with |X | ≥ q ≥ ℵ0 and I(X) is the symmetric inverse semigroup defined on X. In 1984, Levi and Wood determined a class of maximal subsemigroups MA (using certain subsets A of X) of the Baer-Levi semigroup BL(q) = {α ∈ I(X): dom α = X and |X∖Xα| = q}. Later, in 1995, Hotzel showed that there are many other classes of maximal subsemigroups of BL(q), but these are far more complicated to describe. It is known that BL(q) is a subsemigroup of the partial Baer-Levi semigroup PS(q) = {α ∈ I(X) : |X∖Xα| = q}. In this paper, we characterize all maximal subsemigroups of PS(q) when |X | > q, and we extend MA to obtain maximal subsemigroups of PS(q) when |X | = q.
1. Introduction
2. Preliminaries
In this paper, means Y is a disjoint union of sets A and B. As usual, ∅ denotes the empty (one-to-one) mapping which acts as a zero for P(X). For each nonempty A⊆X, we write idA for the identity transformation on A: these mappings constitute all the idempotents in I(X) and belong to PS(q) precisely when |X∖A | = q.
- (a)
〈M ∪ {a}〉 = S for all a ∈ S∖M;
- (b)
for any a, b ∈ S∖M, a can be written as a finite product of elements of M ∪ {b} (note that a is not expressible as a product of elements of M since a ∉ M).
Throughout this paper, we will use this fact to show the maximality of subsemigroups of PS(q).
3. Maximal Subsemigroups of PS(q) When p > q
The characterisation of maximal subsemigroups of a given semigroup is a natural topic to consider when studying its structure. Sometimes, it is difficult to describe all of them (see [6, 7], e.g.), but for a semigroup with some special properties, we can easily describe some of its maximal subsemigroups.
Lemma 3.1. Let S be a semigroup and suppose that S is a disjoint union of a subsemigroup T and an ideal I of S. Then,
- (a)
for any maximal subsemigroup M of T, M ∪ I is a maximal subsemigroup of S;
- (b)
for any maximal subsemigroup N of S such that T∖N ≠ ∅ and T∩N ≠ ∅, the set T∩N is a maximal subsemigroup of T.
Proof. To see that (a) holds, let M be a maximal subsemigroup of T. Since I is an ideal, we have M ∪ I is a subsemigroup of S. Clearly, M ∪ I⊊T ∪ I = S. If a ∈ S∖(M ∪ I), then a ∈ T∖M and thus T = 〈M ∪ {a}〉⊆〈M ∪ I ∪ {a}〉. Since 〈M ∪ I ∪ {a}〉 contains I, we have S = T ∪ I = 〈M ∪ I ∪ {a}〉 and so M ∪ I is maximal in S as required.
To prove (b), let N be a maximal subsemigroup of S, where T∖N ≠ ∅ and T∩N ≠ ∅, and let a ∈ T∖N. Since N is maximal in S, we have 〈N ∪ {a}〉 = S. Thus, for each b ∈ T∖N, b = c1c2 ⋯ cn for some natural n and some ci ∈ N ∪ {a} for all i = 1,2, …, n. Since b ∉ N, we have ci = a for some i. Moreover, since b ∉ I, we have cj ∈ T∩N for all j ≠ i. It follows that T∖N⊆〈(T∩N)∪{a}〉, therefore
Corollary 3.2. Suppose that p > r > q ≥ ℵ0. If M is a maximal subsemigroup of Sr, then is a maximal subsemigroup of PS(q).
Lemma 3.3. Let p > q ≥ ℵ0 and suppose that M is a maximal subsemigroup of PS(q). Then,
- (a)
Sr∩M ≠ ∅ for all q ≤ r < p;
- (b)
if there exists α ∉ M with g(α) < p, then Sk∖M ≠ ∅ for some q ≤ k < p.
Proof. To show that (a) holds, we first note that Sq is contained in Sr for all q ≤ r < p. If Sq∩M = ∅, then and thus by the maximality of M. But where is a subsemigroup of PS(q) (since is an ideal), so we get a contradiction. Therefore, ∅ ≠ Sq∩M⊆Sr∩M for all q ≤ r < p.
To show that (b) holds, suppose there is α ∉ M with g(α) = k < p. If k < q, then α ∈ Sr∖M for all q ≤ r ≤ p. Otherwise, if q ≤ k, then α ∈ Sk∖M. Hence (b) holds.
Then G0 = BL(q) and Gq = R(q). Moreover, if p > q and r > q, then Gr = Sr∩Tr, and so Gr is a subsemigroup of Sr (since it is the intersection of two semigroups). Also, Gr is bisimple and idempotent-free, when p > q and r > q (see [3, Corollary 3]).
From [3, Theorem 5], if p ≥ q, then Sq = α · R(q) for each α ∈ BL(q), and by [3, Theorem 6], Sq = BL(q) · μ · BL(q) for each μ ∈ R(q) when p ≠ q.
This motivates the following result.
Lemma 3.4. Suppose that p ≥ r > q ≥ ℵ0. Then Gr = BL(q) · α · BL(q) for each α ∈ Gr.
Proof. Let α ∈ Gr and β, γ ∈ BL(q). Since
For the converse, if α, β ∈ Gr, then |X∖dom α | = r = |X∖dom β|. Since p > q, every element in PS(q) has rank p, so we write
Now we can describe all maximal subsemigroups of PS(q) when p > q.
Theorem 3.5. Suppose that p > q ≥ ℵ0. Then M is a maximal subsemigroup of PS(q) if and only if M equals one of the following sets:
- (a)
PS(q)∖Gp = {α ∈ PS(q) : g(α) < p};
- (b)
, where q ≤ r < p and N is a maximal subsemigroup of Sr.
Proof. Let α, β ∈ PS(q) be such that g(α) < p and g(β) < p. Clearly |Xα∖dom β | ≤|X∖dom β | = g(β) < p. Then
We now suppose that M is a maximal subsemigroup of PS(q) such that M ≠ PS(q)∖Gp. Then there exists α ∉ M with g(α) < p. Thus, Lemma 3.3 implies that Sk∖M ≠ ∅ and Sk ∩ M ≠ ∅ for some k, where q ≤ k < p. Since , Lemma 3.1(b) implies that Sk∩M is maximal in Sk. We also see that
By the previous theorem, when p > q, most of the maximal subsemigroups of PS(q) are induced by maximal subsemigroups of Sr where q ≤ r < p. Hence we now determine some maximal subsemigroups of Sr.
- (a)
A ⊈ Xα, or
- (b)
A⊆Xα and either Aα⊆A⊆dom α, or |Xα∖A | < q.
The next result gives more detail on .
Lemma 3.6. Suppose that p ≥ q ≥ ℵ0, and let A be a nonempty subset of X such that |X∖A | ≥ q. Then,
- (a)
for any cardinal k such that 0 ≤ k ≤ p, there exist α, β ∈ PS(q) such that g(α) = k = g(β) and ;
- (b)
for each , |dom γ∖Aγ−1 | = |X∖A | = |Xγ∖A| and |Aγ−1 | = |A|.
Proof. To show that (a) holds, let |X∖A| = r ≥ q, and let k be a cardinal such that 0 ≤ k ≤ p. We write where |R | = r and |Q | = q. If r = p, then |A ∪ R | ≥ r = p; if not, then |X∖A | < p, and this implies |A | = p, and so |A ∪ R | = p. Fix a ∈ A and let B = (A∖{a}) ∪ R. Then, |B | = p and |X∖B | = |Q ∪ {a}| = q. We write where |K | = k and |L | = p. Then there exists a bijection α : L → B and so g(α) = k, d(α) = q. Also, since A ⊈ B = Xα, we have .
To find with g(β) = k, we consider two cases. First, if r = p, we write where |P| = p, |Q| = q, | K | = k. Fix a ∈ A and define
To see that (b) holds, suppose that there is , then A⊆Xγ and |Xγ∖A | ≥ q. So |Aγ−1 | = |A| since γ is injective. Also,
In [6, Theorem 1], the authors proved that MA is a maximal subsemigroup of BL(q) for every nonempty subset A of X such that |X∖A | ≥ q. Using a similar argument, we show that is a subsemigroup of PS(q).
Lemma 3.7. Suppose that p ≥ q ≥ ℵ0, and let A be a nonempty subset of X such that |X∖A | ≥ q. Then is a proper subsemigroup of PS(q).
Proof. Let . If A ⊈ Xαβ, then . Now we suppose that A⊆Xαβ. Then, A⊆Xβ and since , we either have Aβ⊆A⊆dom β, or |Xβ∖A | < q. If |Xβ∖A | < q, then
Remark 3.8. For any cardinal r such that q ≤ r ≤ p, is a proper subsemigroup of Sr but it is not maximal when q < r. To see this, suppose is maximal and choose such that g(α) = r and g(β) = 0 (possible by Lemma 3.6(a)). Then where dom β = X. Moreover , and so
Since MA is maximal in BL(q), a subsemigroup of PS(q), it is natural to think that is maximal in PS(q). But when p > q, by taking r = p, the above observation shows that this claim is false since Sp = PS(q). Thus, is not always a maximal subsemigroup of PS(q).
The proof of the next result follows some ideas from [6, Theorem 1].
Theorem 3.9. Suppose that p ≥ r ≥ q ≥ ℵ0, and let A be a nonempty subset of X such that |X∖A | ≥ q. Then is a maximal subsemigroup of Sr precisely when r = q.
Proof. In Remark 3.8, we have shown that is not maximal in Sr when r > q. It remains to show is maximal in Sq. Let . Then g(α), g(β) ≤ q and Lemma 3.6(b) implies that
Now define γ ∈ P(X) by
On the other hand, if A ⊈ dom β, then there exists w ∈ A∩(X∖dom β). In this case, we rewrite and where |J | = s, |K | = q. Like before, we write A = {ai} and where {bj} = dom α∖Aα−1, then
We now determine some other classes of maximal subsemigroups of Sr.
Lemma 3.10. Suppose that p ≥ r ≥ q ≥ ℵ0. Let k be a cardinal such that k = 0 or q ≤ k ≤ r. Then
Proof. Since k ≤ r, we have Sr∖Gk⊊Sr. If k = 0, then Sr∖G0 = Sr∖BL(q), and this is a subsemigroup of Sr since, for α, β ∈ Sr∖BL(q), dom (αβ)⊆dom α⊊X, and this implies αβ ∈ Sr∖BL(q). Now suppose q ≤ k ≤ r and let α, β ∈ Sr be such that g(αβ) = k. We claim that g(α) = k or g(β) = k. To see this, assume that g(α) ≠ k. Since
Note that
Remark 3.11. Observe that, if 0 < k < q then Sr∖Gk is not a semigroup for all q ≤ r ≤ p. To see this, let α ∈ BL(q) and β = idXα∖K for some subset K of Xα such that |K | = k (possible since |Xα | = p > k), then α, β ∈ PS(q) since d(β) = d(α) + k = q. Moreover, since g(α) = 0 and g(β) = q ≠ k, we have α, β ∈ Sr∖Gk. But
Theorem 3.12. Suppose that p ≥ r ≥ q ≥ ℵ0. Then the following statements hold:
- (a)
Sr∖G0 is a maximal subsemigroup of Sr;
- (b)
if p > q, then for each cardinal k such that q ≤ k ≤ r, Sr∖Gk is a maximal subsemigroup of Sr.
Proof. By Lemma 3.10, Sr∖G0 is a subsemigroup of Sr. To see that it is maximal, let α, β ∈ G0 = BL(q)⊆Sq. By [3, Theorem 5], Sq = β · R(q), and this implies that α = βγ for some γ ∈ R(q)⊆Sr∖G0. Hence Sr∖G0 is maximal in Sr.
Now suppose that p > q and let q ≤ k ≤ r. Let α, β ∈ Gk. If k = q, then Gk = R(q)⊆Sq and, by [3, Theorem 6], Sq = BL(q) · β · BL(q). If k > q, then Gk = BL(q) · β · BL(q) (by Lemma 3.4). Therefore, α = γβμ for some γ, μ ∈ BL(q)⊆Sr∖Gk, and so Sr∖Gk is maximal in Sr.
Corollary 3.13. Suppose that p > q ≥ ℵ0 and let A be a nonempty subset of X such that |X∖A | ≥ q. Then the following sets are maximal subsemigroups of PS(q):
- (a)
;
- (b)
Nk = {α ∈ PS(q) : g(α) ≠ k} where k = 0 or q ≤ k ≤ p.
Proof. By Theorem 3.9, is maximal in Sq. Then Corollary 3.2 implies that is maximal in PS(q). But
Theorem 3.14. Suppose that p > q ≥ ℵ0 and k equals 0 or q. Let A be a nonempty subset of X such that |X∖A | ≥ q. Then the two classes of maximal subsemigroups and Sq∖Gk of Sq are always disjoint.
4. Maximal Subsemigroups of PS(q) When p = q
Lemma 4.1. Suppose that p = q ≥ ℵ0 and M is a maximal subsemigroup of PS(q). Then the following statements hold:
- (a)
M contains all α ∈ PS(q) with r(α) < p,
- (b)
if R(q)⊆M, then M∩BL(q) = ∅.
Proof. Suppose that there exists α ∉ M with r(α) = k < p. Then g(α) = p, and we write in the usual way
To show that (b) holds, suppose that R(q)⊆M. If there exists α ∈ M∩BL(q), then [3, Theorem 5] implies that PS(q) = α · R(q)⊆M (note that Sq = PS(q) when p = q), so M = PS(q), contrary to the maximality of M. Thus M∩BL(q) = ∅.
Remark 4.2. If p > q, then every α ∈ PS(q) has rank p. This contrasts with Lemma 4.1(a). Also, by Corollary 3.13, if p > q and q < k ≤ p, Nk is a maximal subsemigroup of PS(q) containing R(q) ∪ BL(q), this contrasts with Lemma 4.1(b).
Theorem 4.3. Suppose that p = q ≥ ℵ0, and let A be a nonempty subset of X such that |X∖A | ≥ q. The following are maximal subsemigroups of PS(q):
- (a)
;
- (b)
N0;
- (c)
Np ∪ Jp.
Proof. If p = q, then Sq = PS(q), and so (a) holds by Theorem 3.9. Also, by taking r = p in Theorem 3.12(a), we see that (b) holds. To show that (c) holds, take r = p = k in Lemma 3.10, we have Np = Sp∖Gp is a subsemigroup of PS(q). Moreover, Np ∪ Jp is also a subsemigroup of PS(q) since Jp is an ideal. To show the maximality of Np ∪ Jp, let α, β ∈ PS(q)∖(Np ∪ Jp). Then g(α) = g(β) = p = r(α) = r(β). Write in the usual way
Remark 4.4. When p = q, if M is a maximal subsemigroup containing R(q), then
Remark 4.5. As we showed in Section 3, to see all maximal subsemigroups of PS(q) when p > q, it is necessary to describe all maximal subsemigroups of Sr where q ≤ r < p. So we leave this as a direction for future research.
Acknowledgments
The authors wish to thank the referees for the careful review and the valuable comments, which helped to improve the readability of this paper. B. Singha thanks the Office of the Higher Education Commission, Thailand, for its support by a grant. He also thanks the Graduate School, Chiang Mai University, Chiangmai, Thailand, for its financial support during the preparation of this paper. J. Sanwong thanks the National Research University Project under the Office of the Higher Education Commission, Thailand, for its financial support.