Abstract
For \(n\in \mathbb {N}\), let \(O_{n}\) be the semigroup of all order-preserving transformations on the finite chain \(X_{n}=\{1,\ldots ,n\}\), under its natural order. For any non-empty subset A of \(X_{n}\), let \(O_{n}(A)\) and \(O_{n}^+(A)\) be the subsemigroups of all order-preserving and A-decreasing, and of all order-preserving and A-increasing transformations on \(X_{n}\), respectively. In this paper we obtain formulae for the number of elements and for the number of idempotents in \(O_{n}(A)\). Moreover, we show that \(O_{n}(A)\) contains a zero element if and only if \(1\in A\), and then we obtain the number of nilpotents in \(O_{n}(A)\) when \(1\in A\).
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction
For \(n\in \mathbb {N}\), let \(T_{n}\) be the (full) transformations semigroup (under composition) on the finite chain \(X_{n}=\{1,\ldots ,n\}\), under its natural order. A transformation \(\alpha \in T_{n}\) is called order-preserving if \(x\le y\) implies (\(\forall x,y \in X_{n} \)) \(x\alpha \le y\alpha \), and the semigroup of all order-preserving transformations in \(T_{n}\) is denoted by \(O_{n}\). The problem to find certain combinatorial properties of subsemigroups of \(T_{n}\), in particular of \(O_{n}\), is an important problem in Semigroup theory, and has been much studied in a literature, see for example [1, 2, 5, 8,9,10].
A transformation \(\alpha \in T_{n}\) is called decreasing (increasing) if \(x\alpha \le x\) (\(x\alpha \ge x\)) for each \(x \in X_n\) and the semigroup of all order-preserving and decreasing (increasing) transformations in \(T_n\) is denoted by \(C_n\) (\(C_n^+\)). It is a well-known fact from [12, Corollary 2.7] that \(C_n\) and \(C_n^+\) are isomorphic semigroups. For any non-empty subset A of \(X_{n}\), a transformation \(\alpha \in T_{n}\) is called A-decreasing (A-increasing) if \(x\alpha \le x\) (\(x\alpha \ge x\)) for each \(x \in A\). Then the subsemigroup of all order-preserving and A-decreasing transformations in \(T_{n}\) is denoted by \(O_{n}(A)\), and the subsemigroup of all order-preserving and A-increasing transformations in \(T_{n}\) is denoted by \(O_{n}^+(A)\), say
respectively. Notice that
-
\(O_{n}(A)=O_{n}\) if \(A=\{n\}\),
-
\(O_{n}(A)=C_{n}\) if \(A=X_n{\setminus } \{n\}\) or \(A=X_n\), and
-
\(O_{n}(A)= O_{n}(A\cup \{n\})\) for any non-empty subset A of \(X_n\).
Similarly, we have
-
\(O_{n}^+(A)=O_{n}\) if \(A=\{1\}\),
-
\(O_{n}^+(A)=C_{n}^+\) if \(A=X_n{\setminus } \{1\}\) or \(A=X_n\), and
-
\(O_{n}^+(A)= O_{n}^+(A\cup \{1\})\) for each non-empty subset A of \(X_n\).
Also notice that \(C_{n}\subseteq O_{n}(A)\subseteq O_{n}\) and \(C_{n}^+\subseteq O_{n}^+(A)\subseteq O_{n}\).
Let S be a semigroup. An element \(x\in S\) is called idempotent if \(x^2=x\), and the set of all idempotents in S is denoted by E(S). Moreover, if S contains a zero element, denoted in general by 0, that implies \(0y=y0=0\) for each \(y\in S\), then an element \(x\in S\) is called nilpotent if \(x^k=0\) for some positive integer k, and the set of all nilpotents in S is denoted by N(S).
Recall that nth Fibonacci number \(f_n\) is defined by the linear recurrence relation
where \(f_{1}=f_{2}=1\) (see for example [6]), and nth Catalan number \(\mathcal {C}_{n}\) is defined by
(see for example [8]). From [6, Theorem 2.1 and Theorem 2.3] we know that
Also, from [8, Theorem 2.1 and Proposition 2.3] and from [5, Theorem 3.19] we know that
In general a transformation \(\alpha \in T_{n}\) is represented by the following tabular form:
The height, the kernel and the fix of any transformation \(\alpha \in T_n\) are defined by
respectively. Recall that \(\ker (\alpha )\) is an equivalence relation on \(X_{n}\) and the equivalence classes obtained by \(\ker (\alpha )\) are all of the pre-image sets of elements in \(\mathrm {im\, }(\alpha )\), say \(\{ y\alpha ^{-1}: y\in \mathrm {im\, }(\alpha )\}\), that forms a partition of \(X_n\). (For the other terms in semigroup theory, which are not explained here, we refer to [7]).
Let \(P=\{ I_{1}, \ldots ,I_{p} \}\) be a partition of a set \(X_n\) for \(1\le p \le n\). P is called an ordered partition, and we write \(P=\{I_{1}< \cdots <I_{p}\}\), if \(x < y\) for all \(x\in I_{i}\) and \(y\in I_{i+1}\) (\(1\le i\le p-1\)), (the idea of ordering a family of sets appeared on p. 335 of [11]). Notice that the kernel classes of \(\alpha \in O_{n}\) are convex subsets C of \(X_{n}\) in the sense that
(see [4, p. 187]). Thus, if \(\mathrm {h\, }(\alpha )=p\) (\(2\le p\le n\)), there exist \(x_1,\ldots , x_{p-1}\in X_n\) such that the kernel classes of \(\alpha \) are \(A_i=\{x_{i-1}+1,\ldots ,x_i\}\) for \(1\le i\le p\) where \(x_0=0\) and \(x_p=n\), then the ordered partition \(\{A_1<\cdots < A_k\}\) of \(X_n\) is called the kernel partition of \(\alpha \). Moreover, there exist \(a_1,\ldots ,a_p\in X_n\) such that \(\mathrm {im\, }(\alpha )=\{a_1<\cdots < a_p\}\). Then we can also use the following tabular form:
to represent \(\alpha \in O_{n}\).
Recall that, for \(k\in \mathbb {N^+}\) and \(m\in \mathbb {N}\), the number of integer solutions of the equation
is \(\left( {\begin{array}{c}m+k-1\\ k-1\end{array}}\right) =\left( {\begin{array}{c}m+k-1\\ m\end{array}}\right) \), and throughout the next section we use the notation \(x_i=| i\alpha ^{-1}|\) for \(1 \le i \le n\). Also we use the equation
that can be shown easily by inductive hypothesis on \(t\ge 1\). Moreover, if \(A=\{r\}\) for any \(1\le r \le n\), we use the notation \(O_n(r)\)\((O_n^+(r))\) instead of \(O_n(\{r\})\)\((O_n^+(\{r\}))\).
The rank and the idempotent rank of a finite semigroup S are defined by \(\mathrm {rank\, }(S)=\min \{ |A| :A \subseteq S, \langle A \rangle = S \}\) and \(\mathrm {idrank\, }(S)=\min \{ |A|:A\subseteq E(S), \langle A \rangle =S \}\), respectively. Zhao, investigated the subsemigroups \(O_{n}(A)\) for any non-empty subset A of \(X_n\) in [13], and examined in terms of certain algebraic properties for example, the Green’s \(*\)-Relations and the (idempotent) rank of \(O_{n}(A)\). Now we examine the combinatorial properties of \(O_{n}(A)\), and obtain formulaes for the numbers of elements, idempotents and (when \(1\in A\)) nilpotents in \(O_{n}(A)\). Moreover, we show that the subsemigroup \(O_{n}^+(A)\) is isomorphic to the subsemigroup \(O_{n}(\overline{A})\) where \(\overline{A}=\{n-a+1:a\in A\}\), and state some properties of \(O_{n}^+(A)\) by using this isomorphism.
2 The Number of Elements in \(O_n(A)\)
Lemma 2.1
For \(1\le r \le n\), we have \(|O_{n}(r) |=\sum _{s=1}^{r}\left( {\begin{array}{c}r+s-2\\ s-1\end{array}}\right) \left( {\begin{array}{c}2n-(r+s)\\ n-s\end{array}}\right) \).
Proof
Let \(1\le r \le n-1\), and let \(\alpha \in O_n(r)\). Then there exists \(s\in X_n\) such that \(1 \le s \le r\) and that \(r\alpha =s\). That is \(\alpha \in O_n\) with the properties that
-
(i)
\(1\le t\alpha \le s\) for each \(1 \le t \le r\),
-
(ii)
\(r\alpha =s\) and
-
(iii)
\(s\le t\alpha \le n\) for each \(r\le t \le n\).
Consider and fix an arbitrary s such that \(1\le s \le r\). It is easy to see that the number of elements with the properties given above in \(O_n\) is equals to the product of the numbers of integer solutions of the equations
by considering the possible image sets. This product is
and so the result is clear for \(1\le r \le n-1\) by considering each \(1\le s\le r\) separately. Moreover, \(|O_n(n)|=\left( {\begin{array}{c}2n-1\\ n-1\end{array}}\right) \) since \(O_n(n)=O_n\), and notice that
by using the well-known basic recurrence that \(\left( {\begin{array}{c}m-1\\ k-1\end{array}}\right) +\left( {\begin{array}{c}m-1\\ k\end{array}}\right) =\left( {\begin{array}{c}m\\ k\end{array}}\right) ,\) for \(m,k\in \mathbb {Z^+},\) as required. \(\square \)
Theorem 2.2
For any subset \(A=\{r_1,\ldots ,r_k\}\)\((k\ge 2)\) of \(X_n\), we have
Proof
Let \(A=\{r_1,\ldots ,r_k\}\subseteq X_n\)\((k\ge 2)\), and let \(\alpha \in O_n(A)\). Then there exist \(s_1,\ldots , s_k\in X_n\) such that \(s_{i-1}\le s_i \le r_i\) and \(r_i\alpha =s_i\) for each \(1 \le i \le k\) where \(s_0=1\). That is \(\alpha \in O_n\) with the properties that
-
(i)
\(r_i\alpha =s_i\), for each \(1\le i \le k\)
-
(ii)
\(1\le t\alpha \le s_1\) for each \(1 \le t \le r_1\),
-
(iii)
\(s_{i-1}\le t\alpha \le s_i\) for each \(r_{i-1}+1 \le t \le r_i\)\((2 \le i \le k)\), and
-
(iv)
\(s_k\le t\alpha \le n\) for \(r_k \le t \le n\).
Consider and fix an ordered k-tuple \((s_1,\ldots ,s_k )\) such that \(s_{i-1}\le s_i \le r_i\) for each \(1\le i \le k\) where \(s_0=1\). It is easy to see that the number of elements with the properties given above in \(O_n\) equals to the product of the numbers of integer solutions of the equations
by considering the possible image sets. This product is
and so the result is clear by considering each ordered k-tuple \((s_1,\ldots ,s_k )\) such that \(1\le s_1\le \cdots \le s_{k}\le r_k\). \(\square \)
3 The Number of Idempotents in \(O_n(A)\)
It is shown in [13, Lemma 2.10] that \(\alpha \in O_{n}(A)\) is idempotent if and only if for each \(t\in \mathrm {im\, }(\alpha )\), \(t\alpha =t\) and \(t=\min \{x:x\in t\alpha ^{-1}\cap (A\cup \{t\})\}\). Let \(\alpha \in T_n\) and let \(\emptyset \ne X \subseteq X_n\). Then the restriction \(\alpha \) to the set X is denoted by \(\alpha |_{X}\).
Lemma 3.1
For \(1\le r \le n\), we have \(|E(O_{n}(r))|=f_{2r}\,f_{2(n-r)+1}\).
Proof
Let \(1\le r \le n-1\). Since \(r \alpha \le r\) for each \(\alpha \in E(O_{n}(r))\), we have
and so \(|I|=f_{2r}\).
Now consider and fix an arbitrary element \(\sigma \in I\), and let \(r\sigma =s\) for any \(1 \le s \le r\). Then let
Notice that \(I_\sigma =\bigcup _{i=r}^{n}I_{\sigma }^i\) where
for each \(r \le i \le n-1\), and
Also notice that for each \(\alpha |_{\{r+1,\ldots , n\}}\in I_{\sigma }^i\) we have
since \(\alpha \in E(O_n(r))\), and so
for each \(r\le i\le n-1\). Thus we have
for \(r\le i \le n-1\), and so
It is easy to see that \(|I_{\sigma }|=f_{2(n-r)+1}\) for each \(\sigma \in I\), and so we have
for \(1\le r \le n-1\).
Moreover, \(|E(O_n(n))|=f_{2n}\) since \(O_n(n)=O_n\), and notice that \(|E(O_{n}(n))|= f_{2n} = f_{2n} \, f_{2(n-n)+1}\), as required. \(\square \)
Theorem 3.2
For any subset \(A=\{r_1,\ldots ,r_k\}\)\((k\ge 2)\) of \(X_n\), we have
where \(r_{k+1}=n\).
Proof
Let \(A=\{r_1,\ldots ,r_k\}\subseteq X_n\)\((k\ge 2)\). Since \(r_1\alpha \le r_1\), for each \(\alpha \in E(O_{n}(A))\), we have
and so \(|I|=f_{2r_1}\).
Now consider and fix an arbitrary element \(\sigma _1\in I\) and let \(r_1\sigma _1=s_1\) for any \(1 \le s_1 \le r_1\). Then let
As in the proof of Lemma 3.1 it is clear that \(I_{\sigma _1}=\bigcup _{i=r_{1}}^{r_{2}}I_{\sigma _1}^{i}\) where
for \(r_{1}\le i\le r_{2}-1\) and
and that
for \(r_1\le i \le r_2-1\). Thus
Also it is easy to see that \(|I_{\sigma }|=f_{2(r_2-r_1)+1}\) for each \(\sigma \in I\).
With similar steps, for each \(2\le j \le k\) consider and fix an arbitrary element \(\sigma _j\in I_{\sigma _{j-1}}\) and let \(r_j\sigma _j=s_j\) for any \(s_{j-1}\le s_j \le r_j\). Then let
where \(r_{k+1}=n\). Similarly, we have \(I_{\sigma _j}=\bigcup _{i=r_{j}}^{r_{j+1}}I_{\sigma _{j}}^{i}\) where
for \(r_{j}\le i\le r_{j+1}-1\) and
and
for each \(r_j\le i \le r_{j+1}-1\). Thus we have
Also it is easy to see that \(|I_{\sigma }|=f_{2(r_{j+1}-r_{j})+1}\) for each \(\sigma \in I_{\sigma _{j-1}}\) (\(2\le j \le k\)). Thus we have
where \(r_{k+1}=n\), as required. \(\square \)
4 The Number of Nilpotents in \(O_n(A)\)
It is well known that \(\varepsilon =\left( \begin{array}{ccc} 1 &{} \cdots &{} n\\ 1 &{} \cdots &{} 1 \end{array}\right) \) and \(\varepsilon ^+=\left( \begin{array}{ccc} 1 &{} \cdots &{} n\\ n &{} \cdots &{} n \end{array}\right) \) are the zero elements of \(C_{n}\) and \(C_n^+\), respectively. Also, from [8, Lemma 2.2], we know that an element \(\alpha \) in \(C_{n}\)\((C_n^+)\) is nilpotent if and only if \(\mathrm {Fix\, }(\alpha )=\{1\}\)\((\mathrm {Fix\, }(\alpha )=\{n\})\).
Theorem 4.1
For any non-empty subset A of \(X_{n}\)\((n\ge 2)\), \(O_{n}(A)\) contains a zero element if and only if \(1\in A\).
Proof
(\(\Leftarrow \)) Let A be a subset of \(X_{n}\) contains 1. Then clearly
is a zero element in \(O_{n}(A)\), as required.
(\(\Rightarrow \)) Suppose that \(O_{n}(A)\) contains a zero element denoted by \(\xi \). Since \(O_{n}(A)\) contains at least one constant transformation, and since \(\alpha \xi =\xi \alpha =\xi \) for each \(\alpha \in O_{n}(A)\), there exists \(1\le i\le n\) such that \(\xi =\left( \begin{array}{ccc} 1 &{} \cdots &{} n\\ i &{} \cdots &{} i \end{array}\right) .\) Now assume that \(i\ne 1\). Then it is clear that there exist at least one element \(\alpha \in O_{n}(A)\) such that \(i\alpha \ne i\), and so we have
which is a contradiction. Hence \(i=1\), and so \(\xi =\varepsilon \).
Next assume that \(1\notin A\). Then there exist at least one element \(\beta \in O_{n}(A)\) such that \(1\beta \ge 2\), and so we have \(\xi \beta \ne \xi \) which is a contradiction. Thereby \(1\in A\), as required. \(\square \)
Lemma 4.2
For any subset A of \(X_n\) contains 1, an element \(\alpha \in O_{n}(A)\) is nilpotent if and only if \(x\alpha \ne x\) for each \(2\le x\le n\).
Proof
We prove this claim as in the proof of [3, Lemma 2.1].
\((\Rightarrow )\) Let \(\alpha \in N(O_{n}(A))\). If there exists \(2\le x\le n\) such that \(x\alpha =x\), it is clear that
and so \(\alpha ^k\ne \varepsilon \) for all \(k\in \mathbb {Z^+}\), which is a contradiction.
\((\Leftarrow )\) Let \(\alpha \in O_{n}(A)\), and let \(x\alpha \ne x\) for each \(2\le x\le n\). Then we have
and so for each \(k\ge 2\)
Consequently we have \(x\alpha ^k\ne x\) for each \(2\le x\le n\) and for each \(k\in \mathbb {Z}^+\).
If \(\mathrm {im\, }(\alpha )=\{1\}\) then we have \(\alpha =\varepsilon \), and so the result is clear. Now let \(\{1\}\subset \mathrm {im\, }(\alpha )\), and let
where \(\{A_1<\cdots <A_r\}\) is the kernel partition of \(\alpha \) and \(\mathrm {im\, }(\alpha )=\{1<a_2<\cdots <a_r\}\subseteq X_n\) for any \(2\le r\le n\). It is clear that \(a_r\notin A_r\) since \(a_r\alpha \ne a_r\), and so \(a_r < \min \{A_r\}\). Thus we have \(\mathrm {im\, }(\alpha )\cap A_r=\emptyset \) and so \(a_r\notin \mathrm {im\, }(\alpha ^2)\), that is
If \(\mathrm {im\, }(\alpha ^2)=\{1\}\) then clearly \(\alpha ^2=\varepsilon \), and so the result is clear. Now let \(\{1\}\ne \mathrm {im\, }(\alpha ^2)\subset \mathrm {im\, }(\alpha )\), and let
where \(\{B_1<\cdots <B_t\}\) is the kernel partition of \(\alpha ^2\) and \(\mathrm {im\, }(\alpha ^2)=\{1< b_2<\cdots<b_t\}\subset \{1<a_2<\cdots <a_r\}\) for \(2\le t< r\le n\). Similarly we have
since \(x\alpha ^2\ne x\) for each \(2\le x\le n\). If we continue with this procedure, we will obtain a strict descent sequence:
Since \(|\mathrm {im\, }(\alpha )|\) is finite, there exists \(m>2\) such that
as required. \(\square \)
Lemma 4.3
For any subset A of \(X_n\) contains 1, \(N(O_n(A))=N(C_n)\).
Proof
Let A be a subset of \(X_{n}\) contains 1. First recall that \(C_n \subseteq O_n(A)\), and that the zero element \(\varepsilon \) of \(C_n\) is also the zero element of \(O_n(A)\). Hence, we immediately have \(N(C_n) \subseteq N(O_n(A))\).
Conversely, let \(\alpha \in N(O_n(A))\). Then clearly \(1\alpha =1\). If \(2\in A\) we immediately have \(2\alpha =1\) from Lemma 4.2. Now let \(2\notin A\) and assume that \(2\alpha >2\). Then, by using Lemma 4.2 in each step, we have
and so \(n\alpha =n\), which is in a contradict with Lemma 4.2. Thus, even \(2\notin A\) we also have \(2\alpha =1\).
If \(3\in A \), then we immediately have \(3\alpha =1\) or \(3\alpha =2\) from Lemma 4.2. Now let \(3\notin A\). Similarly, it can be shown that to be \(3\alpha >3\) implies that \(n\alpha =n\), which is in a contradict with Lemma 4.2. Thus even \(3\notin A\) we also have \(3\alpha =1\) or \(3\alpha =2\).
If we continue with this procedure, we will obtain that \(1\le x\alpha <x\) for each \(1\le x\le n\), and so \(\alpha \in C_n\). Moreover, from Lemma 4.2 and [8, Lemma 2.2], we have \(\alpha \in N(C_n)\), as required. \(\square \)
Corollary 4.4
For any subset A of \(X_n\) contains 1, we have
\(\square \)
5 Remark
First we state the following theorem which can be proved easily and will be used throughout this section.
Theorem 5.1
For any non-empty subset A of \(X_{n}\), let \(\overline{A}=\{n-a+1:a\in A\}\). Then the map
is an isomorphism where \(x\alpha ^*=n-(n-x+1)\alpha +1\) for each \(x\in X_n\). \(\square \)
This isomorphism proves to be powerful tool in translating results on \(O_{n}(\overline{A})\) to \(O_{n}^+(A)\). For example, we can immediately translate all results, stated in [13], to \(O_{n}^+(A)\), more precisely, for \(n\ge 3\) we have
Moreover, we have
-
\(\alpha \in E(O_{n}^+(A))\) if and only if \(\alpha ^*\in E(O_{n}(\overline{A}))\),
-
\(O_{n}^+(A)\) has a zero element if and only if \(n\in A\), and
-
\(\alpha \in N(O_{n}^+(A))\) if and only if \(\alpha ^*\in N(O_{n}(\overline{A}))\) when \(n\in A\).
Thereby we can state certain combinatorial results on \(O_{n}^+(A)\), for any non-empty subset A of \(X_{n}\), as in the following theorem.
Theorem 5.2
For any non-empty subset A of \(X_{n}\), we have
-
(i)
\(|O_{n}^+(A)|=|O_{n}(\overline{A})|\),
-
(ii)
\(|E(O_{n}^+(A))|=|E(O_{n}(\overline{A}))|\), and
-
(iii)
\(|N(O_{n}^+(A))|=|N(O_{n}(\overline{A}))|=|N(C_n)|=\mathcal {C}_{n-1}\) when \(n\in A\). \(\square \)
References
Ayık, G., Ayık, H., Koç, M.: Combinatorial results for order-preserving and order-decreasing transformations. Turk. J. Math. 35, 1–9 (2011)
Fernandes, V.H., Gomes, G.M.S., Jesus, M.M.: The cardinal and the idempotent number of various monoids of transformations on a finite chain. Bull. Malays. Math. Sci. Soc. 34, 79–85 (2011)
Garba, G.U.: Nilpotents in semigroups of partial order-preserving transformations. Proc. Edinb. Math. Soc. 37, 361–377 (1994)
Garba, G.U.: On the idempotent ranks of certain semigroups of order-preserving transformations. Port. Math. 51, 185–204 (1994)
Higgins, P.M.: Combinatorial results for semigroups of order-preserving mappings. Math. Proc. Camb. Philos. Soc. 113, 281–296 (1993)
Howie, J.M.: Products of idempotents in certain semigroups of transformations. Proc. Edinb. Math. Soc. 17, 223–236 (1970/71)
Howie, J.M.: Fundamentals of Semigroup Theory. Oxford University Press, New York (1995)
Laradji, A., Umar, A.: Combinatorial results for semigroups of order-preserving full transformations. Semigroup Forum 72, 51–62 (2006)
Laradji, A., Umar, A.: On the number of nilpotents in the partial symmetric semigroup. Commun. Algebra 32, 3017–3023 (2004)
Laradji, A., Umar, A.: On certain finite semigroups of order-decreasing transformations I. Semigroup Forum 69, 184–200 (2004)
Sornsanam, S., Sullivan, R.P.: Regularity conditions on order-preserving transformation semigroups. Southeast Asian Bull. Math. 28, 333–342 (2004)
Umar, A.: Semigroups of order-decreasing transformations: the isomorphism theorem. Semigroup Forum 53, 220–224 (1996)
Zhao, P.: On the semigroups of order-preserving and \(A\)-decreasing finite transformations. Algebra Colloq. 21, 653–662 (2014)
Author information
Authors and Affiliations
Corresponding author
Additional information
Communicated by Kar Ping Shum.
Rights and permissions
About this article
Cite this article
Bugay, L., Yağcı, M. & Ayık, H. Combinatorial Results for Semigroups of Order-Preserving and A-Decreasing Finite Transformations. Bull. Malays. Math. Sci. Soc. 42, 921–932 (2019). https://doi.org/10.1007/s40840-017-0529-1
Received:
Revised:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s40840-017-0529-1