betaFeedback

WorksMathematics

Automorphism of reduced finitary power monoids

Abstract

Let HH be a multiplicatively written monoid with identity element 1H1_H. The family Pfin,1(H)\mathcal{P}_{\mathrm{fin},1}(H) of all finite subsets of HH containing 1H1_H is a monoid with identity {1H}\{1_H\} under the setwise multiplication induced by HH. We prove that Aut⁡(Pfin,1(H))≅Aut⁡(H)\operatorname{Aut}(\mathcal{P}_{\mathrm{fin},1}(H)) \cong \operatorname{Aut}(H) whenever HH satisfies either of the following two conditions: (1) HH is commutative, cancellative, and contains an invertible element of infinite order; (2) HH is an abelian group which is not isomorphic to the Klein four-group.

Introduction

Let HH be a semigroup. Endowed with the setwise multiplication induced by HH, the family of all nonempty subsets of HH forms a semigroup, hereinafter denoted by P(H)\mathcal{P}(H). The family Pfin(H)\mathcal{P}_{\mathrm{fin}}(H) of all nonempty finite subsets of HH is a subsemigroup of P(H)\mathcal{P}(H). More explicitly, the operation is defined by

XY:={xy:x∈X,y∈Y},XY := \{xy : x \in X, y \in Y\},

for all X,Y∈P(H)X, Y \in\mathcal{P}(H). We call P(H)\mathcal{P}(H) and Pfin(H)\mathcal{P}_{\mathrm{fin}}(H) the large power semigroup and the finitary power semigroup of HH, respectively.

In particular, if HH is a monoid with identity 1H1_H then P(H)\mathcal{P}(H) and Pfin(H)\mathcal{P}_{\mathrm{fin}}(H) are monoids too; their identity is the singleton {1H}\{1_H\}. Moreover,

P1(H):={X⊆H:1H∈X}andPfin,1(H):=P1(H)∩Pfin(H)\mathcal{P}_1(H) := \{X \subseteq H : 1_H \in X\} \quad\text{and}\quad \mathcal{P}_{\mathrm{fin},1}(H) := \mathcal{P}_1(H) \cap\mathcal{P}_{\mathrm{fin}}(H)

are submonoids of P(H)\mathcal{P}(H) and Pfin(H)\mathcal{P}_{\mathrm{fin}}(H), respectively. We call P1(H)\mathcal{P}_1(H) the reduced large power monoid and Pfin,1(H)\mathcal{P}_{\mathrm{fin},1}(H) the reduced finitary power monoid of HH.

Remark 1.1. When HH is an additively written semigroup, the same convention is adopted for P(H)\mathcal{P}(H) and its subsemigroups. The operation on P(H)\mathcal{P}(H) is then called setwise addition, and it is explicitly defined by

X+Y:={x+y:x∈X,y∈Y},X + Y := \{x + y : x \in X, y \in Y\},

for all X,Y∈P(H)X, Y \in\mathcal{P}(H). Accordingly, if HH is a monoid, we use P0(H)\mathcal{P}_0(H) in place of P1(H)\mathcal{P}_1(H) and Pfin,0(H)\mathcal{P}_{\mathrm{fin},0}(H) in place of Pfin,1(H)\mathcal{P}_{\mathrm{fin},1}(H).

Power semigroups first appeared explicitly in a 1950 paper of Ballieu [2], and their systematic study was started by Tamura and Shafer [15, 16] in the late 1960s. In recent years, they have been studied from several perspectives, including their algebraic structure, their factorization properties, and their connections with arithmetic combinatorics; see [1, 7, 8]. For more details about the arithmetic of power monoids, we refer the reader to the survey [17].

Let N\mathbb{N} and Z\mathbb{Z} be the additive monoid of nonnegative integers and the additive group of integers, respectively. Denote by Aut⁡(H)\operatorname{Aut}(H) the automorphism group of a semigroup HH. The study of the automorphisms of power semigroups can be traced back to Byrd et al. In particular, it is proved in [5 Theorem 4] that Aut⁡(Pfin(H))\operatorname{Aut}(\mathcal{P}_{\mathrm{fin}}(H)) is isomorphic to Aut⁡(H)\operatorname{Aut}(H) for a locally finite group HH with ∣H∣>5|H| > 5, and in [6 Theorem 9] that Aut⁡(Pfin(Z))\operatorname{Aut}(\mathcal{P}_{\mathrm{fin}}(\mathbb{Z})) is a splitting extension of Z\mathbb{Z} by the Klein four-group. The latter result does not completely determine the structure of Aut⁡(Pfin(Z))\operatorname{Aut}(\mathcal{P}_{\mathrm{fin}}(\mathbb{Z})). This was recently achieved by Tringali and Wen [18 Theorem 3.5], who proved that Aut⁡(Pfin(Z))\operatorname{Aut}(\mathcal{P}_{\mathrm{fin}}(\mathbb{Z})) is isomorphic to the direct product of a cyclic group of order two and the infinite dihedral group. Wong et al. [24] showed that for a numerical semigroup HH (that is, a cofinite subsemigroup of the nonnegative integers under addition), Aut⁡(Pfin(H))\operatorname{Aut}(\mathcal{P}_{\mathrm{fin}}(H)) is a cyclic group of order 22 if HH is a discrete interval; otherwise Aut⁡(Pfin(H))\operatorname{Aut}(\mathcal{P}_{\mathrm{fin}}(H)) is trivial. Tringali and Wen [19] proved that Aut⁡(P(H))\operatorname{Aut}(\mathcal{P}(H)) is trivial for every numerical semigroup HH, and that Aut⁡(P0(H))\operatorname{Aut}(\mathcal{P}_{0}(H)) is trivial whenever 0∈H0 \in H.

In this paper, we focus on the relation between the automorphism group of a monoid HH and the automorphism group of the reduced finitary power monoid Pfin,1(H)\mathcal{P}_{\mathrm{fin},1}(H). Each f∈Aut⁡(H)f \in\operatorname{Aut}(H) induces an automorphism f∗f^{*} of Pfin,1(H)\mathcal{P}_{\mathrm{fin},1}(H) defined as follows:

f∗(X)=f[X]:={f(x):x∈X},for all X∈Pfin,1(H).f^{*}(X)=f[X]:=\{f(x):x\in X\},\qquad\text{for all }X\in\mathcal{P}_{\mathrm{fin},1}(H).

We call f∗f^{*} the augmentation of ff and we say that an automorphism of Pfin,1(H)\mathcal{P}_{\mathrm{fin},1}(H) is inner if it is the augmentation of an automorphism of HH.

It is easy to verify that this gives a group monomorphism

Φ:Aut⁡(H)⟶Aut⁡(Pfin,1(H)),f⟼f∗.\Phi:\operatorname{Aut}(H)\longrightarrow\operatorname{Aut}(\mathcal{P}_{\mathrm{fin},1}(H)),\qquad f\longmapsto f^{*}.

It is natural to ask whether Φ\Phi is an isomorphism. In other words, is each automorphism of Pfin,1(H)\mathcal{P}_{\mathrm{fin},1}(H) inner? Tringali and Yan [21] showed that the only non-trivial automorphism of Pfin,0(N)\mathcal{P}_{\mathrm{fin},0}(\mathbb{N}) is given by X⟼max⁡X−XX\longmapsto\max X-X. This proves that, in general, Φ\Phi is not surjective since Aut⁡(N)\operatorname{Aut}(\mathbb{N}) is trivial. In fact, the same result was also obtained in [6 Theorem 1] but this paper went unnoticed until April 2026. More recently, Bhowmik and Tringali [4 Theorem 1.2] showed that Aut⁡(Pfin,0(H))\operatorname{Aut}(\mathcal{P}_{\mathrm{fin},0}(H)) is trivial for every numerical monoid H≠NH\ne\mathbb{N}, which confirmed a conjecture proposed in [21].

Wong et al. [23 Theorem 1.3] proved that all automorphisms of Pfin,0(Z)\mathcal{P}_{\mathrm{fin},0}(\mathbb{Z}) are inner and hence gave an affirmative answer to a conjecture proposed by Tringali and Wen [18]. In a remarkable feat, Rago [12 Theorem 13] established that if HH is a finite abelian group, then all automorphisms of Pfin,1(H)\mathcal{P}_{\mathrm{fin},1}(H) are inner, except when HH is the Klein four-group, in which case Aut⁡(Pfin,1(H))\operatorname{Aut}(\mathcal{P}_{\mathrm{fin},1}(H)) is isomorphic to the direct product of two copies of the symmetric group of degree three.

In this paper, we show that all automorphisms of Pfin,1(H)\mathcal{P}_{\mathrm{fin},1}(H) are inner if HH is a commutative and cancellative monoid containing an invertible element of infinite order or HH is an infinite abelian group.

Notation and preliminary results

We use the conventions of [10] for semigroups and monoids. Let HH be a monoid written multiplicatively. Then an element x∈Hx \in H is right (resp. left) cancellative if yx=zx⇒y=zyx = zx \Rightarrow y = z (resp. xy=xz⇒y=zxy = xz \Rightarrow y = z) for any y,z∈Hy, z \in H. We call HH cancellative if each of its elements is both right and left cancellative. Let XX be a nonempty subset of HH and let a∈Ha \in H. We write

aX:={ax:x∈X}.aX := \{ax : x \in X\}.

We denote by CnC_n the cyclic group of order nn. The cardinality of a set XX is denoted by ∣X∣\lvert X\rvert.

We start with a result of Tringali and Yan [22].

Lemma 2.1. [22 Corollary 3.3] Let HH and KK be monoids, and let F:Pfin,1(H)→Pfin,1(K)F : \mathcal{P}_{\mathrm{fin},1}(H) \to\mathcal{P}_{\mathrm{fin},1}(K) be an isomorphism. Then there is a unique bijection f:H→Kf : H \to K, called the pullback of FF, such that

F({1H,x})={1K,f(x)},for all x∈H.F(\{1_H, x\}) = \{1_K, f(x)\}, \qquad\text{for all } x \in H.

In particular, f(1H)=1Kf(1_H) = 1_K.

The pullback ff of the isomorphism FF has many interesting properties and is remarkably helpful in the study of reduced finitary power monoids. In particular, we will need the following result by Tringali and Yan [22].

Theorem 2.2. [22 Theorem 5.1] Let HH and KK be cancellative monoids, and suppose at least one of them is torsion. Let FF be an isomorphism from Pfin,1(H)\mathcal{P}_{\mathrm{fin},1}(H) to Pfin,1(K)\mathcal{P}_{\mathrm{fin},1}(K). Then the pullback of FF is an isomorphism from HH to KK, and both HH and KK are groups.

Rago [12, 14] continued the study of the properties of the pullback and proved the following results.

Theorem 2.3. [12 Theorem 13] Let HH be a finite abelian group with H≇C2×C2H \not\cong C_2 \times C_2. Then all automorphisms of Pfin,1(H)\mathcal{P}_{\mathrm{fin},1}(H) are inner.

Theorem 2.4. [14 Theorem 14] Let HH and KK be commutative and cancellative monoids, each containing an element of infinite order. Suppose that at least one of them has a nontrivial unit group. Then the pullback of every isomorphism F:Pfin,1(H)→Pfin,1(K)F : \mathcal{P}_{\mathrm{fin},1}(H) \to\mathcal{P}_{\mathrm{fin},1}(K) is an isomorphism.

Let HH be a finite abelian group with H≇C2×C2H \not\cong C_2 \times C_2. Indeed, Rago [12] showed that every F∈Aut⁡(Pfin,1(H))F \in\operatorname{Aut}(\mathcal{P}_{\mathrm{fin},1}(H)) whose pullback is the identity map is itself the identity map, and Theorem 2.3 comes from this fact and Theorem 2.2. We summarize it as follows.

Lemma 2.5. Let HH be a monoid. Assume that the pullback of every automorphism of Pfin,1(H)\mathcal{P}_{\mathrm{fin},1}(H) is an automorphism of HH. Then the following are equivalent:

(1) Φ:Aut⁡(H)⟶Aut⁡(Pfin,1(H))\Phi: \operatorname{Aut}(H) \longrightarrow\operatorname{Aut}(\mathcal{P}_{\mathrm{fin},1}(H)), f⟼f∗f \longmapsto f^*, is an isomorphism;

(2) every F∈Aut⁡(Pfin,1(H))F \in\operatorname{Aut}(\mathcal{P}_{\mathrm{fin},1}(H)) whose pullback is the identity map is itself the identity map;

(3) if the pullbacks of F1,F2∈Aut⁡(Pfin,1(H))F_1,F_2 \in\operatorname{Aut}(\mathcal{P}_{\mathrm{fin},1}(H)) coincide, then F1=F2F_1 = F_2;

(4) if ff is the pullback of an automorphism FF of Pfin,1(H)\mathcal{P}_{\mathrm{fin},1}(H), then F(X)={f(x):x∈X}F(X) = \{f(x) : x \in X\} for any X∈Pfin,1(H)X \in\mathcal{P}_{\mathrm{fin},1}(H).

Proof. (1) ⟹\Longrightarrow (2) is obvious.

(2) ⟹\Longrightarrow (3). Let ff be pullback of F1F_1. By hypothesis, ff is also the pullback of F2F_2. On the other hand, it is clear that f−1f^{-1} is the pullback of F1−1F_1^{-1}, and hence f−1∘ff^{-1} \circ f is the pullback of F1−1∘F2F_1^{-1} \circ F_2. So F1−1∘F2F_1^{-1} \circ F_2 is the identity map and F1=F2F_1 = F_2.

(3) ⟹\Longrightarrow (4). By assumption, ff is also an automorphism of HH. So FF and f∗f^* share the same pullback ff. Hence, F(X)=f∗(X)={f(x):x∈X}F(X) = f^*(X) = \{f(x) : x \in X\} for any X∈Pfin,1(H)X \in\mathcal{P}_{\mathrm{fin},1}(H).

(4) ⟹\Longrightarrow (1) is clear. □

We also need the following results about the reduced finitary power monoid of a cancellative monoid containing an element of infinite order. The results were first proved by Rago in the commutative setting, see the proof of [14 Lemma 4] and then generalized to the non-commutative setting by Bhowmik and Tringali [4 Lemma 2.2].

Lemma 2.6. Let HH be a cancellative monoid. Suppose a∈Ha \in H is invertible of infinite order and XX is a finite nonempty subset of HH. Then there exists a positive integer NN such that

ainX∩ajnX=∅a^{in}X \cap a^{jn}X = \varnothing

for all n>Nn > N and i,j∈Zi,j \in\mathbb{Z} with i≠ji \ne j.

Proof. Lemma 2.2 in [4] gives that there exists a positive integer NN such that ainX∩ajnX=∅a^{in}X \cap a^{jn}X = \varnothing for all integers n>Nn > N and i,j∈Ni,j \in\mathbb{N} with i≠ji \ne j. Therefore, taking into account that aa is a unit of HH, we get that a(i+k)nXa^{(i+k)n}X and a(j+k)nXa^{(j+k)n}X are also disjoint for all i,j∈Ni,j \in\mathbb{N} with i≠ji \ne j, n>Nn > N and k∈Zk \in\mathbb{Z}, which is the desired conclusion. □

Lemma 2.7. [4 Proposition 2.4] Let HH and KK be cancellative monoids and assume that HH contains an element of infinite order. Then each isomorphism F:Pfin,1(H)⟶Pfin,1(K)F : \mathcal{P}_{\mathrm{fin},1}(H) \longrightarrow\mathcal{P}_{\mathrm{fin},1}(K) is cardinality-preserving, that is, ∣F(X)∣=∣X∣|F(X)| = |X| for any X∈Pfin,1(H)X \in\mathcal{P}_{\mathrm{fin},1}(H).

Main Results

Let HH be a semigroup and let h∈Hh \in H. The “correlation function”

νh:Pfin(H)⟶N,A⟼∣{g∈A:hg∈A}∣\nu_h : \mathcal{P}_{\mathrm{fin}}(H) \longrightarrow\mathbb{N}, \qquad A \longmapsto\lvert\{g \in A : hg \in A\}\rvert

was first introduced in [4 Definition 2.5]. This function is closely related to the discrete covariogram in convex geometry [3 Section 6.1] and to difference representation functions in additive combinatorics [11 Section 6.1]. We need the following properties of νh\nu_h.

Lemma 3.1. Let HH be a commutative and cancellative monoid. Then for any y,z∈Hy,z \in H and A∈Pfin(H)A \in\mathcal{P}_{\mathrm{fin}}(H),

(1) νz(A)=∣A∩zA∣\nu_z(A)=\lvert A\cap zA\rvert;

(2) νz(A)=νz(yA)\nu_z(A)=\nu_z(yA);

(3) if A=⋃i=1nAiA=\bigcup_{i=1}^{n} A_i is a disjoint union, and zAi∩Aj=∅zA_i\cap A_j=\varnothing for all i,j∈{1,2,…,n}i,j\in\{1,2,\ldots,n\} with i≠ji\ne j, then

νz(A)=∑i=1nνz(Ai).\nu_z(A)=\sum_{i=1}^{n}\nu_z(A_i).

Proof. (1) is a special case of [4 Lemma 2.6].

(2) Since HH is commutative and cancellative, the map

A∩zA⟶yA∩yzA:x⟼xy,A\cap zA\longrightarrow yA\cap yzA:\qquad x\longmapsto xy,

is a bijection, so (2) follows from (1).

(3) By the assumptions, we see that

A∩zA=(⋃i=1nAi)∩(⋃j=1nzAj)=⋃1≤i,j≤n(Ai∩zAj)=⋃1≤i≤n(Ai∩zAi).A\cap zA=\left(\bigcup_{i=1}^{n}A_i\right)\cap\left(\bigcup_{j=1}^{n}zA_j\right)=\bigcup_{1\le i,j\le n}(A_i\cap zA_j)=\bigcup_{1\le i\le n}(A_i\cap zA_i).

Hence,

νz(A)=∣A∩zA∣=∣⋃1≤i≤n(Ai∩zAi)∣=∑i=1n∣Ai∩zAi∣=∑i=1nνz(Ai).\nu_z(A)=\lvert A\cap zA\rvert=\left\lvert\bigcup_{1\le i\le n}(A_i\cap zA_i)\right\rvert=\sum_{i=1}^{n}\lvert A_i\cap zA_i\rvert=\sum_{i=1}^{n}\nu_z(A_i).

□

Lemma 3.2. [4 Corollary 2.8] Let HH be a commutative and cancellative monoid containing an element of infinite order, and let FF be an automorphism of Pfin,1(H)\mathcal{P}_{\mathrm{fin},1}(H) with identity pullback. Then

νz(A)=νz(F(A))\nu_z(A)=\nu_z(F(A))

for any z∈Hz\in H and A∈Pfin,1(H)A\in\mathcal{P}_{\mathrm{fin},1}(H).

Now we can state and prove our main results.

Theorem 3.3. Let HH be a commutative and cancellative monoid, and suppose that HH contains an invertible element of infinite order. Then all automorphisms of Pfin,1(H)\mathcal{P}_{\mathrm{fin},1}(H) are inner.

Proof. Let Ph={1H,h}P_h=\{1_H,h\} for any h∈Hh\in H and let g∈Hg\in H be an invertible element of infinite order.

By Theorem 2.4, the pullback of every isomorphism of Pfin,1(H)P_{\mathrm{fin},1}(H) is an automorphism of HH. Thus, by Lemma 2.5, it suffices to show that if an automorphism FF of Pfin,1(H)P_{\mathrm{fin},1}(H) has trivial pullback, then FF is the identity map. Accordingly, we assume below that

F(Ph)=Phfor every h∈H.(1)F(P_h)=P_h \qquad\text{for every }h\in H. \tag*{(1)}

Fix S∈Pfin,1(H)S\in P_{\mathrm{fin},1}(H) and let T=F(S)T=F(S). We proceed to show that S⊂TS\subset T. Take z∈Sz\in S. Suppose z≠1Hz\ne1_H, otherwise the conclusion is obvious. The proof is divided into two cases.

Case 1: zz is invertible. Since U:=T∪z−1TU:=T\cup z^{-1}T is also finite, by Lemma 2.6 there exists a positive integer NN such that giNU∩gjNU=∅g^{iN}U\cap g^{jN}U=\varnothing for any pairwise distinct i,j∈Zi,j\in\mathbb{Z}. Put w=gNw=g^N. Then we have

wiR1∩wjR2=∅,for R1,R2∈{T,z−1T} and i,j∈Z with i≠j.(2)w^iR_1\cap w^jR_2=\varnothing,\qquad\text{for }R_1,R_2\in\{T,z^{-1}T\}\text{ and }i,j\in\mathbb{Z}\text{ with }i\ne j. \tag*{(2)}

Put A=w−1S∪wz−1S∪{1H}A=w^{-1}S\cup wz^{-1}S\cup\{1_H\}. Since 1H,z∈S1_H,z\in S, we have Pw⊂S∪wz−1SP_w\subset S\cup wz^{-1}S. Thus,

Pw⋅A=A∪wA=w−1S∪wz−1S∪S∪w2z−1S={1H,w−1,wz−1,w2z−1}⋅S=Pw−1⋅Pw2z−1⋅S.\begin{aligned} P_w\cdot A&=A\cup wA\\ &=w^{-1}S\cup wz^{-1}S\cup S\cup w^2z^{-1}S\\ &=\{1_H,w^{-1},wz^{-1},w^2z^{-1}\}\cdot S\\ &=P_{w^{-1}}\cdot P_{w^2z^{-1}}\cdot S. \end{aligned}

Applying FF to the above equation and using (1) gives that

Pw⋅B=Pw−1⋅Pw2z−1⋅T,(3)P_w\cdot B=P_{w^{-1}}\cdot P_{w^2z^{-1}}\cdot T, \tag*{(3)}

where B=F(A)B=F(A). That is,

B∪wB=w−1T∪T∪wz−1T∪w2z−1T.(4)B\cup wB=w^{-1}T\cup T\cup wz^{-1}T\cup w^2z^{-1}T. \tag*{(4)}

By (2) and (4), we obtain that

w−2T∩B⊂w−2T∩(w−1T∪T∪wz−1T∪w2z−1T)=⋃i=−10(w−2T∩wiT)∪(⋃j=12(w−2T∩wjz−1T))=∅,\begin{aligned} w^{-2}T\cap B &\subset w^{-2}T\cap\left(w^{-1}T\cup T\cup wz^{-1}T\cup w^2z^{-1}T\right)\\ &=\bigcup_{i=-1}^{0}\left(w^{-2}T\cap w^iT\right)\cup\left(\bigcup_{j=1}^{2}\left(w^{-2}T\cap w^jz^{-1}T\right)\right)\\ &=\varnothing, \end{aligned}

and thus w−1T∩wB=∅w^{-1}T\cap wB=\varnothing. So w−1T⊆Bw^{-1}T\subseteq B. By the same argument, we have w3z−1T∩wB=∅w^3z^{-1}T\cap wB=\varnothing and thus w2z−1T∩B=∅w^2z^{-1}T\cap B=\varnothing. It follows that w2z−1T⊆wBw^2z^{-1}T\subseteq wB and wz−1T⊆Bwz^{-1}T\subseteq B.

We have shown w−1T,wz−1T⊆Bw^{-1}T,wz^{-1}T\subseteq B and w2z−1T∩B=∅w^2z^{-1}T\cap B=\varnothing. Combining these with (4) gives

B=w−1T∪Z∪wz−1T,B=w^{-1}T\cup Z\cup wz^{-1}T,

where Z=B∩TZ=B\cap T. Substituting BB into equation (4) yields that

wZ⊆w−1T∪T∪wz−1T∪w2z−1T.wZ\subseteq w^{-1}T\cup T\cup wz^{-1}T\cup w^2z^{-1}T.

Since Z⊆TZ \subseteq T, we have wZ∩wiT=wZ∩wjz−1T=∅wZ \cap w^{i}T = wZ \cap w^{j}z^{-1}T = \varnothing for any integers i,j≠1i,j \ne1 by (2). We conclude that wZ⊆wz−1TwZ \subseteq wz^{-1}T and Z⊆z−1TZ \subseteq z^{-1}T. Combining with that 1H∈Z1_H \in Z, we obtain 1H∈z−1T1_H \in z^{-1}T and thus z∈Tz \in T.

Case 2: zz is not invertible. By Lemma 2.6 and the same arguments at the begin of Case 1, we may choose N>0N > 0 such that

wiR1∩wjR2=∅,for R1,R2∈{S,T,zS,zT} and i,j∈Z with i≠j,(5)w^{i}R_1 \cap w^{j}R_2 = \varnothing,\qquad\text{for } R_1,R_2 \in\{S,T,zS,zT\}\text{ and }i,j \in\mathbb{Z}\text{ with }i \ne j, \tag*{(5)}

where w=gNw = g^N.

Put

C={w−2,w−1,1H,w}.C = \{w^{-2},w^{-1},1_H,w\}.

We have the following claim.

Claim: For given R∈{S,T}R \in\{S,T\} and X∈Pfin,1(H)X \in\mathcal{P}_{\mathrm{fin},1}(H),

Pw−1⋅X=C⋅R(6)P_{w^{-1}} \cdot X = C \cdot R \tag*{(6)}

if and only if

X=w−1R∪V∪wR,1H∈V⊆R.(7)X = w^{-1}R \cup V \cup wR,\qquad1_H \in V \subseteq R. \tag*{(7)}

Proof of the Claim. It is easy to verify that the form given in (7) satisfies (6) and belongs to Pfin,1(H)\mathcal{P}_{\mathrm{fin},1}(H). Conversely, suppose A∈Pfin,1(H)A \in\mathcal{P}_{\mathrm{fin},1}(H) and Pw−1⋅A=C⋅RP_{w^{-1}} \cdot A = C \cdot R. That is,

A∪w−1A=⋃i=−21wiR.(8)A \cup w^{-1}A = \bigcup_{i=-2}^{1} w^{i}R. \tag*{(8)}

We need to show that w−2R∩A=∅w^{-2}R \cap A = \varnothing, and w−1R,wR⊆Aw^{-1}R,wR \subseteq A. If w−2R∩A≠∅w^{-2}R \cap A \ne\varnothing and x∈w−2R∩Ax \in w^{-2}R \cap A, then w−1x∈w−3R∩w−1Aw^{-1}x \in w^{-3}R \cap w^{-1}A. By (8), it follows that

w−1x∈w−3R∩(⋃i=−21wiR)=⋃i=−21(w−3R∩wiR),w^{-1}x \in w^{-3}R \cap\left(\bigcup_{i=-2}^{1} w^{i}R\right) = \bigcup_{i=-2}^{1}\left(w^{-3}R \cap w^{i}R\right),

which contradicts to (5). So w−2R∩A=∅w^{-2}R \cap A = \varnothing, and thus w−2R⊆w−1Aw^{-2}R \subseteq w^{-1}A by (8). We obtain w−1R⊆Aw^{-1}R \subseteq A.

Similarly, if wR∩w−1A≠∅wR \cap w^{-1}A \ne\varnothing, then wy∈w2R∩Awy \in w^{2}R \cap A for any y∈wR∩w−1Ay \in wR \cap w^{-1}A. By (8), it follows that wy∈w2R∩(⋃i=−21wiR)wy \in w^{2}R \cap\left(\bigcup_{i=-2}^{1} w^{i}R\right), which contradicts to (5). So wR∩w−1A=∅wR \cap w^{-1}A = \varnothing and we get wR⊆AwR \subseteq A by (8).

Combining (8) with that w−2R∩A=∅w^{-2}R \cap A = \varnothing, and w−1R,wR⊆Aw^{-1}R,wR \subseteq A, we have

A=⋃i=−21(A∩wiR)=w−1R∪V∪wR,1H∈V=A∩R⊆R.A = \bigcup_{i=-2}^{1}(A \cap w^{i}R) = w^{-1}R \cup V \cup wR,\qquad1_H \in V = A \cap R \subseteq R.

This completes the proof of the Claim.

Let X=w−1S∪Pz∪wSX = w^{-1}S \cup P_z \cup wS. Since Pz⊆SP_z \subseteq S, by the Claim we have

Pw−1⋅X=C⋅S.(9)P_{w^{-1}} \cdot X = C \cdot S. \tag*{(9)}

Note that C=Pw−1⋅Pw−1⋅PwC = P_{w^{-1}} \cdot P_{w^{-1}} \cdot P_w. Applying FF to (9) and using (1) gives that

Pw−1⋅Y=C⋅T,P_{w^{-1}} \cdot Y = C \cdot T,

where Y=F(X)Y = F(X). By the Claim again,

Y=w−1T∪V∪wT,1H∈V⊆T.Y = w^{-1}T \cup V \cup wT,\qquad1_H \in V \subseteq T.

By {(xref)}(3.5), both X=w−1S∪Pz∪wSX=w^{-1}S\cup P_z\cup wS and Y=w−1T∪V∪wTY=w^{-1}T\cup V\cup wT are disjoint unions. Since FF is cardinality-preserving by Lemma 2.7, we assert that ∣S∣=∣T∣|S|=|T| and

∣X∣=2∣S∣+2=∣Y∣=2∣T∣+∣V∣.|X|=2|S|+2=|Y|=2|T|+|V|.

It follows that ∣V∣=2|V|=2 and thus V=PvV=P_v for some non-identity element v∈Tv\in T. We are now in a position to show that v=zv=z. By {(xref)}(3.5), we have R1∩zR2=∅R_1\cap zR_2=\varnothing for distinct R1,R2∈{w−1S,Pz,wS}R_1,R_2\in\{w^{-1}S,P_z,wS\}. By (2), (3) of Lemma 3.1,

νz(X)=νz(w−1S∪Pz∪wS)=νz(w−1S)+νz(Pz)+νz(wS)=2νz(S)+νz(Pz).\begin{aligned} \nu_z(X)&=\nu_z(w^{-1}S\cup P_z\cup wS)\\ &=\nu_z(w^{-1}S)+\nu_z(P_z)+\nu_z(wS)\\ &=2\nu_z(S)+\nu_z(P_z). \end{aligned}

By the same discussion, we also have

νz(Y)=2νz(T)+νz(Pv).\nu_z(Y)=2\nu_z(T)+\nu_z(P_v).

By Lemma 3.2, νz(S)=νz(T)\nu_z(S)=\nu_z(T) and νz(X)=νz(Y)\nu_z(X)=\nu_z(Y). Hence

νz(Pz)=νz(Pv)∈{1,2}.\nu_z(P_z)=\nu_z(P_v)\in\{1,2\}.

Since zz is not invertible, z2∉Pzz^2\notin P_z and νz(Pz)=1\nu_z(P_z)=1. Therefore,

νz(Pv)=∣{1H,v}∩{z,zv}∣=1.\nu_z(P_v)=|\{1_H,v\}\cap\{z,zv\}|=1.

It is clear that zv∈{1H,v}zv\in\{1_H,v\} implies zz is invertible. So zv∉Pvzv\notin P_v and z∈Pvz\in P_v. We thus get z=v∈Tz=v\in T.

We have shown that S⊆TS\subseteq T. Since ∣S∣=∣T∣<∞|S|=|T|<\infty, we have S=TS=T. Therefore, FF is the identity map. □

Theorem 3.4. Let HH be an abelian group which is not isomorphic to the Klein four-group. Then

Φ:Aut⁡(H)⟶Aut⁡(Pfin,1(H)),f⟼f∗\Phi:\operatorname{Aut}(H)\longrightarrow\operatorname{Aut}(P_{\mathrm{fin},1}(H)),\quad f\longmapsto f^*

is an isomorphism.

Proof. By Theorem 2.3 and Theorem 3.3, we only need to deal with the case that HH is infinite and torsion. By Theorem 2.2, the pullback of an automorphism of Pfin,1(H)P_{\mathrm{fin},1}(H) is also an automorphism of HH since HH is torsion. By Lemma 2.5, it suffices to prove that if FF is an automorphism of Pfin,1(H)P_{\mathrm{fin},1}(H) with identity pullback, then FF is the identity map.

Let S∈Pfin,1(H)S\in P_{\mathrm{fin},1}(H) and let Ph={1H,h}P_h=\{1_H,h\} for h∈Hh\in H. Since HH is infinite, abelian, and torsion, there exists a finite subgroup GG of HH such that S⊆GS\subseteq G and ∣G∣>4|G|>4. Since G=∏g∈GPgG=\prod_{g\in G}P_g, and FF fixes PgP_g, it follows that

F(G)=F(∏g∈GPg)=∏g∈GF(Pg)=∏g∈GPg=G.F(G)=F\left(\prod_{g\in G}P_g\right)=\prod_{g\in G}F(P_g)=\prod_{g\in G}P_g=G.

Since GG is a subgroup of HH, we have A⋅G=GA\cdot G=G if and only if A⊆GA\subseteq G for any A∈Pfin,1(H)A\in P_{\mathrm{fin},1}(H). Therefore FF maps Pfin,1(G)P_{\mathrm{fin},1}(G) into itself. The same argument applied to F−1F^{-1} shows that the restriction F∣Pfin,1(G)F|_{P_{\mathrm{fin},1}(G)} is also an automorphism of Pfin,1(G)P_{\mathrm{fin},1}(G) with identity pullback. Since ∣G∣>4|G| > 4, by Theorem 2.3 the restriction F∣Pfin,1(G)F|_{\mathcal{P}_{\mathrm{fin},1}(G)} itself is the identity map. Hence, F(S)=SF(S) = S, which completes the proof. □

Bhowmik and Tringali [4, Theorem 1.2] showed that Aut⁡(Pfin,0(H))\operatorname{Aut}(\mathcal{P}_{\mathrm{fin},0}(H)) is trivial for every numerical monoid H≠NH \ne\mathbb{N}. Together with our results, this motivates the following conjecture.

Conjecture 3.5. Let HH be a commutative and cancellative monoid with nontrivial torsion unit group. Suppose that HH is not isomorphic to C2×C2C_2 \times C_2. Then all automorphisms of Pfin,1(H)\mathcal{P}_{\mathrm{fin},1}(H) are inner.

AI Declaration

During the preparation of this manuscript, the authors used ChatGPT and Kimi to assist with checking the set equation involving four translates in the proof of Theorem 3.3, verifying reference formats, and improving English grammar. The main mathematical results and their proofs were developed by the authors independently of these tools. The authors reviewed and edited all AI-assisted material and takes full responsibility for the final manuscript.

References

References

  1. [1]A. A. Antoniou and S. Tringali, On the Arithmetic of Power Monoids and Sumsets in Cyclic Groups, Pacific J. Math. 312 (2021), No. 2, 279–308.DOI
  2. [2]R. Ballieu, Sur les groupes de parties d’un demi-groupe (French), Ann. Soc. Sci. Bruxelles, Sér. I 64 (1950), 139–147.
  3. [3]G. Bianchi, The covariogram problem, pp. 37–82 in: A. Koldobsky and A. Volberg (eds.), Harmonic Analysis and Convexity, De Gruyter, Boston, 2023.DOI
  4. [4]A. Bhowmik and S. Tringali, On the automorphisms of numerical power monoids, preprint (arXiv: 2607.23301).
  5. [5]R. D. Byrd, J. T. Lloyd, F. D. Pedersen, and J. W. Stepp, Automorphisms of the semigroup of finite complexes of locally finite groups, J. Reine Angew. Math. 299/300 (1978), 151–160.
  6. [6]R. D. Byrd, J. T. Lloyd, F. D. Pedersen, and J. W. Stepp, The automorphism group of some semigroups, Fund. Math. 124 (1984), 187–195.
  7. [7]L. Cossu and S. Tringali, On the arithmetic of power monoids, J. Algebra 686 (2026), 793–813.DOI
  8. [8]Y. Fan and S. Tringali, Power monoids: A bridge between Factorization Theory and Arithmetic Combinatorics, J. Algebra 512 (2018), 252–294.DOI
  9. [10]J. M. Howie, Fundamentals of Semigroup Theory, London Math. Soc. Monogr. Ser. 12, Oxford Univ. Press, 1995.DOI
  10. [11]M. B. Nathanson, Problems in additive number theory, III, pp. 279–297 in: A. Geroldinger and I.Z. Ruzsa, Combinatorial Number Theory and Additive Group Theory, Birkhäuser, 2009.
  11. [12]B. Rago, The automorphism group of reduced power monoids of finite abelian groups, preprint (arXiv: 2510.17533).arxiv.org/abs/2510.17533
  12. [14]B. Rago, The isomorphism problem for reduced finitely power monoids, preprint (arXiv: 2601.22469).arxiv.org/abs/2601.22469
  13. [15]J. Shafer, Note on power semigroups, Math. Japon. 12 (1967), 32.
  14. [16]T. Tamura and J. Shafer, Power semigroups, Math. Japon. 12 (1967), 25–32; Errata, ibid. 29 (1984), No. 4, 679.
  15. [17]S. Tringali, Power monoids and their arithmetic: a survey, Amer. Math. Monthly, to appear (https://arxiv.org/abs/2602.15754).
  16. [18]S. Tringali and K. Wen, The additive group of integers and the automorphisms of its power monoids, preprint(arXiv:2504.12566).
  17. [19]S. Tringali and K. Wen, On the automorphisms of the power semigroups of a numerical semigroup, Trans. London Math. Soc. 13 (2026), No. 1, Art. ID e70030, 11 pp.DOI
  18. [21]S. Tringali and W. Yan, On power monoids and their automorphisms, J. Combin. Theory Ser. A 209 (2025), 105961, 16 pp.DOI
  19. [22]S. Tringali and W. Yan, Torsion groups and the Bienvenu-Geroldinger conjecture, Bull. London Math. Soc. 58 (2026), No. 7, Art. ID e70439, 16 pp.DOI
  20. [23]D. Wong, S. Xu, C. Zhang, and Z. Wang, On automorphism group of the reduced finitely power monoid of the additive group of integers, preprint (arXiv: 2602.19493).arxiv.org/abs/2602.19493
  21. [24]D. Wong, S. Xu, C. Zhang and J. Zhao, On automorphism groups of numerical power semigroups, J. Algebra 713 (2027), 356–367.DOI

Paper details

vlnc:W_uxk3h8sk
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Mathematics

Not recorded for this paper

View statement
Looking for related papers…
Contents