The representation theory of the wreath product of a finite group with the monoid of all partial functions on a finite set as an EI-category algebra

Let $\mathcal{A}$ be a small category. We denote by $\mathcal{A}^{0}$ and $\mathcal{A}^{1}$ the sets of objects and morphisms of $\mathcal{A}$, respectively. For $a,b\in\mathcal{A}^{0}$, we write $\mathcal{A}(a,b)$ for the hom-set of morphisms with domain $a$ and range $b$. Recall that a monoid can be viewed as a category with a single object. Following [25], we denote by $\operatorname{\mathbf{Pfn}}$ the category whose objects are finite sets and whose morphisms are partial functions, and by $\operatorname{\bf{Set}}$ the subcategory consisting of total functions as morphisms.

Let $G$ be a finite group with identity element $1_{G}$, and let $\operatorname{PT}(\mathcal{X},G)$ denote the set of all partial functions from $\mathcal{X}$ to $G$. For every $f\in\operatorname{PT}(\mathcal{X},G)$, we denote by $\operatorname{dom}(f)$ its domain. The set $\operatorname{PT}(\mathcal{X},G)$ is a monoid under pointwise multiplication, where for any $f_{1},f_{2}\in\operatorname{PT}(\mathcal{X},G)$, the product $f_{1}\cdot f_{2}$ has domain $\operatorname{dom}(f_{1})\cap\operatorname{dom}(f_{2})$ and is defined by:

$$(f_{1}\cdot f_{2})(x)=f_{1}(x)\cdot f_{2}(x)\quad\text{for all }x\in\operatorname{dom}(f_{1})\cap\operatorname{dom}(f_{2}).$$

The identity element of this monoid is the constant function mapping every $x\in\mathcal{X}$ to $1_{G}$.

Let $H:\mathcal{A}\to\operatorname{\mathbf{Pfn}}$ be a functor. Define a new category $G\wr_{H}\mathcal{A}$ in the following way. The set of objects is the same as the set of objects of $\mathcal{A}$, that is, $(G\wr_{H}\mathcal{A})^{0}=\mathcal{A}^{0}$. Given two objects $a,b\in\mathcal{A}^{0}$, the hom-set $(G\wr_{H}\mathcal{A})(a,b)$ is

$$\{(f,m)\mid f\in\operatorname{PT}(H(a),G),\,m\in\mathcal{A}(a,b),\,\text{where}\,\operatorname{dom}(f)=\operatorname{dom}(H(m))\}.$$

Now, given two morphisms $(f,m)\in(G\wr_{H}\mathcal{A})(a,b)$ and $(f^{\prime},m^{\prime})\in(G\wr_{H}\mathcal{A})(b,c)$ the composition is

$$(f^{\prime},m^{\prime})\cdot(f,m)=((f^{\prime}\circ H(m))\cdot f,m^{\prime}m).$$

It is routine to verify that this composition is well defined and that $G\wr_{H}\mathcal{A}$ is indeed a category. If $\operatorname{id}_{a}$ is the identity morphism of $a\in\mathcal{A}^{0}$ and ${\bf 1}_{H(a)}:H(a)\to G$ is the function defined by ${\bf 1}_{H(a)}(x)=1_{G}$ for every $x\in H(a)$, then the identity morphism of the object $a\in(G\wr_{H}\mathcal{A})^{0}$ is $({\bf 1}_{H(a)},\operatorname{id}_{a})$. The category $G\wr_{H}\mathcal{A}$ is called the partial wreath product of $G$ and $\mathcal{A}$ with respect to $H$. We will apply this construction in two special cases. In the first case, $H$ is a functor $H:\mathcal{A}\to\operatorname{\bf{Set}}$. In this case, the construction reduces to the standard wreath product of a group with a category (see, for example, [23, 34]). In the second case, $\mathcal{A}$ is a monoid $M$ with identity element $1_{M}$. Here the construction is well-known, even when $G$ and $M$ are semigroups. It appears in [5] in the language of transformation semigroups, and in [10] under the name of 0-wreath products. See also [2, Section 9.1] and references therein. In this case, a functor $H:M\to\operatorname{\mathbf{Pfn}}$ is an incomplete $M$-action. That is, it consists of a set $\mathcal{X}$ and a monoid homomorphism $\varphi:M\to\operatorname{PT}_{\mathcal{X}}$. For $m\in M$ and $x\in\mathcal{X}$, we will write $m\bullet x$ instead of $\varphi(m)(x)$. For every $m\in M$, we write

$$\operatorname{dom}(m)=\operatorname{dom}(\varphi(m))\subseteq\mathcal{X}.$$

The monoid $M$ acts on the right of $\operatorname{PT}(\mathcal{X},G)$ by $f\ast m=f\circ\varphi(m)$. Explicitly, for $x\in\mathcal{X}$, the partial function $f\ast m$ is given by

$$(f\ast m)(x)=\begin{cases}f(m\bullet x)&m\bullet x\kern 5.0pt\text{and}\kern 5.0ptf(m\bullet x)\kern 5.0pt\text{are both defined}\\ \text{undefined}&\text{otherwise}.\end{cases}$$

In this case, the partial wreath product $G\wr_{\mathcal{X}}M$ is a monoid whose underlying set is

$$G\wr_{\mathcal{X}}M=\{(f,m)\mid m\in M,\quad f\in\operatorname{PT}(\mathcal{X},G),\quad\operatorname{dom}(f)=\operatorname{dom}(m)\}.$$

The operation is given by

$$(f_{1},m_{1})\cdot(f_{2},m_{2})=((f_{1}\ast m_{2})\cdot f_{2},m_{1}m_{2}),$$

where $(f_{1}\ast m_{2})(x)=f_{1}(m_{2}\bullet x)$ and the identity element is $({\bf 1}_{\mathcal{X}},1_{M})$.

The case where $M$ is a group is of great importance. If $f:\mathcal{X}\to G$ is a (total) function, we denote by $f^{-1}:\mathcal{X}\to G$ the function defined by $f^{-1}(x)=\left(f(x)\right)^{-1}$ for every $x\in\mathcal{X}.$ If $M$ is a group, then $G\wr_{\mathcal{X}}M$ is a group, and the inverse of $(f,m)\in G\wr_{\mathcal{X}}M$ is given by

$$(f,m)^{-1}=(\left(f\ast m^{-1}\right)^{-1},m^{-1}).$$

There is a natural incomplete action of the monoid $\operatorname{PT}_{\mathcal{X}}$ on the set $\mathcal{X}$. Formally, the action is given by the identity function $\varphi:\operatorname{PT}_{\mathcal{X}}\to\operatorname{PT}_{\mathcal{X}}$. In this case, we simply write $G\wr\operatorname{PT}_{\mathcal{X}}$ for $G\wr_{\mathcal{X}}\operatorname{PT}_{\mathcal{X}}$. If $\mathcal{X}=[n]=\{1,\ldots,n\}$, we denote the corresponding wreath product by $G\wr\operatorname{PT}_{n}$. In this special case, the wreath product has a natural description using matrices over $G\cup\{0\}$. For $\alpha\in\operatorname{PT}_{n}$ and $f\in\operatorname{PT}([n],G)$ with $\operatorname{dom}(f)=\operatorname{dom}(\alpha)$, we denote by $[f,\alpha]$ an $n\times n$ matrix over $G\cup\{0\}$ defined by

$$[f,\alpha]_{i,j}=\begin{cases}f(j)&\alpha(j)=i,\\ 0&\text{otherwise}.\end{cases}$$

Let $\mathcal{M}$ denote the set of all such matrices. Note that each $[f,\alpha]\in\mathcal{M}$ has at most one non-zero entry in each column, so matrix multiplication is well-defined. It is straightforward to verify that the function $\psi:G\wr\operatorname{PT}_{n}\to\mathcal{M}$ defined by $\psi((f,\alpha))=[f,\alpha]$ is a monoid isomorphism.

Let $S$ be a semigroup and let $E\subseteq S$ be a subset of idempotents. We define two equivalence relations $\widetilde{\mathcal{L}}_{E}$ and $\widetilde{\mathcal{R}}_{E}$ on $S$.

$$a\widetilde{\mathcal{L}}_{E}b\iff(\forall e\in E\quad be=b\Leftrightarrow ae=a)$$

$$a\widetilde{\mathcal{R}}_{E}b\iff(\forall e\in E\quad eb=b\Leftrightarrow ea=a).$$

A subset $E\subseteq S$ of idempotents is called a subsemilattice if it is a commutative subsemigroup. It is well known that any commutative semigroup of idempotents has the structure of a semilattice (i.e. a poset where every two elements have a meet) if one defines $a\leq b$ whenever $ab=ba=a$. A semigroup $S$ with a subsemilattice $E\subseteq S$ is called right $E$-Ehresmann if every $\widetilde{\mathcal{L}}_{E}$-class contains a unique idempotent from $E$ and $\widetilde{\mathcal{L}}_{E}$ is a right congruence. We denote the unique idempotent in the $\widetilde{\mathcal{L}}_{E}$-class of $a$ by $a^{\ast}$. Note that $a^{\ast}$ is the unique minimal element $e\in E$ such that $ae=a$. It is well known that $\widetilde{\mathcal{L}}_{E}$ is a right congruence if and only if the identity $(ab)^{\ast}=(a^{\ast}b)^{\ast}$ holds for every $a,b\in S$.

Dually, we can consider semigroups for which every $\widetilde{\mathcal{R}}_{E}$ class contains a unique idempotent. We denote the unique idempotent in the $\widetilde{\mathcal{R}}_{E}$ class of $a$ by $a^{+}$. Such a semigroup is called left $E$-Ehresmann if $\widetilde{\mathcal{R}}_{E}$ is a left congruence, or equivalently if $(ab)^{+}=(ab^{+})^{+}$ for every $a,b\in S$. A semigroup $S$ with a subsemilattice $E\subseteq S$ is called $E$-Ehresmann if it is both left and right $E$-Ehresmann. The semilattice $E$ is also called the set of projections of $S$.

A right (left) Ehresmann semigroup $S$ is called right (respectively, left) restriction if the “right ample” (respectively, “left ample”) identity $ea=a(ea)^{\ast}$ (respectively, $ae=(ae)^{+}a$) holds for every $a\in S$ and $e\in E$.

For every Ehresmann semigroup $S$, we can associate a category ${\bf C}(S)$ in the following way. The object set of ${\bf C}(S)$ is the set $E$ of projections and the morphisms of ${\bf C}(S)$ are in one-to-one correspondence with elements of $S$. For every $a\in S$, we associate a morphism $C(a)\in{\bf C}(S)^{1}$ with domain $a^{\ast}$ and range $a^{+}$ and if the range of $C(a)$ is the domain of $C(b)$, the composition $C(b)\cdot C(a)$ is defined to be $C(ba)$. For more facts and proofs on Ehresmann semigroups and Ehresmann categories, the reader is referred to [6, 7].

Let $A$ be an algebra. We will only discuss unital, finite-dimensional $\mathbb{C}$-algebras in this paper. Likewise, when we say that $V$ is a module over $A$ (or an $A$-module or an $A$-representation), we mean that $V$ is a finite-dimensional left module over $A$. For $a\in A$ and $v\in V$, we write $a\bullet v$ for the action of $a$ on the module element $v$. An $A$-module $V$ is simple (or irreducible) if $0$ is its only proper submodule. The ordinary quiver $Q$ of a finite dimensional algebra $A$ is a directed graph defined in the following way: The vertices of $Q$ are in a one-to-one correspondence with the simple modules of $A$ (up to isomorphism). If $V_{i}$ and $V_{j}$ are simple modules of $A$ (identified with two vertices of the quiver), then the number of arrows from $V_{i}$ to $V_{j}$ is

$$\dim\operatorname{Ext}^{1}(V_{i},V_{j})$$

where $\operatorname{Ext}^{1}(V,-)$ is the first right derived functor of $\operatorname{Hom}(V,-)$, see [20, Chapters 6-7]. More about modules of algebras and quivers can be found in [1].

In this paper, we discuss complex algebras of finite categories. Let $\mathcal{A}$ be a finite category. The category algebra $\mathbb{C}\mathcal{A}$ is a vector space over $\mathbb{C}$ with the morphisms of $\mathcal{A}$ as its basis. It consists of all formal linear combinations:

$$\left\{\sum_{i=1}^{n}k_{i}m_{i}\mid k_{i}\in\mathbb{C},\,m_{i}\in\mathcal{A}^{1}\right\}.$$

The multiplication in $\mathbb{C}\mathcal{A}$ is the linear extension of the following:

$$m^{\prime}\cdot m=\begin{cases}m^{\prime}m&\text{if }m^{\prime}m\text{ is defined in }\mathcal{A},\\ 0&\text{otherwise}.\end{cases}$$

The algebra $\mathbb{C}\mathcal{A}$ has a unit element given by the sum of the identity morphisms of all objects in $\mathcal{A}$:

$$1_{\mathbb{C}\mathcal{A}}=\sum_{c\in\mathcal{A}^{0}}\operatorname{id}_{c}.$$

Since a monoid is a category with a single object, this construction naturally specializes to the definition of a monoid algebra. In this case, the algebra $\mathbb{C}M$ consists of all formal linear combinations of elements of the monoid $M$, with multiplication defined as the linear extension of the monoid operation.

Let $G$ be a finite group. If $V$ is a $\mathbb{C}$G-module, we will usually simply say that $V$ is a $G$-module (or a $G$-representation). Equivalently, a $G$-module is a pair $(V,\rho)$ of a $\mathbb{C}$-vector space $V$ and a group homomorphism $\rho:G\to\operatorname{GL}(V)$. We denote the set of simple modules of $G$ (up to isomorphism) by $\operatorname{IRep}G$. It is well known that every $G$-module is a finite direct sum of simple modules and that the number of different simple $G$-modules (up to isomorphism) is the number of conjugacy classes of $G$. We denote the trivial module of any group $G$ by $\operatorname{tr}_{G}$. Recall that if $V$ is a $G$-module, then $V^{\ast}=\operatorname{Hom}(V,\mathbb{C})$ is also a $G$-module with operation $(g\bullet\varphi)(v)=\varphi(g^{-1}\bullet v)$. Let $U$ and $V$ be $G$-modules. The inner tensor product $U\otimes V$ is again a $G$-module with action defined by $g\bullet(u\otimes v)=(g\bullet u)\otimes(g\bullet v)$ and extending linearly. Now, assume that $U_{1}$ and $U_{2}$ are modules of $G_{1}$ and $G_{2}$, respectively. The outer tensor product $U_{1}\otimes U_{2}$ of $U_{1}$ and $U_{2}$ is the ($G_{1}\times G_{2}$)-module where $(g_{1},g_{2})\bullet(u_{1}\otimes u_{2})=(g_{1}\bullet u_{1})\otimes(g_{2}\bullet u_{2})$. Although the two types of tensor product can be distinguished by context, we prefer to use a different notation for the outer tensor product, denoting it by $\boxtimes$. Similarly, the simple tensors of $U\boxtimes V$ will be denoted by $u\boxtimes v$. It is well known that $\operatorname{IRep}(G_{1}\times G_{2})=\{U\boxtimes V\mid U\in\operatorname{IRep}G_{1},V\in\operatorname{IRep}G_{2}\}$. The character $\chi_{U}$ of the $G$-module $(U,\rho)$ is the function $\chi_{U}:G\to\mathbb{C}$ defined by $\chi_{U}(g)=\operatorname{trace}(\rho(g))$. Recall that the multiplicity of $U\in\operatorname{IRep}G$ as a simple constituent in some $G$-module $V$ is given by the inner product of characters

$$\langle\chi_{U},\chi_{V}\rangle_{G}=\frac{1}{|G|}\sum_{g\in G}\chi_{U}(g)\overline{\chi_{V}(g)}.$$

We may omit the subscript $G$ when the group is clear from the context. Recall also that $\chi_{V^{\ast}}(g)=\overline{\chi_{V}(g)}$, $\chi_{U\boxtimes V}((g_{1},g_{2}))=\chi_{U}(g_{1})\chi_{V}(g_{2})$ and $\chi_{U\otimes V}(g)=\chi_{U}(g)\chi_{V}(g)$. In order to simplify notation, we will usually omit $\chi$ and write $U$ also for the character of $U$. Hence the above inner product will be written as

$$\langle U,V\rangle=\frac{1}{|G|}\sum_{g\in G}U(g)\overline{V(g)}.$$

Let $(U,\rho)$ be a $G$-module and let $H\leq G$ be a subgroup. The restriction of $(U,\rho)$ to $H$ is the $H$-module $(\operatorname{Res}_{H}^{G}U,\operatorname{Res}_{H}^{G}\rho)$ defined by

$$\operatorname{Res}_{H}^{G}\rho(h)(u)=\rho(h)(u)$$

that is, restricting the homomorphism to the subgroup $H$. Note that $\dim\operatorname{Res}_{H}^{G}U\allowbreak=\dim U$ and if $U$ is a simple $G$-module, then $\operatorname{Res}_{H}^{G}U$ does not have to be a simple $H$-module. Let $(U,\rho)$ be an $H$-module, the induction to $G$, denoted $(\operatorname{Ind}_{H}^{G}U,\operatorname{Ind}_{H}^{G}\rho)$ is the tensor product

$$\operatorname{Ind}_{H}^{G}U=\mathbb{C}G\underset{\mathbb{C}H}{\otimes}U$$

where the $G$ action is given by

$$g\bullet(s\otimes u)=(gs)\otimes u$$

where $s\in\mathbb{C}G$ and $u\in U$. However, we will also use the following more concrete description. Choose $S=\{s_{1},\ldots,s_{l}\}$ to be representatives of the left cosets of $H$ in $G$ (where $l=[G:H]$). Note that any element $g\in G$ can be written in a unique way as $g=s_{i}h$ where $s_{i}\in S$ and $h\in H$. Every element of $\operatorname{Ind}_{H}^{G}U$ is a formal sum of the form

$$\alpha_{1}(s_{1},u_{1})+\ldots+\alpha_{l}(s_{l},u_{l})$$

where $u_{i}\in U$ and $\alpha_{i}\in\mathbb{C}$. In other words, as a vector space $\operatorname{Ind}_{H}^{G}U$ is ${\displaystyle\bigoplus_{i=1}^{l}U}$, that is, $l$ copies of $U$. The action is defined on elements of the form $(s_{i},u)$ by

$$g\bullet(s_{i},u)=(s_{j},h\bullet u)$$

where $s_{j}$ and $h$ are unique such that $gs_{i}=s_{j}h$. The required action is given by extending linearly. Note that $\dim\operatorname{Ind}_{H}^{G}U=[G:H]\dim U$. It is important to mention that the modules $\operatorname{Ind}_{H}^{G}U$ and $\operatorname{Res}_{H}^{G}V$ depend not only on the groups $G$ and $H$ but also on the specific embedding of $H$ into $G$. Hence, we will have to give the specific embeddings when discussing these modules. Both induction and restriction are transitive and additive, that is, if $K\leq H\leq G$ then

$$\operatorname{Ind}_{H}^{G}\operatorname{Ind}_{K}^{H}U\cong\operatorname{Ind}_{K}^{G}U,\quad\operatorname{Ind}_{H}^{G}(U\oplus V)\cong\operatorname{Ind}_{H}^{G}U\oplus\operatorname{Ind}_{H}^{G}V$$

and

$$\operatorname{Res}_{K}^{H}\operatorname{Res}_{H}^{G}U\cong\operatorname{Res}_{K}^{G}U,\quad\operatorname{Res}_{H}^{G}(U\oplus V)\cong\operatorname{Res}_{H}^{G}U\oplus\operatorname{Res}_{H}^{G}V.$$

For the restriction this is a trivial statement, and for the induction the proof is [3, Propositions 1.1.10 and 1.1.11]. An important fact that relates induction to restriction is the following one (for a proof, see [3, Corollary 1.1.20 ]).

Using characters, Frobenius reciprocity can be written as the following equality

$$\langle\operatorname{Ind}_{H}^{G}V,U\rangle_{G}=\langle V,\operatorname{Res}_{H}^{G}U\rangle_{H}.$$

Let $U$ be a $G$-module. Consider the swap transformation $S:U\otimes U\to U\otimes U$ defined on simple tensors by $S(u_{1}\otimes u_{2})=u_{2}\otimes u_{1}$. We define the symmetric square $\operatorname{Sym}^{2}U$ and the alternating square $\operatorname{Alt}^{2}U$ as the following submodules of the tensor product $U\otimes U$:

	$\displaystyle\operatorname{Sym}^{2}U$	$\displaystyle=\{v\in U\otimes U\mid S(v)=v\},$
	$\displaystyle\operatorname{Alt}^{2}U$	$\displaystyle=\{v\in U\otimes U\mid S(v)=-v\}.$

As $G$-modules, $U\otimes U\cong\operatorname{Sym}^{2}U\oplus\operatorname{Alt}^{2}U$. The characters of these modules are given by:

$$(\operatorname{Sym}^{2}U)(g)=\frac{1}{2}\left(U(g)^{2}+U(g^{2})\right)\quad\text{and}\quad(\operatorname{Alt}^{2}U)(g)=\frac{1}{2}\left(U(g)^{2}-U(g^{2})\right).$$

We will recall some elementary facts regarding the representation theory of the symmetric group. More details can be found in [9, 21]. Recall that an integer composition of $n$ is a tuple $\lambda=[\lambda_{1},\ldots,\lambda_{k}]$ of non-negative integers such that $\lambda_{1}+\cdots+\lambda_{k}=n$ while an integer partition of $n$ (denoted $\lambda\vdash n$) is an integer composition such that $\lambda_{1}\geq\lambda_{2}\geq\cdots\geq\lambda_{k}>0$. From now on, when dealing with a partition $\lambda$ we will write its elements in superscript $\lambda=[\lambda^{1},\ldots,\lambda^{k}]$ because we want to reserve the subscript for multipartitions. Note that $0$ has one partition, namely the empty partition, denoted by $\emptyset$. We can associate to any partition $\lambda$ a graphical description called a Young diagram, which is a table with $\lambda^{i}$ boxes in its $i$-th row. For instance, the Young diagram associated to the partition $[3,3,2,1]$ of $9$ is:

We will identify the two notions and regard integer partition and Young diagram as synonyms. It is well-known that simple modules of $S_{n}$ are indexed by integer partitions of $n$. We denote the simple module associated to the partition $\lambda$ (also called its Specht module) by $S^{\lambda}$. An explicit description of $S^{\lambda}$ can be found in [21, Section 2.3]. It will be often convenient to draw the diagram $\lambda$ instead of writing $S^{\lambda}$. For instance we may write

$$\hbox{\vtop{\halign{&\opttoksa@YT={\font@YT}\getcolor@YT{\save@YT{\opttoksb@YT}}\nil@YT\getcolor@YT{\startbox@@YT\the\opttoksa@YT\the\opttoksb@YT}#\endbox@YT\cr\lower 0.39993pt\vbox{\kern 0.19997pt\hbox{\kern 0.39993pt\vbox to15.39995pt{\vss\hbox to15.00002pt{\hss$$\hss}\vss}\kern-15.39995pt\vrule height=15.39995pt,width=0.39993pt\kern 15.00002pt\vrule height=15.39995pt,width=0.39993pt}\kern-0.19997pt\kern-15.39995pt\hrule width=15.79988pt,height=0.39993pt\kern 15.00002pt\hrule width=15.79988pt,height=0.39993pt}&\lower 0.39993pt\vbox{\kern 0.19997pt\hbox{\kern 0.39993pt\vbox to15.39995pt{\vss\hbox to15.00002pt{\hss$$\hss}\vss}\kern-15.39995pt\vrule height=15.39995pt,width=0.39993pt\kern 15.00002pt\vrule height=15.39995pt,width=0.39993pt}\kern-0.19997pt\kern-15.39995pt\hrule width=15.79988pt,height=0.39993pt\kern 15.00002pt\hrule width=15.79988pt,height=0.39993pt}&\lower 0.39993pt\vbox{\kern 0.19997pt\hbox{\kern 0.39993pt\vbox to15.39995pt{\vss\hbox to15.00002pt{\hss$$\hss}\vss}\kern-15.39995pt\vrule height=15.39995pt,width=0.39993pt\kern 15.00002pt\vrule height=15.39995pt,width=0.39993pt}\kern-0.19997pt\kern-15.39995pt\hrule width=15.79988pt,height=0.39993pt\kern 15.00002pt\hrule width=15.79988pt,height=0.39993pt}\cr}}\kern 31.99976pt}\oplus\hbox{\vtop{\halign{&\opttoksa@YT={\font@YT}\getcolor@YT{\save@YT{\opttoksb@YT}}\nil@YT\getcolor@YT{\startbox@@YT\the\opttoksa@YT\the\opttoksb@YT}#\endbox@YT\cr\lower 0.39993pt\vbox{\kern 0.19997pt\hbox{\kern 0.39993pt\vbox to15.39995pt{\vss\hbox to15.00002pt{\hss$$\hss}\vss}\kern-15.39995pt\vrule height=15.39995pt,width=0.39993pt\kern 15.00002pt\vrule height=15.39995pt,width=0.39993pt}\kern-0.19997pt\kern-15.39995pt\hrule width=15.79988pt,height=0.39993pt\kern 15.00002pt\hrule width=15.79988pt,height=0.39993pt}&\lower 0.39993pt\vbox{\kern 0.19997pt\hbox{\kern 0.39993pt\vbox to15.39995pt{\vss\hbox to15.00002pt{\hss$$\hss}\vss}\kern-15.39995pt\vrule height=15.39995pt,width=0.39993pt\kern 15.00002pt\vrule height=15.39995pt,width=0.39993pt}\kern-0.19997pt\kern-15.39995pt\hrule width=15.79988pt,height=0.39993pt\kern 15.00002pt\hrule width=15.79988pt,height=0.39993pt}\cr\lower 0.39993pt\vbox{\kern 0.19997pt\hbox{\kern 0.39993pt\vbox to15.39995pt{\vss\hbox to15.00002pt{\hss$$\hss}\vss}\kern-15.39995pt\vrule height=15.39995pt,width=0.39993pt\kern 15.00002pt\vrule height=15.39995pt,width=0.39993pt}\kern-0.19997pt\kern-15.39995pt\hrule width=15.79988pt,height=0.39993pt\kern 15.00002pt\hrule width=15.79988pt,height=0.39993pt}\cr}}\kern 31.99976pt}$$

instead of: $S^{\lambda}\oplus S^{\delta}$ for partitions $\lambda=[3]$ and $\delta=[2,1]$.

Let ${\bf n}=[n_{1},\ldots,n_{l}]$ be an integer composition of $n$. A tuple $\Lambda=(\lambda_{1},\ldots,\lambda_{l})$ such that $\lambda_{i}\vdash n_{i}$ for every $i$ is called a multipartition of $n$ with $l$ components. We also call it a multipartition of the composition ${\bf n}$ and denote this by $\Lambda\Vdash{\bf n}$. We define a multi-Young diagram to be a tuple of Young diagrams. As we identify partitions with Young diagrams, we also identify multipartitions with multi-Young diagrams.

Let $G$ be a finite group with $l$ conjugacy classes. It is well-known ([3, Theorem 2.6.1]) that multi-Young diagrams with $n$ boxes and $l$ components index the simple modules of the wreath product $G\wr S_{n}$. If $\Lambda\Vdash{\bf n}$ is a multi-Young diagram, then we denote by $S^{\Lambda}$ its associated $G\wr S_{n}$-module.

Let $S_{2}=\{\operatorname{id},s\}$ be the symmetric group of order $2$ where $s$ will always be the swap permutation. We denote by $\operatorname{tr}_{2}$ the trivial module of $S_{2}$ and by $\operatorname{sgn}_{2}$ the sign module of $S_{2}$. The group $G\times S_{2}$ can be embedded in $G\wr S_{2}$ by $\varphi(g,\sigma)=((g,g),\sigma)$. In this section, we study the restriction $\operatorname{Res}_{G\times S_{2}}^{G\wr S_{2}}$. Every simple module of $G\times S_{2}$ is of the form $U_{r}\boxtimes\operatorname{tr}_{2}$ or $U_{r}\boxtimes\operatorname{sgn}_{2}$ where $U_{r}\in\operatorname{IRep}G$ ($r\in\{1,\ldots l\}$). The description of simple modules of $G\wr S_{2}$ is more complicated, but well-known (see [3, Chapter 2] for the description of simple modules of $G\wr S_{n}$ in general). If $U_{i},U_{j}\in\operatorname{IRep}G$ are two non-isomorphic simple modules of $G$ then $U_{i}\boxtimes U_{j}$ is a $G\times G$ simple module. Clearly, $G\times G$ embeds in $G\wr S_{2}$ by $\varphi(g_{1},g_{2})=((g_{1},g_{2}),\operatorname{id})$. One type of simple $G\wr S_{2}$ module is obtained by induction $W_{i,j}=\operatorname{Ind}_{G\times G}^{G\wr S_{2}}(U_{i}\boxtimes U_{j})$. This gives us $\binom{l}{2}$ simple modules. Note that $\dim W_{i,j}=2\dim U_{i}\dim U_{j}$. The module $W_{i,j}$ corresponds to the multipartition $\Lambda$ with $\lambda_{i}=\lambda_{j}=[1]$ and $\lambda_{k}=\emptyset$ for $k\neq i,j$. Given $U_{i}\in\operatorname{IRep}G$, another type of simple module is the tensor product $U_{i}\boxtimes U_{i}$ where the action is

$$((g_{1},g_{2}),\sigma)\bullet(u_{1}\boxtimes u_{2})=(g_{\sigma(1)}\bullet u_{\sigma(1)})\boxtimes(g_{\sigma(2)}\bullet u_{\sigma(2)}).$$

We denote this simple module by $W_{i}^{+}$. This gives another $l$ simple modules. The module $W_{i}^{+}$ corresponds to the multipartition $\Lambda$ with $\lambda_{i}=[2]$ and $\lambda_{k}=\emptyset$ for $k\neq i$. Finally, we denote by $\operatorname{Inf}(\operatorname{sgn}_{2})$ the inflation of $\operatorname{sgn}_{2}$ to a $G\wr S_{2}$ module. This means that $((g_{1},g_{2}),\sigma)$ acts like $\sigma$. The last $l$ simple modules of $G\wr S_{2}$ are obtained by the tensor product $W_{i}^{+}\otimes\operatorname{Inf}(\operatorname{sgn}_{2})$ and we denote them by $W_{i}^{-}$. The module $W_{i}^{-}$ corresponds to the multipartition $\Lambda$ with $\lambda_{i}=[1,1]$ and $\lambda_{k}=\emptyset$ for $k\neq i$. In total, we have $\frac{l^{2}+3l}{2}$ simple modules for $G\wr S_{2}$.

The characters of these modules are well-known and essential for computing the desired restriction. Recall that we use the same notation for a module and its character. If we take $((1_{G},1_{G}),\operatorname{id})$ and $((1_{G},1_{G}),s)$ as representatives of the cosets of $G\times G$ in $G\wr S_{2}$, it is easy to see that $((g,g),s)$ swaps the cosets and $((g,g),\operatorname{id})$ fixes them. The character of the induction is summation of the base character over the fixed cosets (see [21, Section 1.12]). Therefore, the character $W_{i,j}$ for an element $((g,g),\sigma)\in G\wr S_{2}$ is given by:

$$W_{i,j}((g,g),\sigma)=\begin{cases}2U_{i}(g)U_{j}(g)&\text{if }\sigma=\operatorname{id},\\ 0&\text{if }\sigma=s.\end{cases}$$

For the modules $W_{i}^{+}$ and $W_{i}^{-}$, the character values on the elements $((g,g),\sigma)$ are given as follows (see [9, 4.3.10 (vi), p. 150]):

$$W_{i}^{+}((g,g),\sigma)=\begin{cases}U_{i}(g)^{2}&\text{if }\sigma=\operatorname{id},\\ U_{i}(g^{2})&\text{if }\sigma=s,\end{cases}$$

and

$$W_{i}^{-}((g,g),\sigma)=\begin{cases}U_{i}(g)^{2}&\text{if }\sigma=\operatorname{id},\\ -U_{i}(g^{2})&\text{if }\sigma=s.\end{cases}$$

Let $G$ be a group with $l$ conjugacy classes. The group $(G\wr S_{k})\times(G\wr S_{r})$ is naturally embedded in $G\wr S_{k+r}$. If $f_{1}:\{1,\ldots,k\}\to G$ and $f_{2}:\{1,\ldots,r\}\to G$ we define $f:\{1,\ldots,k+r\}\to G$ by

$$f(i)=\begin{cases}f_{1}(i)&i\leq k\\ f_{2}(i-k)&i>k\end{cases}.$$

Then, the natural embedding $\varphi:G\wr S_{k}\times G\wr S_{r}\to G\wr S_{k+r}$ is defined by $\varphi((f_{1},\sigma_{1}),(f_{2},\sigma_{2}))=(f,\sigma_{1}\sigma_{2})$ where $\sigma_{1}$ ($\sigma_{2}$) can be regarded as an element of $S_{k+r}$ that fixes $\{k+1,\ldots,k+r\}$ ($\{1,\ldots,k\}$). The branching rules for describing the induction from $G\wr S_{k}\times G\wr S_{r}$ to $G\wr S_{k+r}$ are known (see [8, Theorem 4.7] or [27, Theorem 4.5]), but we will give here only the cases for $r=1,2$ as this is what we need in this paper.