2 Category of FRBSU Monoidal Categories and Crossed Modules

Assume that \mathcal{A} is a k linear abelian category and X, Y are nonzero simple objects. Let f:~X\rightarrow Y is a nonzero morphism in \mathcal{A}. Ker(f)\cong 0 since X is simple and f\neq 0, so that morphism is a monomorphism. As a result, X\cong Y that is a contradiction, hence f=0.

We define the tensor product as X\otimes^{op} Y=Y\otimes X and associativity constraint a^{op} as a family of natural isomorphisms

in \mathcal{A}^{op} for all objects X, Y and Z in \mathcal{A}. This is same as the family of natural isomorphisms

in \mathcal{A}^{op} which can be obtained by inverting the arrows

in \mathcal{A} and as a result, we get a^{op}_{XYZ}=a^{-1}_{ZYX} for all objects X, Y and Z in \mathcal{A}.

Here, I^{op}=I. l^{op} is a left unit constraint which is a family of natural isomorphisms

in \mathcal{A}^{op} which is same as

for all objects X in \mathcal{A}. So, we take l^{op}_X=r_X and l^{op}=r in \mathcal{A}. Similarly, we can take r^{op}=l.

We define a monoidal functor \mathcal{F}:~\mathcal{A} \rightarrow \mathcal{A}^{rev} by sending an object A in \mathcal{A} to itself, \gamma as a family of natural isomorphisms \gamma_{XY}:~X\otimes^{rev} Y=Y\otimes X \rightarrow X\otimes Y for all objects X, Y in \mathcal{A} and also \varphi=id_I. We define \gamma_{XY}=c^{rev}_{XY}. Then, we need to show that the following diagram commutes in \mathcal{A}^{rev}.

This diagram is same as the following diagram in \mathcal{A}.

The first and third squares commute by definition and the middle one commutes by using naturality of the braiding in the following diagram.

As a result, that diagram commutes.

a_{XYZ}\circ c_{Z(XY)}\circ (id_Z\otimes c_{YX})=a_{XYZ}\circ a^{-1}_{XYZ}\circ (id_X\otimes c_{ZY})\circ a_{XZY}\circ (c\otimes id_Y)\circ a^{-1}_{ZXY}\circ (id_Z\otimes c_{YX})=(id_X\otimes c_{ZY})\circ a_{XZY}\circ (c\otimes id_Y)\circ a^{-1}_{ZXY}\circ (id_Z\otimes c_{YX}).

c_{(YZ)X}\circ (c_{ZY}\otimes id_X)\circ a^{-1}_{ZYX}=a_{XYZ}\circ (c_{YX}\otimes id_Z)\circ a^{-1}_{YXZ}\circ (id_Y\otimes c_{ZX})\circ a_{YZX}\circ (c_{ZY}\otimes id_X)\circ a^{-1}_{ZYX}.

These two equations are same by the commutativity of Diagram ?. As a result, Diagram Proof commutes. The commutativity of other diagrams are easy to show, also it is a braided monoidal functor. The reader can show that the conditions for equivalence are satisfied.

See [baki] for the proof.

If X is an object in a rigid monoidal category \mathcal{A}, then X has both a left and a right dual objects ^+X and X^+ which are unique up to a unique isomorphism by Lemma 1.3.1 such that the following compositions are the identities by definition where ev_{rX}:~X_+\otimes X \rightarrow I, coev_{rX}:~I \rightarrow X\otimes X^+, ev_{lX}:~X\otimes {^+X}\rightarrow I and coev_{lX}:~I \rightarrow {^+X}\otimes X are morphisms in \mathcal{A}.

We know that ev_{rX}\otimes id_{X^+} \circ id_{X^+}\otimes coev_{rX}=id_{X^+} in \mathcal{A} by the first composition, so (ev_{rX}\otimes id_{X^+} \circ id_{X^+}\otimes coev_{rX})^{op}=(id_{X^+})^{op}=id_{X^+}. This implies that

As a result, the following composition is the identity of X^+ in \mathcal{A}^{op}

which is same as the following one

Also, we get an identity of X by the composition

which is same as the following one

These two identities show that X^+ is right dual for X in \mathcal{A}^{op}. By using same technique, we may show that ^+X is left dual for X in \mathcal{A}^{op}. Those are unique objects in \mathcal{A} by Lemma 1.3.1, so they are also unique in \mathcal{A}^{op} as objects. This shows that \mathcal{A}^{op} is a rigid category.

The composition \xymatrix{I\ar[r] & I^+\ar[r] & I} is the identity by ?. Also, the other condition is satisfied. It is easy to see that I satisfies the required conditions, too. Hence, I=I^+ by uniqueness of a right dual.

If the above maps are isomorphisms, then we get X^+\otimes X\cong I\cong X\otimes X^+, so X^+ is the required object in the definition of an invertible object. Thus, X is invertible. Similarly, we see that X^+ is invertible.

Conversely, if X is invertible, then there exists an object Z such that X\otimes Z\cong Z\otimes X\cong I, so we can use Z as a right dual, hence Z\cong X^+ by uniqueness of a right dual. Then the above maps are isomorphisms. With the same idea, we may consider Z\cong {^+X} by the isomorphism and we reverse the arrows if required and see Z is a left dual. As a result, X^+\cong {^+X}.

Now, assume that X and Y are two invertible objects in the category. Then, X^+\otimes X\cong I\cong X\otimes X^+ and Y^+\otimes Y\cong I\cong Y\otimes Y^+. So, we get Y^+\otimes X^+\otimes X\otimes Y\cong I\cong X\otimes Y\otimes Y^+\otimes X^+. This shows that Y^+\otimes X^+ is the inverse of X\otimes Y.

(Z,~\gamma_Z)\otimes (W,~\gamma_W)=(Z\otimes W,~(\gamma_Z \otimes id_W)\circ (id_Z\otimes \gamma_W)) is a tensor product for all objects Z and W in \mathcal{A} and c_{(Z,~\gamma_Z)(W,~\gamma_W)}: (Z,~\gamma_Z)\otimes (W,~\gamma_W) \rightarrow (W,~\gamma_W)\otimes (Z,~\gamma_Z) is a braiding which is same as

We define the action as

for all objects M\in \mathcal{M}^{op} and for all objects X\in \mathcal{A} where M\otimes_{r\mathcal{M}^{op}} X=X^+\otimes_{l\mathcal{M}} M. \otimes_{r\mathcal{M}^{op}} is an exact bifunctor since \otimes_{l\mathcal{M}} is an exact bifunctor.

Can we find an associativity constraint a_{r\mathcal{M}^{op}} consisting of associativity isomorphisms a_{MXY}:~M\otimes_{r\mathcal{M}^{op}} (X\otimes_{\mathcal{A}} Y) \rightarrow (M\otimes_{r\mathcal{M}^{op}} X)\otimes_{r\mathcal{M}^{op}} Y for all objects X, Y in \mathcal{A}, M in \mathcal{M} and a right unit constraint r_{\mathcal{M}^{op}} which is a family of right unit isomorphisms r_M:~M\otimes_{r\mathcal{M}^{op}} I\rightarrow M such that for any X, Y, Z in \mathcal{A}, M in \mathcal{M}, the Diagram ? and Diagram ? commute?

We know that there is an isomorphism

since \mathcal{M} is a left \mathcal{A} module category, so, we can take a_{MXY}=a_{Y^+X^+M}.

M\otimes_{r\mathcal{M}^{op}} I=I^+\otimes_{l\mathcal{M}} M and there is an isomorphism l_M:~I^+\otimes_{l\mathcal{M}} M\rightarrow M, so we get the family of right unit constraints r_M=l_M since I^+=I. The reader can show that the Diagram ? and Diagram ? commute.

Assume that \mathcal{M} is a module category over \mathcal{A} given as above. Let A be an object in \mathcal{A} and P be projective in \mathcal{M}. For all epimorhisms f:~X\rightarrow Y and morphisms g:~A\otimes P\rightarrow Y in \mathcal{M}, can we find a morphism k:~A\otimes P\rightarrow X such that g=k\circ f.

We find an epimorphism f':~{^+A}\otimes X\rightarrow {^+A}\otimes Y corresponding to f and a morphism g':~P\rightarrow {^+A}\otimes Y corresponding to g. Since P is projective, then we get a morphism k':~P\rightarrow {^+A}\otimes X such that g'=f'\circ k'. After that, we get a morphism k:~A\otimes P\rightarrow X corresponding to k' such that f\circ k=g. As a result, A\otimes P is a projective object in \mathcal{M}.

We want to show that I is projective in \mathcal{A}. Assume that we are given an epimorphism f:~A\rightarrow B and a map g:~I\rightarrow B. Can we find a map h:~I\rightarrow A such that f\circ h=g?

A=\oplus_i A_i, I=\oplus_j I_j and B=\oplus_k B_k for simple objects A_i, I_j and B_k for all i, j, and k. We can decompose f and g as f=~\oplus_{ik} f_{ik} and g=~\oplus_{jk} g_{jk} where f_{ik}:~A_i\rightarrow B_k, g_{jk}:~I_j\rightarrow B_k are morphisms in \mathcal{A}. By Proposition 1.0.2, we get f_{ik}=g_{jk}=0, so any morphism I_j\rightarrow A_i works, then we take h as the direct sum of those morphisms and the result follows afterwards.

A=I\otimes A for all objects A in \mathcal{A} by left unit associativity. \mathcal{A} is an exact left module category over itself by Lemma 1.6.2. I is projective, thus I\otimes A is projective by exactness. Hence, every object is projective in \mathcal{A}.

Assume that \mathcal{M} is a semisimple module category over a finite monoidal category \mathcal{A}. Any object in a semisimple category is projective, so \mathcal{M} is exact.

Assume that \mathcal{A} is given as above and \mathcal{M} is an exact left module category over \mathcal{A}. Then, we find a projective object P in \mathcal{A} with an epimorphism P\rightarrow I since \mathcal{A} is a finite category. Hence, for all objects M in \mathcal{M}, we get an epimorphism P\otimes M\rightarrow I\otimes M\cong M in \mathcal{M}. \mathcal{M} is exact, so P\otimes M is projective by exactness. As a result, we see that \mathcal{M} has enough projectives and there exists a projective cover for every simple object in \mathcal{M}.

Assume that \mathcal{M} is an (\mathcal{A}-\mathcal{B}) bimodule category. In that situation \mathcal{M} is a left \mathcal{A} module category and a right \mathcal{B} module category, so \mathcal{M}^{op} is a left \mathcal{B} module category and right \mathcal{A} module category by Proposition 1.5.1.

We have an associativity constraint a consisting of a family of isomorphisms

natural in X\in \mathcal{A}, Y\in \mathcal{B}, M\in \mathcal{M} as in ? which satisfies the commutativity of the required diagrams for all X,~Y\in \mathcal{A}, Z,~W\in \mathcal{B} and M\in \mathcal{M} to be an (\mathcal{A}-\mathcal{B}) bimodule. We need to define a^{op} consisting of associativity constraints

? is obtained by reversing the morphism Y\otimes (M\otimes X)\rightarrow (Y\otimes M)\otimes X in \mathcal{M} which is same as a_{YMX}. We can prove the compatibility conditions without no difficulty.

Assume that \mathcal{A} is a finite, rigid monoidal category. We can take \mathcal{M}=\mathcal{A}. We have a bifunctor \mathcal{F}:~\mathcal{A} \times \mathcal{A} \rightarrow \mathcal{A} taking (X,~Y) to X\otimes Y. \mathcal{F} is exact in each variable by Remark 1.3.1.

We can use the associativity constraint a and left unit constraint l in the definition of a monoidal category. We can see that these satisfy the commutativity of the required diagrams to be a left \mathcal{A} module category. Similarly, it is a right \mathcal{A} module category with the same associativity constraint and right unit constraint r by the definition of a monoidal category. Also, we use same associativity constraint and a middle associativity constraint. These satisfy the compatibility conditions.

Every (A-B) bimodule M is a left A module and a right B module in \mathcal{A} that satisfies the compatible conditions, so M is an object in A\mathcal{A} and an object in \mathcal{A}B. This means that A\mathcal{A} B is a subcategory of A\mathcal{A} and a subcategory of \mathcal{A} B. A\mathcal{A} is a right \mathcal{A} module category and \mathcal{A} B is left \mathcal{A} module category. As a result, A\mathcal{A} B is both a left \mathcal{A} and right \mathcal{A} module category. We need to define an associativity constraint a consisting of associativity isomorphisms as in ? for all objects X, Y in \mathcal{A}, M in A\mathcal{A} B that satisfies the required conditions.

We have two actions \mathcal{A} \times A\mathcal{A} B\rightarrow A\mathcal{A} B taking (X,~M) to X\otimes_{l(\mathcal{A} B)} M and A\mathcal{A} B\times \mathcal{A} \rightarrow A\mathcal{A} B taking (M,~Y) to M\otimes_{r(A\mathcal{A})} Y.

We know that X\otimes_{l(\mathcal{A} B)} M is a right B module and we may show that it is a left A module which means that it is an (A-B) bimodule. Similarly, M\otimes_{r(A\mathcal{A})} Y is an (A-B) bimodule.

X\otimes_{l(\mathcal{A}B)}(M\otimes_{r(A\mathcal{A})} Y)\rightarrow (X\otimes_{l(\mathcal{A}B)} M)\otimes_{r(A\mathcal{A})} Y is an isomorphism since M is an object in \mathcal{A} and the above actions are exactly same as the tensor product in \mathcal{A}, we can use the associativity constraint in \mathcal{A} as a middle associativity constraint. It is obvious, this gives the commutativity of the diagrams in the definition.

We must show that

is an isomorphism and \mathcal{F} is essentially surjective for all (A-B) bimodules M and N.

We send each morphism \xymatrix{M\ar[r]^a & N} in the category of (A-B) bimodules to a natural transformation \xymatrix{-\otimes_A M\ar[r]^{f(a)} & -\otimes_A N} in the category of module functors Hom^{re}_{\mathcal{A}}(\mathcal{A}A,~\mathcal{A}B). Here, -\otimes_A M,~-\otimes_A N:~\mathcal{A}A \rightarrow \mathcal{A}B are \mathcal{A} module functors for all (A-B) bimodules M and N in \mathcal{A}.

To show that they are module functors, we show that K\otimes_A N=K\otimes N is a right B module in the category \mathcal{A} at first for all right A modules K in \mathcal{A}. We define the action as a_{(K\otimes N)}=(K\otimes N)\times B\rightarrow (K\otimes N) by sending the pair ((k\otimes n),~b) to k\otimes (n\otimes b) for all elements k\in K, n\in N, b\in B. As a result, the following diagrams commute since N is a right B module.

We want to show that f is an injection. Let a,~b:~M\rightarrow N be two (A-B) bimodule homomorphisms and f(a)=f(b).

f(a),~f(b):~-\otimes_A M\rightarrow -\otimes_A N are natural transformations and for all objects K in \mathcal{A}A, we get f(a)(K)=f(b)(K):~K\otimes_A M\rightarrow K\otimes_A N.

For all elements k in K and m in M, we have k\otimes a(m)=k\otimes b(m), so k\otimes (a(m)-b(m))=0 and a(m)=b(m). This says that a=b. Surjectivity is clear. As a result f is an isomorphism.

For all right exact \mathcal{A} module functors \mathcal{G}:~\mathcal{A}A \rightarrow \mathcal{A}B, can we find an (A-B) bimodule M such that \mathfrak{F}(M) \cong \mathcal{G}? We take M=\mathcal{G}(A). It is (A-B) bimodule.

Now we want to show the commutativity of the required diagram for the natural isomorphism. For all morphism \alpha:~T\rightarrow S in \mathcal{A}A, we get the following commutative diagram.

T\otimes_A \mathcal{G}(A)\cong T\otimes_A A\cong T\cong \mathcal{G}(T) since \mathcal{G}(A)\cong A and \mathcal{G}(T)\cong T. Similarly, \theta is an isomorphism, so the diagram commutes.

\mathcal{M} \simeq \mathcal{A}A and \mathcal{N} \simeq \mathcal{A}B for some algebras A and B in \mathcal{A}. So, the category Hom^{re}_{\mathcal{A}}(\mathcal{M},~\mathcal{N}) is equivalent to the category A\mathcal{A} B and A\mathcal{A} B is finite.

For given two module categories \mathcal{F},~\mathcal{G}:~\mathcal{M} \rightarrow \mathcal{M}, we define their tensor product as the composition \mathcal{G} \circ \mathcal{F}:~\mathcal{M} \rightarrow \mathcal{M} and the unit functor as the identity of \mathcal{M} which is \mathcal{I}=id_{\mathcal{M}}:~\mathcal{M} \rightarrow \mathcal{M}. We get an associativity constraint a which is a family of associative isomorphisms {a_{\mathcal{F} \mathcal{G} \mathcal{H}}:~(\mathcal{F} \circ \mathcal{G}) \circ \mathcal{H} \rightarrow \mathcal{F} \circ (\mathcal{G} \circ \mathcal{H})}, left and right unit constraints satisfying the required compatibility conditions.

The right and left duals are the right and left adjoint functors.

We define the action as Hom^{re}_{\mathcal{A}}(\mathcal{M},~\mathcal{M})\times Hom^{re}_{\mathcal{A}}(\mathcal{M},~\mathcal{N})\rightarrow Hom^{re}_{\mathcal{A}}(\mathcal{M},~\mathcal{N}) with (\mathcal{F},~\mathcal{G})\mapsto \mathcal{G} \circ \mathcal{F}.

Assume that we have an exact sequence 0\rightarrow A\rightarrow B\rightarrow C\rightarrow 0 for all objects A, B and C in \mathcal{A}. Then, the sequence A\otimes M\rightarrow B\otimes M\rightarrow C\otimes M\rightarrow 0 is exact since -\otimes M is an exact functor by 1.5.1. So, the sequence

is exact since Hom_{\mathcal{M}}(-,~N) is left exact controvariant functor by Example ?. This proves the left exactness.

Hom_{\mathcal{M}}(X\otimes M,~N)\cong Hom_{\mathcal{A}}(X,~\underline{Hom}_{\mathcal{M}}(M,~N)) for all objects X in \mathcal{A} by existence of representing object. We know that this functor is left exact controvariant functor by Lemma 2.0.1. We want to show that it is right exact. For this, we need to show that the controvariant functor Hom_{\mathcal{A}}(-,~\underline{Hom}_{\mathcal{M}}(M,~N)):~\mathcal{A} \rightarrow Set is right exact which means that the covariant functor Hom_{\mathcal{A}^{op}}(\underline{Hom}_{\mathcal{M}}(M,~N),~-):~\mathcal{A}^{op} \rightarrow Set is right exact.

Assume that \xymatrix{ 0\ar[r] & X\ar[r]^f & Y\ar[r]^g & Z\ar[r] & 0} is an exact sequence in \mathcal{A}^{op}. We want to show that the sequence

\xymatrix{0\ar[r] & Hom_{\mathcal{A}^{op}}(\underline{Hom}_{\mathcal{M}}(M,~N),~X)\ar[r]^F & Hom_{\mathcal{A}^{op}}(\underline{Hom}_{\mathcal{M}}(M,~N),~Y)\\ & \ar[r]^G & Hom_{\mathcal{A}^{op}}(\underline{Hom}_{\mathcal{M}}(M,~N),~Z)\ar[r] & 0}

is exact in Set. We just need to show that G is an epimorphism since that sequence is left exact. It is obvious since the internal hom is projective.

For all morphisms M\rightarrow X\otimes N in \mathcal{M}, we find a morphism

by using rigidity of \mathcal{A}. Here, we use the evaluation map ev_{rX}:~X^+\otimes X\rightarrow I. So,

For all morphisms X^+\rightarrow \underline{Hom}_{\mathcal{M}}(M,~N), we get a morphism

by using the coevaluation map. As a result, we get an isomorphism

[os] We have

Hom_{\mathcal{A}}(K,~\underline{Hom}_{\mathcal{M}}(M,~N)\otimes X^+) \cong Hom_{\mathcal{A}}(K\otimes X,~\underline{Hom}_{\mathcal{M}}(M,~N))\cong \\ Hom_{\mathcal{M}}((K\otimes X)\otimes M,~N)\cong Hom_{\mathcal{M}}(K\otimes (X\otimes M),~N)\cong Hom_{\mathcal{A}}(K,~\underline{Hom}_{\mathcal{M}}(X\otimes M,~N))

for all K in \mathcal{M}, so \underline{Hom}_{\mathcal{M}}(M,~N)\otimes X^+\cong \underline{Hom}_{\mathcal{M}}(X\otimes M,~N) canonically.

See [os].

We need to define a multiplication morphism

and a unit morphism u:~I\rightarrow \underline{Hom}_{\mathcal{M}}(M,~M) satisfying the required compatibility conditions.

[os] finds a multiplication morphism m as below.

id_{\underline{Hom}_{\mathcal{M}}(M,~M)} is in Hom_{\mathcal{A}}(\underline{Hom}_{\mathcal{M}}(M,~M),~\underline{Hom}_{\mathcal{M}}(M,~M)) since \underline{Hom}_{\mathcal{M}}(M,~M) is an object in \mathcal{A}.

by definition. So, we get a unique morphism f:~\underline{Hom}_{\mathcal{M}}(M,~M)\otimes M\rightarrow M corresponding to id_{\underline{Hom}_{\mathcal{M}}(M,~M)}. Using this morphism, we get a composition

This is same as the morphism \xymatrix{ (\underline{Hom}_{\mathcal{M}}(M,~M)\otimes \underline{Hom}_{\mathcal{M}}(M,~M))\otimes M\rightarrow M} since

This defines a multiplication morphism

\underline{Hom}_{\mathcal{M}}(M,~M) \otimes \underline{Hom}_{\mathcal{M}}(M,~M) \rightarrow \underline{Hom}_{\mathcal{M}}(M,~M) as shown in [os].

This multiplication is associative since \underline{Hom}_{\mathcal{M}}(M,~M) is an object in \mathcal{A} and we have the associativity constraint.

For the compatibility conditions, we need to show that the following diagrams commute.

We have the isomorphisms

m is a morphism in the first hom set and id is a morphism in the second one.

Also, we have a composition of hom sets

Hom_{\mathcal{A}}(A\otimes A\otimes A,~A\otimes A)\times Hom_{\mathcal{A}}(A\otimes A,~A)\rightarrow Hom_{\mathcal{A}}(A \otimes A\otimes A,~A)

for A=\underline{Hom}_{\mathcal{M}}(M,~M) taking (m\otimes id,~m) to m\circ (m\otimes id) and (id\otimes m,~m) to m\circ (id\otimes m). We may prove that m\circ (m\otimes id)=m\circ (id\otimes m) for the commutativity of the first diagram.

Now, we want to find a unit morphism u:~I\rightarrow \underline{Hom}_{\mathcal{M}}(M,~M) satisfying the commutativity of the required diagrams.

Hom_{\mathcal{M}}(M,~X\otimes N)\cong Hom_{\mathcal{A}}(I,~X\otimes \underline{Hom}_{\mathcal{M}}(M,~N)) for all objects X in \mathcal{A}. Taking X=I and M=N, we get an isomorphism Hom_{\mathcal{M}}(M,~M)\cong Hom_{\mathcal{A}}(I,~\underline{Hom}_{\mathcal{M}}(M,~M)). So, for the identity morphism id_M:~M\rightarrow M, we get a unique morphism u:~I\rightarrow \underline{Hom}_{\mathcal{M}}(M,~M).

We have an isomorphism

The diagram

corresponds to the diagram

And we get another diagram

Obviously, this diagram commutes. As a result, the first one commutes and we get the result. We follow the similar way for the right associativity constraint.

We need to define a multiplication morphism

satisfying the commutativity of the following diagrams.

So the identity morphism id_{\underline{Hom}_{\mathcal{M}}(M,~N)} corresponds to a morphism

and a corresponds to a composition

which is k\circ (id_{\underline{Hom}_{\mathcal{M}}(M,~N)}\otimes f) and f is the morphism \underline{Hom}_{\mathcal{M}}(M,~M) \otimes M\rightarrow M that corresponds to id_{\underline{Hom}_{\mathcal{M}}(M,~M)}.

It is easy to show that those diagrams commute.

f is a family of natural isomorphisms f_{XN}:~\underline{Hom}_{\mathcal{M}}(M,~X\otimes N) \rightarrow X\otimes \underline{Hom}_{\mathcal{M}}(M,~N) for all objects X in \mathcal{A}, N in \mathcal{M} such that for all X, Y in \mathcal{A}, N in \mathcal{M}, the following diagrams commute.

Exactness comes from 1.7.3 and this proves the proposition.

We need to show that

is an isomorphism for all objects N and K in \mathcal{M} and \mathcal{F} is essentially surjective for the equivalence. [os] proves the isomorphism for all objects N of the form X\otimes M and for all objects K in \mathcal{M} first, then the author proves it for all objects N and K in \mathcal{M} by using the exactness of the functor \underline{Hom}_{\mathcal{M}}(M,~-). After that, [os] shows that \underline{Hom}_{\mathcal{M}}(M,~-) is essentially surjective.

We may apply similar way to prove the proposition. In [os], the proposition is given whenever the category is semisimple, hence exact means semisimple in such a category. \mathcal{M} is indecomposable module category there instead of the assumption for \mathcal{M} in here.

To prove that Hom^{re}_{\mathcal{B}}(\mathcal{M}^{op},~\mathcal{N}) is an (\mathcal{A}-\mathcal{C}) bimodule category, we define the left action of \mathcal{A} on Hom^{re}_{\mathcal{B}}(\mathcal{M}^{op},~\mathcal{N}) by

for all objects A in \mathcal{A} and right exact module functors \mathcal{F}:~\mathcal{M}^{op} \rightarrow \mathcal{N}. Here, we have (A\otimes \mathcal{F})(M)=\mathcal{F}(M\otimes A) for all objects M in \mathcal{M}^{op}. \mathcal{M}^{op} is a right \mathcal{A} module category, so M\otimes A is an object in \mathcal{M}^{op}. We need to show that \mathfrak{F} is a biexact bifunctor, that is \mathfrak{F}(-,~\mathcal{F}) is exact for all right exact module functors \mathcal{F}:~\mathcal{M}^{op} \rightarrow \mathcal{N} which means that exact in the first variable and \mathfrak{F}(A,~-) is exact for all objects A in \mathcal{A} which means that exact in the second variable.

To prove that \mathfrak{F}(-,~\mathcal{F}) is an exact functor, we need to show that

is an exact sequence of natural transformations of right exact module functors from \mathcal{M}^{op} to \mathcal{N} whenever the sequence 0\rightarrow A\rightarrow B\rightarrow C\rightarrow 0 is exact. That sequence of natural transformations is a sequence of morphisms

for all objects M in \mathcal{M}^{op} which satisfies the compatibility conditions for all morphisms M\rightarrow N in \mathcal{M}^{op}. This sequence is same as the sequence

by the above action. The sequence 0\rightarrow M\otimes A\rightarrow M\otimes B\rightarrow M\otimes C\rightarrow 0 is exact since the action is an exact functor by definition of module category. As a result, ? is an exact sequence by Lemma 1.7.3.

Assume that 0\rightarrow \mathcal{F} \rightarrow \mathcal{G} \rightarrow \mathcal{H} \rightarrow 0 is an exact sequence of module functors in Hom^{re}_{\mathcal{B}}(\mathcal{M}^{op},~\mathcal{N}). It is clear that the sequence 0\rightarrow A\otimes \mathcal{F} \rightarrow A\otimes \mathcal{G} \rightarrow A\otimes \mathcal{H}\rightarrow 0 is exact. As a result, \mathfrak{F}(A,~-) is an exact functor.

How do we get an associativity constraint a consisting of a family of associativity isomorphism a_{AB\mathcal{F}}:~(A\otimes B)\otimes \mathcal{F} \rightarrow A\otimes (B\otimes \mathcal{F}) for all objects A, B in \mathcal{A}, right exact module functors \mathcal{F}:~\mathcal{M}^{op} \rightarrow \mathcal{N} and a unit constraint l consisting of a family of unit isomorphisms l_{\mathcal{F}}:~I\otimes \mathcal{F} \rightarrow \mathcal{F} for the unit object I in \mathcal{A}, module functor \mathcal{F}:~\mathcal{M}^{op} \rightarrow \mathcal{N} making the required diagrams commute.

For all objects A, B in \mathcal{A}, module functors \mathcal{F}:~\mathcal{M}^{op} \rightarrow \mathcal{N}, we get

\mathcal{M}^{op} is a right \mathcal{A} module category, so M\otimes (A\otimes B)\cong (M\otimes A)\otimes B). We get a short exact sequence 0\rightarrow M\otimes (A\otimes B)\rightarrow (M\otimes A)\otimes B\rightarrow 0. \mathcal{F} is exact by 1.7.3, so the sequence 0\rightarrow \mathcal{F}(M\otimes (A\otimes B))\rightarrow \mathcal{F}((M\otimes A)\otimes B)\rightarrow 0 is exact, hence we get an isomorphism \mathcal{F}(M\otimes (A\otimes B)) \cong \mathcal{F}((M\otimes A)\otimes B) and use this isomorphism as an associativity constraint. Similarly, we define l.

\mathfrak{G}:~Hom^{re}_{\mathcal{B}}(\mathcal{M}^{op},~\mathcal{N})\times \mathcal{C} \rightarrow Hom^{re}_{\mathcal{B}}(\mathcal{M}^{op},~\mathcal{N}) is the right action of \mathcal{C} on Hom^{re}_{\mathcal{B}}(\mathcal{M}^{op},~\mathcal{N}) with (\mathcal{F},~C)\rightarrow \mathcal{F} \otimes C for all objects C in \mathcal{C} and right exact module functors \mathcal{F}:~\mathcal{M}^{op} \rightarrow \mathcal{N}.

Here, \mathcal{F} \otimes C:~\mathcal{M}^{op} \rightarrow \mathcal{N} such that (\mathcal{F} \otimes C)(M)=\mathcal{F}(M) \otimes C for all objects M in \mathcal{M}^{op}. \mathcal{N} is right \mathcal{C} module category, so \mathcal{F}(M) \otimes C is an object in \mathcal{N}. We can show that \mathfrak{G} is a biexact bifunctor in a similar way. After that we find a right associativity constraint and a right unit constraint satisfying the required conditions.

Finally, we find a middle associativity constraint satisfying two commutative diagrams, hence the lemma is proved.

First, we want to show that A\otimes - is a right exact \mathcal{A} module functor for all objects A in \mathcal{A}. The right exactness is clear by definition of module category.

Hom sets are vector spaces, because \mathcal{M} is finite by Lemma 2.0.14. That functor is k linear since all functions f:~Hom_{\mathcal{M}}(A,~B)\rightarrow Hom_{\mathcal{M}}(A\otimes -,~B\otimes -) are linear maps for all objects A and B in \mathcal{M}.

For the functors \mathcal{F}_1,~\mathcal{F}_2:~A\rightarrow B, k_1,~k_2\in k, we get f(k_1\mathcal{F}_1+k_2\mathcal{F}_2)=\mathcal{F} where \xymatrix{((k_1+k_2)A)\otimes -\ar[r]^{\mathcal{F}} & ((k_1+k_2)B)\otimes -} is a natural transformation. This natural transformation is same as

which is same as k_1f(\mathcal{F}_1)+k_2f(\mathcal{F}_2).

For all objects B in \mathcal{A} and for all objects M in \mathcal{M}, we obtain

by using associativity constraint and the braiding as in the following diagram.

It is easy to see that the compatibility conditions are satisfied, hence it is a module functor.

Now, we want to prove that the assignment \mathcal{F}_{\mathcal{M}}:~\mathcal{A} \rightarrow Hom^{re}_{\mathcal{A}}(\mathcal{M},~\mathcal{M}) taking any object A in \mathcal{A} to a right exact module functor \mathcal{F}(A):~\mathcal{M} \rightarrow \mathcal{M} defined by \mathcal{F}(A)(M)=A\otimes M for all objects M in \mathcal{M} is a monoidal functor.

There exists a natural transformation \gamma_{AB}:~\mathcal{F}_{\mathcal{M}}(A) \circ \mathcal{F}_{\mathcal{M}}(B) \rightarrow \mathcal{F}_{\mathcal{M}}(A\otimes B).

(\mathcal{F}_{\mathcal{M}}(A)\circ \mathcal{F}_{\mathcal{M}}(B))(M)=\mathcal{F}_{\mathcal{M}}(A)(B\otimes M)=A\otimes (B\otimes M) and \mathcal{F}_{\mathcal{M}}(A\otimes B)(M)=(A\otimes B)\otimes M for all M in \mathcal{M}.

We have an isomorphism a^{-1}:~A\otimes (B\otimes M)\rightarrow (A\otimes B)\otimes M. This says that

is an isomorphism for all M in \mathcal{M}, so \gamma_{AB} is a natural isomorphism.

Also I\rightarrow \mathcal{F}_{\mathcal{M}}(I)=I\otimes M is an isomorphism by left unit constraint. We may show that the diagrams commute. So, we get the result.

If \mathcal{M} is invertible, then \mathcal{B} \simeq Hom^{re}_{\mathcal{A}}(\mathcal{M},~\mathcal{M}) as (\mathcal{B}-\mathcal{B}) bimodule categories. So, \mathcal{B}^{rev} \simeq Hom^{re}_{\mathcal{A}}(\mathcal{M},~\mathcal{M}).

Similarly, \mathcal{A} \simeq Hom^{re}_{\mathcal{B}}(\mathcal{M}^{op},~\mathcal{M}^{op})=Hom^{re}_{\mathcal{B}}(\mathcal{M},~\mathcal{M}) as (\mathcal{B}-\mathcal{B}) bimodule categories.

Conversely, if \mathcal{R} is an equivalence, then \mathcal{B}^{rev} \simeq Hom^{re}_{\mathcal{A}}(\mathcal{M},~\mathcal{M}) as (\mathcal{B}-\mathcal{B}) bimodule categories, so \mathcal{B} \simeq Hom^{re}_{\mathcal{A}}(\mathcal{M},~\mathcal{M}) and if \mathcal{L} is an equivalence, then

as (\mathcal{A}-\mathcal{A}) bimodule categories. If \mathcal{R} and \mathcal{L} are equivalences, then \mathcal{M} is invertible.

See [etnios] for the rest of the proof.

Let \mathcal{A} be a fusion category and \mathcal{M} be decomposable module category over \mathcal{A} under the given conditions. Then, \mathcal{M} \simeq \mathcal{P} \oplus \mathcal{Q} for indecomposable module categories \mathcal{P} and \mathcal{Q} over \mathcal{A} and A=\underset{i}{\oplus} A_i for simple objects A_i in \mathcal{A} for all objects A in \mathcal{A} since \mathcal{A} is a semisimple monoidal category. The monoidal functor L:~\mathcal{A} \rightarrow Hom^{re}_{\mathcal{B}}(\mathcal{M},~\mathcal{M}),~L(A)(M)=A\otimes M between monoidal categories is an equivalence for all objects A in \mathcal{A} and for all objects M in \mathcal{M} under these conditions.

A\otimes M=(\underset{i}{\oplus} A_i)\otimes_{l\mathcal{M}} M\simeq (\underset{i}{\oplus} A_i)\otimes_{l\mathcal{M}} (P\oplus Q)=((\underset{i}{\oplus} A_i)\otimes_{l\mathcal{P}} P)\oplus ((\underset{i}{\oplus} A_i)\otimes_{l\mathcal{Q}} Q)\\ =(\underset{i}{\oplus}(A_i\otimes_{l\mathcal{P}} P) \oplus (\underset{i}{\oplus}(A_i\otimes_{l\mathcal{Q}} Q)) for all objects P in \mathcal{P} and Q in \mathcal{Q}.

However, \underset{i}{\oplus}(A_i\otimes_{l\mathcal{P}} P) \oplus \underset{i}{\oplus}(A_i\otimes_{l\mathcal{Q}} Q) is an object in \mathcal{P} \oplus \mathcal{Q}. This means that \underset{i}{\oplus}(A_i\otimes_{l\mathcal{P}} P) is an object in \mathcal{P} and \underset{i}{\oplus}(A_i\otimes_{l\mathcal{Q}} Q) is an object in \mathcal{Q}. So, \mathcal{P}=\underset{i}{\oplus}(A_i\otimes \mathcal{P}) and \mathcal{Q}=\underset{i}{\oplus}(A_i\otimes \mathcal{Q}). This is a contradiction since \mathcal{P} and \mathcal{Q} are indecomposable module categories over \mathcal{A}. As a result, \mathcal{M} is indecomposable.

There exists a functor Hom(\mathcal{B},~\mathcal{C})\times Hom(\mathcal{A},~\mathcal{B})\rightarrow Hom(\mathcal{A},~\mathcal{C}) taking any pair (\mathcal{F},~\mathcal{G}) to \mathcal{F} \circ \mathcal{G} for given monoidal functors (\mathcal{F},~\beta,~\varphi_1):~\mathcal{B} \rightarrow \mathcal{C} in Hom(\mathcal{B},~\mathcal{C}) and (\mathcal{G},~\psi,~\varphi_2):~\mathcal{A} \rightarrow \mathcal{B} in Hom(\mathcal{A},~\mathcal{B}).

\beta is a family of natural isomorphisms \beta_{XY}:~\mathcal{F}(X) \otimes \mathcal{F}(Y) \rightarrow \mathcal{F}(X\otimes Y) for all objects X, Y in \mathcal{B} and \psi is a family of natural isomorphisms \psi_{AB}:~\mathcal{G}(A) \otimes \mathcal{G}(B) \rightarrow \mathcal{G}(A\otimes B) for all objects A and B in \mathcal{A}.

We want to show that \mathcal{F} \circ \mathcal{G} is a monoidal category. We define \gamma as a family of natural isomorphisms \gamma_{AB}=\mathcal{F}(\psi_{AB}) \circ \beta_{\mathcal{G}(A) \mathcal{G}(B)} for all objects A and B in \mathcal{A} as in the following diagram.

I\cong \mathcal{G}(I), so I\cong \mathcal{F}(I) \cong (\mathcal{F} \circ \mathcal{G})(I) for the unit object I in \mathcal{A}. It is easy to prove the commutativity of the required diagrams.

If we have natural transformations \theta_1:~\mathcal{F}_1\Rightarrow \mathcal{F}_2 in Hom(\mathcal{B},~\mathcal{C}) and \theta_2:~\mathcal{G}_1 \Rightarrow \mathcal{G}_2 in Hom(\mathcal{A},~\mathcal{B}), then the composition is again a natural transformation. This gives a morphism in Hom(\mathcal{A},~\mathcal{C}) corresponding to the pair (\theta_1,~\theta_2) in Hom(\mathcal{B},~\mathcal{C})\times Hom(\mathcal{A},~\mathcal{B}).

\mathfrak{X}=\mathcal{MCT}. Objects are FRBSU monoidal categories.

\mathcal{MCT}(\mathcal{A},~\mathcal{B}) is a category in which 1 arrows are braided monoidal functors \mathcal{K}:~\mathcal{A} \rightarrow \mathcal{B} and 2 arrows are natural isomorphisms \theta:~\mathcal{K} \Rightarrow \mathcal{L} for all monoidal functors \mathcal{K},~\mathcal{L}:~\mathcal{A} \rightarrow \mathcal{B}. We show 2 arrows as in the following diagram.

\mathcal{F}_{\mathcal{A} \mathcal{B} \mathcal{C}}:~\mathcal{MCT}(\mathcal{B},~\mathcal{C}) \times \mathcal{MCT}(\mathcal{A},~\mathcal{B}) \rightarrow \mathcal{MCT}(\mathcal{A},~\mathcal{C}) is a functor taking the pair (\mathcal{L},~\mathcal{K}) to \mathcal{L} \circ \mathcal{K} and the pair (\theta,~\gamma) to \theta \star \gamma by Lemma 4.0.1.

\mathcal{F}_{\mathcal{A}}:~1\rightarrow \mathcal{MCT}(\mathcal{A},~\mathcal{A}) is a functor sending the object \star in 1 to 1 arrow id_{\mathcal{A}}:~\mathcal{A}\rightarrow \mathcal{A}.

is a natural isomorphism.

for all 1 arrows \mathcal{K},~\mathcal{L},~\mathcal{M}. (\mathcal{K}\circ \mathcal{L})\circ \mathcal{M}=\mathcal{K}\circ (\mathcal{L}\circ \mathcal{M}) and a_{\mathcal{A} \mathcal{B} \mathcal{C} \mathcal{D}}(\mathcal{K},~\mathcal{L},~\mathcal{M})=id_{\mathcal{K} \circ \mathcal{L}\circ \mathcal{M}}.

For all morphisms \alpha:~(\mathcal{K}_1,~\mathcal{L}_1,~\mathcal{M}_1)\rightarrow (\mathcal{K}_2,~\mathcal{L}_2,~\mathcal{M}_2) in

the following diagram

commutes. As a result a_{\mathcal{A} \mathcal{B} \mathcal{C} \mathcal{D}}(\mathcal{K},~\mathcal{L},~\mathcal{M}) is a natural isomorphism for all 1 arrows \mathcal{K}, \mathcal{L} and \mathcal{M}. Similarly, we show r and l are natural isomorphisms. Hence, MCT is a 2-category.

Assume that (f_1,~f_2) is a strict morphism as in Diagram 5.1.1. We get a commutative diagram

which means that the NE-SW sequence is a split extension in that butterfly, that is E=N_2\ltimes M_1 with the product law (n_1,~m_1).(n_2,~m_2)=(n_1.^{f_1(m_1)}n_2,~m_1.m_2) for all m_1,~m_2\in M_1 and n_1,~n_2\in N_2.

Here, we define g as a projection, k(n)=(id_{N_2},~1)(n)=(n,~1) for all n\in N_2, f(n)=(f_2( n^{-1}),~h_1(n)) for all n\in N_1 and t(n,~m)=h_2(n).f_1(m) for all n\in N_2 and m\in M_1.

Conversely, if we are given a split butterfly as in Diagram Proof, we can find a canonical splitting homomorphism s:~M_1\rightarrow N_2\ltimes M_1 taking m to (1,~m) for all m\in M_1. We define f_1=t\circ s. f_2 can be defined from the equation s\circ h_1=f.(k\circ f_2). We can see that those group homomorphisms satisfy the required conditions.

We take \mathfrak{X}=\underline{\mathcal{XM}}. First, we need to show that \underline{\mathcal{XM}}(\mathfrak{C_1},~\mathfrak{C_2}) is a category whose objects are 1 arrows and morphisms are 2 arrows for the crossed modules \xymatrix{\mathfrak{C_1}=[N_1\ar[r]^{h_{\mathfrak{C_1}}} & M_1]} and \xymatrix{\mathfrak{C_2}=[N_2\ar[r]^{h_{\mathfrak{C_2}}} & M_2]} in \underline{\mathcal{XM}}.

The identity morphism is the trivial pointed natural transformation.

We define the composition of two pointed natural transformations PNT_1:~G\Rightarrow F which is a crossed homomorphism \gamma_1 and PNT_2:~F\Rightarrow E which is a crossed homomorphism \gamma_2 between the strict morphisms as \gamma=\gamma_2. \gamma_1:~M_1\rightarrow N_2. For all elements a,~a' in M_1, we get

\gamma(a.a')=\gamma_2(a.a') . \gamma_1(a.a')=(^{e_1(a')}\gamma_2(a))\gamma_2(a') .(^{f_1(a')}\gamma_1(a))\gamma_1(a')

=(^{e_1(a')}\gamma_2(a))\gamma_2(a')(^{e_1(a')h_{\mathfrak{C_2}}(\gamma_2((a')^{-1}))}\gamma_1(a))\gamma_1(a') since f_1(a')=e_1(a')h_{\mathfrak{C_2}}(\gamma_2((a')^{-1}))

=(^{e_1(a')}\gamma_2(a))\gamma_2(a')(^{e_1(a')}\gamma_2((a')^{-1})\gamma_1(a)\gamma_2(a'))\gamma_1(a') since

^{h_{\mathfrak{C_2}}(\gamma_2((a')^{-1}))}\gamma_1(a)=\gamma_2((a')^{-1}) . \gamma_1(a) . \gamma_2(a') by definition of crossed module

=(^{e_1(a')}\gamma_2(a)). \gamma_2(a'). (^{e_1(a')}\gamma_2((a')^{-1})). (^{e_1(a')}\gamma_1(a)). (^{e_1(a')}\gamma_2(a')) . \gamma_1(a')

=(^{e_1(a')}\gamma_2(a)). \gamma_2(a').(^{e_1(a')}\gamma_2((a')^{-1})). \gamma_2(a') . \gamma_2((a')^{-1}) . (^{e_1(a')}\gamma_1(a)) . (^{e_1(a')}\gamma_2(a')) . \gamma_1(a')

=(^{e_1(a')}\gamma_2(a)). \gamma_2(a') . \gamma_2((a')^{-1} . a') . \gamma_2((a')^{-1}) . (^{e_1(a')}\gamma_1(a)) . (^{e_1(a')}\gamma_2(a')) . \gamma_2(a') . \gamma_2((a')^{-1}) . \gamma_1(a')

=(^{e_1(a')}\gamma_2(a)). (^{e_1(a')}\gamma_1(a)). \gamma_2(a' a'). \gamma_2((a')^{-1}). \gamma_1(a')

=(^{e_1(a')}\gamma_2(a). \gamma_1(a)). \gamma_2(a'). \gamma_1(a')

=(^{e_1(a')}\gamma(a)). \gamma(a').

g_1(a)=f_1(a). h_{\mathfrak{C_2}}(\gamma_1(a^{-1})) since PNT_1:~G\Rightarrow F is a pointed natural transformation which is a crossed homomorphism \gamma_1 and f_1(a)=e_1(a). h_{\mathfrak{C_2}}(\gamma_2(a^{-1})) since PNT_2:~F\Rightarrow E is a pointed natural transformation which is a crossed homomorphism \gamma_2. Hence,

as desired. Similarly, we may show the other part. As a result, \gamma is a crossed homomorphism and gives a pointed natural transformation PNT_3:~G\Rightarrow E.

It is clear that the composition is associative.

For all pointed natural transformations PNT:~G\Rightarrow F, the crossed homomorphism \gamma and id:~F\Rightarrow F, the composition id\star PNT:~G\Rightarrow F is equal to PNT. Similarly, we show the other part.

As a result, \underline{\mathcal{XM}}(\mathfrak{C_1},~\mathfrak{C_2}) is a category.

The mapping \mathcal{F}_{\mathfrak{C_1} \mathfrak{C_2} \mathfrak{C_3}}:~\underline{\mathcal{XM}}(\mathfrak{C_2},~\mathfrak{C_3})\times \underline{\mathcal{XM}}(\mathfrak{C_1},~\mathfrak{C_2}) \rightarrow \underline{\mathcal{XM}}(\mathfrak{C_1},~\mathfrak{C_3}) is a functor. We send each pair (G,~F) to G\circ F where G=(g_1,~g_2), F=(f_1,~f_2) and G\circ F=(g_1\circ f_1,~g_2\circ f_2) are strict morphims between the corresponding crossed modules.

We send each pair of pointed natural transformations PNT_2:~E\Rightarrow K which is a crossed homomorphism \gamma_2 and PNT_1:~G\Rightarrow F which is a crossed homomorphism \gamma_1 to their composition PNT_3:~E\circ G\Rightarrow K\circ F. Here, G=(g_1,~g_2), F=(f_1,~f_2), E=(e_1,~e_2) and K=(k_1,~k_2) are strict morphisms. We need to define a crossed homomorphism \gamma_3.

We take \gamma_3=(k_2\circ \gamma_1) . (\gamma_2 \circ g_1) and see it satisfies the required conditions to be a crossed homomorphism. We draw the following diagrams to see the relation between the strict morphisms.

We also draw the following diagrams to understand the group homomorphisms better.

For all a,~a'\in M_1, b\in N_1 and c\in N_2, d,~d'\in M_2, we get the following equalities by definition of \gamma_1 and \gamma_2.

1. \gamma_1(a.a')=(^{f_1(a')}\gamma_1(a)). \gamma_1(a')

2. g_1(a)=f_1(a). h_{\mathfrak{C_2}}(\gamma_1(a^{-1}))

3. g_2(b)=f_2(b). \gamma_1(h_{\mathfrak{C_1}}(b^{-1}))

4. \gamma_2(d.d')=(^{k_1(d')}\gamma_2(d)). \gamma_2(d')

5. e_1(d)=k_1(d). h_{\mathfrak{C_3}}(\gamma_2(d^{-1}))

6. e_2(c)=k_2(c). \gamma_2(h_{\mathfrak{C_2}}(c^{-1}))

For all a'\in M_1, we have

For all a,~a'\in M_1, we have

\gamma_3(a . a')=(k_2\circ \gamma_1)(a.a') . (\gamma_2 \circ g_1)(a. a')=k_2((^{f_1(a')}\gamma_1(a)). \gamma_1(a')) . \gamma_2(g_1(a) . g_1(a')) by ?

=k_2(^{f_1(a')}\gamma_1(a)). k_2(\gamma_1(a')). (^{(k_1\circ g_1)(a')}(\gamma_2 \circ g_1)(a)) . (\gamma_2 \circ g_1)(a') by ?

=k_2(^{f_1(a')}\gamma_1(a)). k_2(\gamma_1(a')). (^{(k_1\circ f_1)(a'). k_1(h_{\mathfrak{C_2}}(\gamma_1((a')^{-1})))}(\gamma_2\circ g_1)(a)) . (\gamma_2 \circ g_1)(a') by ?

=k_2(^{f_1(a')}\gamma_1(a)). k_2(\gamma_1(a')). (^{(k_1\circ f_1)(a'). (h_{\mathfrak{C_3}} (k_2\circ \gamma_1)((a')^{-1}))}(\gamma_2\circ g_1)(a)) . (\gamma_2 \circ g_1)(a') since K is a strict morphism

=k_2(^{f_1(a')}\gamma_1(a)). k_2(\gamma_1(a')). (^{(k_1\circ f_1)(a')}(k_2\circ \gamma_1)((a')^{-1}).(\gamma_2\circ g_1)(a). (k_2\circ \gamma_1)(a')). (\gamma_2 \circ g_1)(a') by definition of \mathfrak{C_3}

=k_2(^{f_1(a')}\gamma_1(a)). k_2(\gamma_1(a')). (^{(k_1\circ f_1)(a')}(k_2\circ \gamma_1)((a')^{-1})) .(^{(k_1\circ f_1)(a')}(\gamma_2\circ g_1)(a)).\\ (^{(k_1\circ f_1)(a')}(k_2\circ \gamma_1)(a')). (\gamma_2 \circ g_1)(a')

=(^{(k_1\circ f_1)(a')}(k_2\circ \gamma_1)(a)).k_2(\gamma_1(a')). (^{(k_1\circ f_1)(a')}(k_2\circ \gamma_1)((a')^{-1})) .(^{(k_1\circ f_1)(a')}(\gamma_2\circ g_1)(a)).\\ (^{(k_1\circ f_1)(a')}(k_2\circ \gamma_1)(a')). (\gamma_2 \circ g_1)(a') since K is a strict morphism

=(^{(k_1\circ f_1)(a')}(k_2\circ \gamma_1)(a)). k_2(\gamma_1(a')). (k_2\circ \gamma_1)((a')^{-1}). (^{(k_1\circ f_1)(a')}(\gamma_2\circ g_1)(a)).\\ (^{(k_1\circ f_1)(a')}(k_2\circ \gamma_1)(a')). (\gamma_2 \circ g_1)(a') by ?

=(^{(k_1\circ f_1)(a')}(k_2\circ \gamma_1)(a)). (^{(k_1\circ f_1)(a')}(\gamma_2\circ g_1)(a)). (^{(k_1\circ f_1)(a')}(k_2\circ \gamma_1)(a')). (\gamma_2 \circ g_1)(a')

=(^{(k_1\circ f_1)(a')}(k_2\circ \gamma_1)(a)). (^{(k_1\circ f_1)(a')}(\gamma_2\circ g_1)(a)). (k_2\circ \gamma_1)(a'). (\gamma_2 \circ g_1)(a') by ?

=(^{(k_1\circ f_1)(a')}(k_2\circ \gamma_1)(a). (\gamma_2\circ g_1)(a)) . (k_2\circ \gamma_1)(a'). (\gamma_2 \circ g_1)(a')

=(^{(k_1\circ f_1)(a')}((k_2\circ \gamma_1). (\gamma_2\circ g_1))(a)). ((k_2\circ \gamma_1) . (\gamma_2 \circ g_1))(a')

=(^{(k_1\circ f_1)(a')}\gamma_3(a)). \gamma_3(a') as required.

The other conditions are satisfied.

For a crossed homomorphism \xymatrix{\mathfrak{C}=[N\ar[r]^{h_{\mathfrak{C}}} & M]}, the mapping \mathcal{F}_{\mathfrak{C}}:~1\rightarrow \underline{\mathcal{XM}}(\mathfrak{C},~\mathfrak{C}) is a functor taking the element \star in 1 to id_h=(id,~id) and morphisms \star \rightarrow \star to a trivial pointed natural transformation PNT:~(id,~id)\Rightarrow (id,~id).

a_{\mathfrak{C_1} \mathfrak{C_2} \mathfrak{C_3} \mathfrak{C_4}}(F,~G,~H):~(F\circ G)\circ H\Rightarrow F\circ (G\circ H) is a trivial pointed natural transformation and (F\circ G)\circ H=F\circ (G\circ H) for all strict morphisms F, G and H. a_{\mathfrak{C_1} \mathfrak{C_2} \mathfrak{C_3} \mathfrak{C_4}}(F,~G,~H) is the identity morphism by assumption, hence it is an isomorphism and the required diagram is commutative. As a result, a_{\mathfrak{C_1} \mathfrak{C_2} \mathfrak{C_3} \mathfrak{C_4}} is a natural isomorphism.

We show the other conditions in a similar way and see \underline{\mathcal{XM}} is a 2-category.