Hopf 2-algebras and Braided Monoidal 2-Categories

This is a straightforward computation using the pairing (?). In particular, the equivariance and Peiffer identity of t^*, as well as the fact that {\@fontswitch\relax\mathcal G}_0^* forms a {\@fontswitch\relax\mathcal G}_{-1}^*-bimodule, follow directly from dualizing (?), (?), (?), (B.3).

What is non-trivial is (?). Define \Delta^*_0 by dualizing the bimodule structure \cdot of {\@fontswitch\relax\mathcal G}, then we have

We now compute using analogues of (?) for \Delta^*, that

and similarly

hence \Delta also satisfies (?). This proves that ({\@fontswitch\relax\mathcal G}^*,\cdot^*,\Delta^*) is an associative 2-bialgebra iff ({\@fontswitch\relax\mathcal G},\cdot,\Delta) also is.

Now consider the units and counits. Given

we see that \eta is a unit for ({\@fontswitch\relax\mathcal G},\cdot) (ie. these quantities all vanish) iff \eta^* is a counit for ({\@fontswitch\relax\mathcal G}^*,\Delta^*). Similarly, \epsilon is a counit for ({\@fontswitch\relax\mathcal G},\Delta) iff \epsilon^* is a unit for ({\@fontswitch\relax\mathcal G}^*,\cdot^*).

This is a direct computation. By the equality in the second row of (?), we have

giving ty\cdot y' = y\cdot ty'. Now by the fact that t is an algebra homomorphism, we have

for which the first row of (?) states yy' = y\cdot ty'. Altogether yields

for any y,y'\in {\@fontswitch\relax\mathcal G}_{-1}, which is precisely the Peiffer identity on {\@fontswitch\relax\mathcal G}. Similarly, if \bar\lhd is the coadjoint representation then (?) reproduces the Peiffer identity on {\@fontswitch\relax\mathcal G}^*.

Let {\@fontswitch\relax\mathcal K}={\@fontswitch\relax\mathcal K}_{-1}\xrightarrow{T}{\@fontswitch\relax\mathcal K}_0 be a 2-bialgebra factorizing into two 2-subbialgebras {\@fontswitch\relax\mathcal G},{\@fontswitch\relax\mathcal H}, with typical elements w\in {\@fontswitch\relax\mathcal K}_0 and e\in {\@fontswitch\relax\mathcal K}_{-1}. Its 2-algebra structure \hat\cdot contains a multiplication ww' in {\@fontswitch\relax\mathcal K}_0 and a {\@fontswitch\relax\mathcal K}_0-bimodule structure w\hat\cdot e,e\hat\cdot w on {\@fontswitch\relax\mathcal K}_{-1}, which are both associative. Since (?) is a span of 2-vector spaces, we have

where t_G,t_H are the t-maps in {\@fontswitch\relax\mathcal G},{\@fontswitch\relax\mathcal H} respectively, and \iota_{-1},\iota_0 are the graded components of the inclusion \iota; similarly for \jmath.

We now separate the bimodule structure \hat\cdot into components according to the span (?),

then for e=\iota_{-1}(y),e'=\jmath_{-1}(g) where y\in G_{-1},g\in H_{-1} we have

By the Peiffer identity in {\@fontswitch\relax\mathcal K}, this should read as a left-multiplication of y on g. We lift this action along t_{\@fontswitch\relax\mathcal H} to create a map \hat\Upsilon_y:{\@fontswitch\relax\mathcal H}_0\rightarrow {\@fontswitch\relax\mathcal H}_{-1}, for which \hat\Upsilon_y(t_{\@fontswitch\relax\mathcal H} g) denotes the left-multiplicaion of y by g. Similarly we have the lift \hat\Upsilon_g:{\@fontswitch\relax\mathcal G}_0\rightarrow {\@fontswitch\relax\mathcal G}_{-1} of the right-multiplication of g on y.

Provided we identify \hat\Upsilon_{y\otimes g} = \Upsilon_y\otimes \tilde\Upsilon_g, the Peiffer conditions in {\@fontswitch\relax\mathcal K} are then equivalent to the 2-representation properties (?), (?). In particular, the multiplication y\cdot g=\Upsilon_y(t^*g) = (ty)\tilde\Upsilon_g is given by the generalized Peiffer identity as shown in Proposition C.2.

Now we prove that (?) is in fact a span of 2-algebras. Due to the linear isomorphsm \hat\cdot\circ(\iota\otimes\jmath), there exists a tuple of well-defined linear maps \Psi = (\Psi_0,\Psi_{-1};\bar\Psi): {\@fontswitch\relax\mathcal G}\otimes {\@fontswitch\relax\mathcal H}\rightarrow {\@fontswitch\relax\mathcal H}\otimes {\@fontswitch\relax\mathcal G}, called the braided transposition, such that

where \Psi_{-1} = \Psi_{-1}^l+\Psi_{-1}^r and x\in {\@fontswitch\relax\mathcal G}_0,y\in {\@fontswitch\relax\mathcal G}_{-1},f\in {\@fontswitch\relax\mathcal H}_0,g\in {\@fontswitch\relax\mathcal H}_{-1}. Due to Peiffer conditions on {\@fontswitch\relax\mathcal K}, these braiding maps are not independent and must satisfy

By collecting all of the graded components of \Psi in accordance with the shorthand notation z=(y,x)\in {\@fontswitch\relax\mathcal G},h=(g,f) \in {\@fontswitch\relax\mathcal H}, the definition of \Psi can be concisely written as

and the relations between its components is summarized as

where T' = t_{\@fontswitch\relax\mathcal H}\otimes t_{\@fontswitch\relax\mathcal G} is the t-map of the 2-bialgebra {\@fontswitch\relax\mathcal K}'\cong{\@fontswitch\relax\mathcal H}\otimes{\@fontswitch\relax\mathcal G} with {\@fontswitch\relax\mathcal G},{\@fontswitch\relax\mathcal H} swapped in the span (?). (?) then implies in particular that \Psi:{\@fontswitch\relax\mathcal K}\rightarrow {\@fontswitch\relax\mathcal K}' is a 2-vector space homomorphism.

We now proceed formally as in the 1-bialgebra case [Majid:1996kd, Majid:1994nw]. The associativity in {\@fontswitch\relax\mathcal K} is

which yields the 2-braiding relations

This then allows us to define the actions

where \epsilon denotes the counit map. Applying \operatorname{id}\otimes \epsilon and \epsilon\otimes\operatorname{id} respectively to the first and second equation of (?) implies that \bar\rhd,\bar\lhd respect the semidirect product structures {\@fontswitch\relax\mathcal G}_{-1}\rtimes {\@fontswitch\relax\mathcal G}_0,{\@fontswitch\relax\mathcal H}_{-1}\rtimes {\@fontswitch\relax\mathcal H}_0, respectively. Together with our above result, (?) is in fact a span of 2-algebras.

We now prove that (?) is actually a span of 2-bialgebras, which proves the theorem. Applying \epsilon\otimes\operatorname{id} and \operatorname{id}\otimes \epsilon respectively to the first and second (?) yields

We now take the coproduct \Delta_K:{\@fontswitch\relax\mathcal K}\rightarrow {\@fontswitch\relax\mathcal K}^{2\otimes} on {\@fontswitch\relax\mathcal K}, given in components and Sweedler notation (see (?), (?)) by

note w_{(1)}^l,w_{(2)}^r\in {\@fontswitch\relax\mathcal K}_{-1}. With the span (?), we can write w=\iota_0(x)\jmath_0(f),e=\iota_{-1}(y)\jmath_{-1}(g) for some appropriate elements x,f,y,g such that

This then allows us to define coproducts on {\@fontswitch\relax\mathcal G},{\@fontswitch\relax\mathcal H} by

whence \Delta_{{\@fontswitch\relax\mathcal K}} = \Delta_{{\@fontswitch\relax\mathcal G}\otimes {\@fontswitch\relax\mathcal H}}, which implies that \hat\cdot\circ (\iota\otimes\jmath) and \hat\cdot\circ(\jmath\otimes\iota) by construction respects the coproducts.

As such, \Psi is a 2-coalgebra map. In particular, we have

where {\@fontswitch\relax\mathcal K}' is the 2-bialgebra with {\@fontswitch\relax\mathcal G},{\@fontswitch\relax\mathcal H} swapped in the span (?). An application of \epsilon\otimes \operatorname{id}\otimes \epsilon\otimes\operatorname{id} and \operatorname{id}\otimes \epsilon\otimes \operatorname{id}\otimes \epsilon to (?) gives

which ensures that \bar\rhd,\bar\lhd are 2-coalgebra maps.

Now applying \epsilon\otimes\operatorname{id}\otimes\operatorname{id}\otimes \epsilon and \operatorname{id}\otimes\epsilon\otimes\epsilon\otimes\operatorname{id} to (?) yields

Using the second equation, together with (?), gives (?) and

which are precisely the matched pair conditions (?), (?) for \hat\cdot|_{\@fontswitch\relax\mathcal G}=\cdot,\hat\cdot|_{{\@fontswitch\relax\mathcal H}}=\cdot^*. On the other hand, using the first equation gives (?). Thus (?) is a span of 2-bialgebras and so K \cong {\@fontswitch\relax\mathcal G} \bar{\bowtie}{\@fontswitch\relax\mathcal H}.

Recall that {\@fontswitch\relax\mathcal R} is quasitriangular iff {\@fontswitch\relax\mathcal R}^l,{\@fontswitch\relax\mathcal R}^r are square and invertible. This pairs {\@fontswitch\relax\mathcal G} with itself and hence \operatorname{dim}{\@fontswitch\relax\mathcal G}_0 = \operatorname{dim}{\@fontswitch\relax\mathcal G}_{-1}. We calculate (\operatorname{id}\otimes \sigma\circ \Delta_0^l){\@fontswitch\relax\mathcal R}^l and (\sigma\circ \Delta_0^l\otimes \operatorname{id} ){\@fontswitch\relax\mathcal R}^r, as well as (\operatorname{id}\otimes \sigma\circ \Delta_0^r){\@fontswitch\relax\mathcal R}^l and (\sigma\circ \Delta_0^r\otimes \operatorname{id} ){\@fontswitch\relax\mathcal R}^r in two ways. First using (?), we have

On the other hand from (?), we have that,

Putting each equation with its above counterpart leads to (C.3).

Let ({\@fontswitch\relax\mathcal G}^*,{\@fontswitch\relax\mathcal T}^*,\Delta^*) be the dual weak 2-bialgebra. By a straightforward computation, it is obvious that the conditions (?), (?), (D.2), (?) on ({\@fontswitch\relax\mathcal G}^*,\Delta^*) hold iff the Peiffer conditions, weak 1-associativity and weak bimodularity hold for the 2-algebra {\@fontswitch\relax\mathcal G}. The fact that (\Delta_{-1}^*,\Delta_0^*) are 2-algebra maps, as well as the units/counits, are treated the same way as in the proof of Proposition C.1.

The two non-trivial identities to check are (D.2), (?). We will use the Sweedler notation for \Delta_1, \Delta_1 x= x_{(1)}\otimes x_{(2)}\otimes x_{(3)}. Note x\in{\@fontswitch\relax\mathcal G}_0, but x_{(i)}\in {\@fontswitch\relax\mathcal G}_{-1} for i=1,2,3.

We begin first with the latter, and check that \Delta_1 is a 2-algebra map. We have

while (recall \Delta_0' dualizes to the multiplication on {\@fontswitch\relax\mathcal G}_0^*, and f_{(i)}\in {\@fontswitch\relax\mathcal G}^*_{-1})

meaning that \Delta_1 is a 2-algebra map on ({\@fontswitch\relax\mathcal G},{\@fontswitch\relax\mathcal T}) iff \Delta_1^* is also one on ({\@fontswitch\relax\mathcal G}^*,{\@fontswitch\relax\mathcal T}^*).

We now check (?). Fix arbitrary elements f_1,\dots,f_4\in{\@fontswitch\relax\mathcal G}_{-1}^*, we compute

on the other hand, we have

By summing these up, the Hochschild 3-cocycle condition for {\@fontswitch\relax\mathcal T}^* is equivalent to

which is exactly (?).

Let us focus first on the left diagram. The goal is to show that \rho_{i\otimes W} defines a cochain homotopy which fits into the following diagram

where the horizontal maps are the differentials given in (?), and the vertical maps are various components of \rho_{V\otimes W}\circ i - i\circ\rho_{V\otimes W}.

The key is the commutation relation (F.2), which allows us to write

for each y\in{\@fontswitch\relax\mathcal G}_{-1}, as well as the definition (?) of \Delta_0'. Directly computing, we have for the rightmost triangle

as maps on V_0\otimes W_0 (with deg = 0), and similarly we have for the leftmost triangle

as maps on V_{-1}\otimes W_{-1} (with deg = -1).

Now consider the middle section. We need to compute

and sum them to find

as maps on V_{-1}\otimes W_0\oplus V_0\otimes W_{-1}. The other diagram is treated identically.

What we need to show is that \rho_{i\otimes j} = (\rho_{i\otimes T}\ast\rho_{V\otimes j})\circ t = (\rho_{U\otimes j}\ast\rho_{i\otimes W})\circ t as 2-morphisms. Recall cochain homotopies q:f\Rightarrow g,\, p:g\Rightarrow h in \mathsf{2Vect}^{hBC} compose by p\ast q = p\partial_U q :f\Rightarrow h, where U is the source 2-vector space of the cochain map g. Indeed, we have

as desired, where W is the target of h and V is the source of f. Notice this is exactly how elements in \mathfrak{End}(V)_{-1} compose, A\ast A' = A\delta A.

The goal is to prove that D^\pm\rho_{i\otimes j}(y) in fact decomposes as described above for each y\in{\@fontswitch\relax\mathcal G}_{-1}. This follows from the coequivariance condition (?). By direct computation, precomposing (?) yields (here we neglect the 2-vector space subscripts for brevity)

where we have used (F.2) to commute the t-map past the 2-representations to the differential \partial. Using the Sweeder notation (?) for \Delta_{-1}, we compute their graded composition to be

as desired, where we have noted the property i_{-1}(\rho_V)_1 = (\rho_U)_1i_0 of the 2-intertwiners i,j to permute them past the \rho’s. This proves that the 2-algebra homomorphisms \rho_{i\otimes j} = \rho_{i\otimes T} \ast \rho_{V\otimes j} coincide. A similar argument shows that the 2-algebra homomorphisms \rho_{i\otimes j} = \rho_{U\otimes j}\ast \rho_{i\otimes W} also coincide.

This is not sufficient to imply that i\otimes T \circ V\otimes j = U\otimes j \circ i\otimes W, however. Indeed, the weak component \varrho of the two decomposed 2-representations in general may differ. After some computations, one can show that we have

The difference \phi_{i,j}= \varrho_{(i\otimes T)\circ (V\otimes j)}\ast \varrho_{(U\otimes j)\circ (i\otimes W)}^{-1} between these 2-morphisms is precisely the 2-isomorphism i\otimes T \circ V\otimes j \cong U\otimes j \circ i\otimes W.

Consider first (?). We precompose (?) with t and reconstruct the associators corresponding to each term according to the definition,

where \operatorname{id}_i:i\Rightarrow i denotes the identity modification on the 2-intertwiner i. Now note that, by coequivariance (?) D_t\circ \Delta_{-1} = \Delta_0\circ t, we have

whence the pentagon relation

is equivalently expressed as

which is nothing but the 2-coassociativity (?) precomposed with t. Now by the coPeiffer identity \Delta_0' = D_t\Delta_0 (?), the same argument shows that the pentagon relations for the rest of the associator 2-morphisms (?), (?) are equivalent to applying the t-map D_t,D_t[2] to (?).

Similarly, under the complete t-map D_t[3] = t\otimes t\otimes t, the 2-coasscociativity condition (?) becomes

which by (?) implies the pentagon relation for the associator 1-morphism (?).

The first statement follows directly from the definitions, and by using the same argument as in the proofs of Lemma F.1, and also later in Lemma G.2. Similarly, by adapting the proof of Lemma F.2, we see that (?), (?) admit the following decompositions

where D_\delta is the tensor triple of the t-map \delta on \mathfrak{End}(V), and "etc." means permutations of the subscripts. This proves the second statement.

Note for each 2-representation \rho, the flip map \text{flip}:V\otimes W\rightarrow W\otimes V is a 2-intertwiner between \rho and \rho'=\rho\circ\sigma. Moreover, we interpret the cochain homotopy defined by (\rho_{V\otimes W})_0^1(x) for each x\in {\@fontswitch\relax\mathcal G}_0 as a modification between the action (\rho_{V\otimes W})_0^0(x) and itself, treated as a 2-intertwiner; similarly for \rho'. Therefore, in order for the mixed braiding map b_{iW} to be a modification in \operatorname{2Rep}({\@fontswitch\relax\mathcal G}), it must commute with the cochain homotopy (\rho_{V\otimes W})_0^1(x) — namely

where \ast denotes the composition of cochain homotopies. With \rho_{W\otimes V}' = (\rho_W\otimes\rho_V) \circ \sigma\circ \Delta, this is satisfied by definition (?) of b_{iW} iff

which is precisely (?).

Similarly, in order for the braiding map b_{VW} to be a 2-intertwiner, it must commute with the action (\rho_{V\otimes W})_0^0(x) for each x\in {\@fontswitch\relax\mathcal G}_0:

where \circ denotes the composition of 2-intertwiners.

First if the 2-representation \rho were strict, then this translates to the algebraic condition

which in fact follows also from (?). To see this, we recall the definitions (?) of R and (?) of the coproduct \Delta_0', and simply apply t\otimes 1 and 1\otimes t respectively to (?). The fact that t is an algebra homomorphism and that (t\otimes 1)\circ\sigma = \sigma\circ(1\otimes t) proves the statement.

Second, if the 2-representation \rho were weak, then in general the component \varrho gives rise to a possibly non-trivial invertible natural 2-morphism

We will not need this 2-morphism in the following so we shall suppose I_{b_{VW},\bullet}=\operatorname{id}.

For brevity, we shall suppress the subscripts V,U,W on the 2-representations. Recall the two equivalent ways R^r,R^l to express R in (?). We can then write

Consider the left diagram. As 2-morphisms in \operatorname{2Rep}({\@fontswitch\relax\mathcal G}) are given by cochain homotopies, we need to show that the definition (?) of the mixed braiding map b_{iW} = b_{iW}^1 + b_{iW}^2 fits into the following diagram

where the vertical arrows are the various graded components of b_{UW}\circ i - i\circ b_{VW}, and the horizontal arrows are the differentials on the three-term tensor product complex (?); for instance, the ones at the top row are given by D^\pm= 1\otimes\partial_W\pm \partial_V\otimes 1.

As in Lemma F.1, the key towards this is the commutative square (F.2), which states that for each y\in {\@fontswitch\relax\mathcal G}_{-1} we have

Let us examine first the commutative triangle on the ends of (?). First, for the right-most triangle, we compute in terms of the components b_{iW}^{1,2} that

as maps on V_0\otimes W_0. Similarly for the left-most triangle, we have

as maps V_{-1}\otimes W_{-1}. Note the sign (-1)^\text{deg} in (?) is non-trivial here as {\@fontswitch\relax\mathcal R} acts on the degree-(-1) part of the tensor product V\otimes W.

We now turn to the middle section of (?). We are required to compute the following,

Summing these and rearranging terms gives, as maps on V_{-1}\otimes W_0\oplus V_0\otimes W_{-1},

The diagram on the right is treated identically, and this establishes the first statement. The second statement directly follows from the fact that (j\circ i)\circ \rho_V = j\circ \rho_U \circ i = \rho_X\circ (j\circ i) for composable 2-intertwiners i,j.

Recall (\mathfrak g,\delta) is a Lie 2-bialgebra iff \delta is a Lie 2-algebra 2-cocycle [Bai_2013]. Therefore it suffices to show that the 2-cochain defined in (?) is a 2-cocycle. This shall follow from the fact that ({\@fontswitch\relax\mathcal G},\cdot,\Delta) is a 2-bialgebra — namely the coproduct map \Delta (?) satisfies (?), (?) and (?).

First note that (?) and (?) for the coproduct \Delta translates directly to the conditions

for the 2-cochain \delta=\delta_{-1}+\delta_0. Now by a direct computation using (?), the condition (?) implies

where we have used the the Peiffer identity and the fact that \bar X_{(1)} = tX_{(1)} inherited from the constraints (B.2), and

where \bar X_{(2)} = X_{(2)}. These are precisely the Lie 2-algebra 2-cocycle conditions for \delta [Bai_2013, chen:2022].

We construct the Lie 2-algebra structure as in Lemma B.1. Let U_3=\mathcal{L}\circ {\@fontswitch\relax\mathcal T}\circ\mathcal{L} denote the induced trilinear map on \mathcal{L}({\@fontswitch\relax\mathcal G}). We apply \mathcal{L} to the Jacobiator J(X,X',X'')=[X,[X',X'']] + [X',[X'',X'] + [X'',[X,X']],

where we have used the weak 1-associativity condition for {\@fontswitch\relax\mathcal G}. Similarly, for J(X,X',Y)=X\rhd (X'\rhd Y) - X'\rhd (X\rhd Y) - [X,X']\rhd Y we have

hence if we define the total skew-symmetrization

then weak 1-associativity implies the 2-Jacobi identity on \mathcal{L}({\@fontswitch\relax\mathcal G}).

Using the Peiffer conditions on this fact, we see that the weak bimodularity condition also implies the 2-Jacobi identity, with two tY’s inserted in U_3 instead. Similar computations show that the Hochschild 3-cocycle condition for {\@fontswitch\relax\mathcal T} implies the Lie 3-cocycle condition for \mu_3.

Finally, let F:({\@fontswitch\relax\mathcal G},{\@fontswitch\relax\mathcal T})\rightarrow ({\@fontswitch\relax\mathcal G}',{\@fontswitch\relax\mathcal T}') denote a weak 2-algebra homomorphism as defined in (?). By applying the Lie-ification functor and appropriately skew-symmetrizing {\@fontswitch\relax\mathcal T},{\@fontswitch\relax\mathcal T}' and the 2-algebra structure, we recover precisely the definition of a weak 2-algebra map \mathcal{L}(F):(\mathfrak g,\mu)\rightarrow (\mathfrak g',\mu') [Baez:2005sn]. Thus \mathcal{L} is functorial.

As foretold, we reconstruct the module associator \alpha and pentagonator \pi of the {\@fontswitch\relax\mathcal G}-modules V\in \mathsf{2Vect}^{hBC} by taking

where \rho= (\varrho,\rho_0,\rho_1): {\@fontswitch\relax\mathcal G}\rightarrow \mathfrak{End}(V) is a weak 2-representation and \mathfrak{T} is the Hochschild 3-cocycle on \mathfrak{End}(V). We now proceed level by level.

paragraph41ex plus1ex minus.2ex-.5em plus -.5em minus -.1emObjects. We identify the action 2-functor \rhd as the weak 2-representation \rho such that x\rhd V = \rho_0(x)V for each x\in {\@fontswitch\relax\mathcal G}_0. An arrow x\rhd V\rightarrow x'\rhd V is therefore expressed as \rho_1(y)V, where y\in {\@fontswitch\relax\mathcal G}_{-1} is interpreted as a 2-morphism x\xRightarrow{y}x' between x,x'=x+ty [Wagemann+2021, Baez:2003fs], or simply by \rho_1(y). What we need to prove is the pentagon relation between \alpha,\pi, as well as the associahedron condition for \pi. The pentagon relation can be written as

Rewriting \pi in terms of the 3-cocycle \mathfrak{T}, we have

which is nothing but the last equation of (?). It is then easy to see that the associahedron condition follows from the Hochschild 3-cocycle condition for \mathfrak{T}.

paragraph41ex plus1ex minus.2ex-.5em plus -.5em minus -.1em2-intertwiners. Recall the notion of weak 2-intertwiners that we have given in Definition F.3. By treating V as a {\@fontswitch\relax\mathcal G}-module 2-category and taking \rhd,\rhd' as the action 2-functors corresponding to the 2-representations \rho,\rho', we equivalently characterize the cochain homotopy I as a collection of invertible natural transformations I_{\bullet,i}: i(\bullet \rhd V) \Rightarrow \bullet \rhd' i(V), such that the following pentagon relation

follows directly from (F.3) This recovers precisely the notion of a {\@fontswitch\relax\mathcal G}-module functor [Delcamp:2023kew]. Notice no pentagonator appears here, as this is a relation on the 2-morphisms in \operatorname{2Rep}^{\@fontswitch\relax\mathcal T}({\@fontswitch\relax\mathcal G}) and hence a pentagonator for it would have to be a 3-morphism.

paragraph41ex plus1ex minus.2ex-.5em plus -.5em minus -.1emModifications. Now let us consider the notion of modifications in \operatorname{2Rep}^{\@fontswitch\relax\mathcal T}({\@fontswitch\relax\mathcal G}) we have defined in Definition F.4. The condition (F.4) is equivalent to the composition of 2-morphisms (\operatorname{id}_{\rho_0(x)}\mu)\ast I_{x,i} = I_{x,i'}\ast \mu, which is exactly a module natural transformation [Delcamp:2023kew].