Generalized (co)homology of symmetric quandles over homogeneous Beck modules

Note that {\mathscr{F} \rtimes X} is a rack, as proved in [MR2155522, Proposition 2.1]. It remains to check that \rho_{\mathscr{F} \rtimes X} is a good involution.

• For (S1), we see that \rho_{\mathscr{F} \rtimes X}^2(a,x)= \rho_{\mathscr{F} \rtimes X}(\eta_x(a), \rho_X(x))=(\eta_{\rho_X(x)}(\eta_x(a)), \rho_X^2(x))=(a,x), where the last equality follows from the (X,\rho_X)-module axiom (M3).

• For (S2), we have

  \begin{aligned} \rho_{\mathscr{F} \rtimes X}\big((a,x)\big) \tilde{\ast} (b,y)&=(\eta_x(a), \rho_X(x)) \tilde{\ast}(b,y)\\ &=(\phi_{\rho_X(x),y}(\eta_x(a))+\psi_{\rho_X(x),y}(b), \rho_X(x) \ast y)\\ &= (\eta_{x \ast y}(\phi_{x,y}(a)+\psi_{x,y}(b)), \rho_X(x \ast y)),\\ && \textrm{by $(X, \rho_X)$-module axioms \hyperref[M4]{(M4)} and \hyperref[M5]{(M5)}, and $\rho_X$ being a good involution}\\ &= \rho_{\mathscr{F} \rtimes X}(\phi_{x,y}(a)+\psi_{x,y}(b), x \ast y)\\ &= \rho_{\mathscr{F} \rtimes X} \big((a,x) \tilde{\ast}(b,y) \big). \end{aligned}

• For (S3), we have

  \begin{aligned} (a,x) \tilde{\ast} \rho_{\mathscr{F} \rtimes X} \big((b,y)\big) &= (a,x) \tilde{\ast} (\eta_y(b), \rho_X(y))\\ &= (\phi_{x, \rho_X(y)}(a)+\psi_{x,\rho_X(y)}(\eta_y(b)), x \ast \rho_X(y)),\\ && \textrm{by $(X, \rho_X)$-module axioms \hyperref[M6]{(M6)} and \hyperref[M8]{(M8)}, and $\rho_X$ being a good involution}\\ &= (\phi_{x \ast^{-1}y,y}^{-1}(a-\psi_{x \ast^{-1}y,y}(b)), x \ast^{-1}y)\\ &= (a,x)\tilde{\ast}^{-1}(b,y). \end{aligned}

Thus, \rho_{\mathscr{F} \rtimes X} is a good involution, and hence ({\mathscr{F} \rtimes X},\rho_{\mathscr{F} \rtimes X}) is a symmetric rack. Finally, suppose that (X,\rho_X) is a symmetric quandle. Then, we have the additional condition \phi_{x,x}(a)+\psi_{x,x}(a)=a for all x\in X and a\in A_x. Thus, we obtain

for all x\in X and a\in A_x, and hence ({\mathscr{F} \rtimes X},\rho_{\mathscr{F} \rtimes X}) is a symmetric quandle.

Let \mathscr{F}=(A,\phi,\psi,\eta) be an (X,\rho_X)-module. Then, we have the semi-direct product ({\mathscr{F} \rtimes X},\rho_{\mathscr{F} \rtimes X}), which is a symmetric rack due to Proposition 4.1. Let p: \mathscr{F} \rtimes X \rightarrow X be the natural projection given as p(a,x)= x for a \in A_x and x \in X. Then p is a homomorphism of symmetric racks. We claim that p: \mathscr{F} \rtimes X \rightarrow X has the canonical structure of an abelian group object. Let

be the categorical product of p: \mathscr{F} \rtimes X \rightarrow X with itself in the slice category over (X,\rho_X). The symmetric rack structure is defined in the same way as in the cartesian product. We define the morphisms

• \mu:(\mathscr{F} \rtimes X )\times_X (\mathscr{F} \rtimes X) \rightarrow (\mathscr{F} \rtimes X) given by ((a_1,x),(a_2,x)) \mapsto (a_1+a_2,x),

• \sigma:X \rightarrow (\mathscr{F} \rtimes X) given by x \mapsto (0,x),

• \nu:(\mathscr{F} \rtimes X) \rightarrow (\mathscr{F} \rtimes X) given by (a,x) \mapsto (-a,x),

for all a_1,a_2, a \in A_x and x \in X. It easy to check that \mu, \sigma and \nu are symmetric rack homomorphisms. Also, the fact that \mu, \sigma and \nu are slice morphisms and that they turn p: \mathscr{F} \rtimes X \rightarrow X into an abelian group object follows directly (see also [MR2155522, Theorem 2.2]).

Let us define a functor \mathcal{T}: \mathbf{SRMod}_{(X, \rho_X)} \rightarrow \mathbf{Ab}(\mathbf{SR}|_{(X,\rho_X)}). Given an (X,\rho_X)-module \mathscr{F}=(A,\phi,\psi,\eta), let \mathcal{T}(\mathscr{F}) denote the object p:\mathscr{F}\rtimes X\rightarrow X. Let \mathscr{F'}=(A',\phi',\psi',\eta') be another (X,\rho_X)-module and f:\mathscr{F}\rightarrow \mathscr{F'} an (X, \rho_X)-map. Define \mathcal{T}(f):\mathscr{F}\rtimes X \rightarrow \mathscr{F'}\rtimes X by

Then, we have

where \star is the rack operation in \mathscr{F}\rtimes X and \star' is the rack operation in \mathscr{F'}\rtimes X. Further, since \eta'_{x}f_x = f_{\rho_X(x)} \eta_{x}, the diagram

commutes. Thus, T(f) is a homomorphism of symmetric racks. Finally, since the diagram

commutes, where p' is the projection to the second coordinate, \mathcal{T}(f) is a slice morphism. In fact, one can check that \mathcal{T} is a functor (see [MR2155522, Theorem 2.2]).

Let p:(Y,\rho_Y) \rightarrow (X, \rho_X) be an abelian group object in \mathbf{Ab}(\mathbf{SR}|_{(X,\rho_X)}). Let R_x=p^{-1}(x) for each x \in X. Then an abelian group structure on R_x is given by

where s(x) is the identity element in R_x and -u:=i(u) for all u,v \in R_x. To define an (X, \rho_X)-module structure, let us define

and

for all u\in R_x, v \in R_y and x,y \in X. It has been proved in [MR2155522, Theorem 2.2] that \phi_{x,y} and \psi_{x,y} are group homomorphisms for all x, y \in X. For the case of \eta_x, if u_1,u_2 \in R_x, then we see that

Next, we need to check that \phi_{x,y}, \psi_{x,y} and \eta_x satisfy the (X,\rho_X)-module identities. Since (M1), (M2) and (M7) are proved in [MR2155522, Theorem 2.2], we verify only the remaining identities.

• Since \rho_Y is a good involution, we have

  \eta_{\rho_X(x)} \eta_x(u)=\eta_{\rho_X(x)}(\rho_Y(u))=\rho_Y^2(u)=u,

  and hence (M3) holds.

• Again, since \rho_Y is a good involution, we have

  \phi_{\rho_X(x),y}(\eta_x(u))=\phi_{\rho_X(x),y}(\rho_Y(u))=\rho_Y(u)\star s(y)=\rho_Y(u \star s(y))=\eta_{x \ast y}(u \star s(y))=\eta_{x \ast y} (\phi_{x,y}(u)),

  and hence (M4) holds.

• Since s is a homomorphism of symmetric racks, we see that

  \psi_{\rho_X(x),y}(v)= s(\rho_X(x))\star v=\rho_Y(s(x)) \star v=\rho_Y(s(x)\star v)=\eta_{x \ast y}(s(x) \star (v))=\eta_{x \ast y}(\psi_{x,y}(v)),

  and hence (M5) holds.

• Since \rho_Y is a good involution and s is a homomorphism of symmetric racks, we obtain

  \phi_{x \ast^{-1}y,y}\phi_{x, \rho_X(y)}(u)=\phi_{x \ast^{-1},y}(u \star s(\rho_X(y)))=\phi_{x \ast^{-1},y}(u \star \rho_Y(s(y)))=(u \star \rho_Y(s(y)))\star s(y)=u,

  and hence (M6) holds.

• Note that \phi_{x\ast^{-1} y,y}\psi_{x,\rho_X(y)}(\eta_{y}(v))=\phi_{x\ast^{-1} y,y}\psi_{x,\rho_X(y)}(\rho_Y(v))=\phi_{x\ast^{-1} y,y}(s(x) \star \rho_Y(v))=(s(x) \star \rho_Y(v) )\star s(y) and \psi_{x\ast \rho_X(y),y}(v)=s(x \ast \rho_X(y))\star v=(s(x) \star \rho_Y(s(y))) \star v. Then, we have

  \begin{aligned} && \phi_{x\ast^{-1} y,y}\psi_{x,\rho_X(y)}(\eta_{y}(v))+\psi_{x\ast \rho_X(y),y}(v)\\ &= (s(x) \star \rho_Y(v)) \star s(y)+(s(x) \star \rho_Y(s(y))) \star v\\ &=m \big((s(x) \star \rho_Y(v)) \star s(y), ~(s(x) \star \rho_Y(s(y))) \star v \big)\\ &=m \big( \big((s(x), s(x)) \star (\rho_Y(v), \rho_Y(s(y))) \big) \star (s(y), v) \big), \quad \textrm{by definition of rack operation in $Y \times_X Y$}\\ &= \big(m(s(x),s(x)) \star m \,(\rho_Y(v),\rho_Y(s(y))) \big)\star m(s(y),v), \quad \textrm{since $m$ is a homomorphism of racks}\\ &= \big(s(x) \star \rho_Y \, m(v, s(y))) \star m(s(y),v), \quad \textrm{since $m$ is a homomorphism of symmetric racks}\\ &= \big(s(x) \star \rho_Y \, m(s(y),v) )\star m(s(y),v), \quad \textrm{since $\rho_Y$ is a good involution}\\ &= s(x), \end{aligned}

and hence (M8) holds. Hence, we have proved that (R, \phi, \psi, \eta) is an (X,\rho_X)-module.

Next, we construct a functor \mathcal{S}: \mathbf{Ab}(\mathbf{SR}|_{(X,\rho_X)}) \to \mathbf{SRMod}_{(X, \rho_X)}. Given an abelian group object p:(Y,\rho_Y) \rightarrow (X, \rho_X), we set \mathcal{S}(p) = (R,\phi,\psi,\eta), which is defined above. Let f be a morphism from the abelian group object p_1:(Y_1,\rho_{Y_1}) \rightarrow (X,\rho_X) to the abelian group object p_2:(Y_2,\rho_{Y_2}) \rightarrow (X,\rho_X). Thus, if m and m' denote the corresponding multiplication maps, then f \, m= m' \,(f, f). Further, f:(Y_1,\rho_{Y_1}) \rightarrow (Y_2,\rho_{Y_2}) is a homomorphism of symmetric racks with p_2\,f=p_1. Let \mathscr{F}_1 and \mathscr{F}_2 denote the (X,\rho_X)-modules as described above. Define g:\mathscr{F}_1=(R_1,\phi,\psi,\eta) \rightarrow \mathscr{F}_2=(R_2,\phi',\psi',\eta') by setting

for all x\in X and u \in {(R_1)}_x. Since f \, m= m' \,(f, f), it follows that g_x is an abelian group homomorphism for each x \in X. The identities \phi'_{x,y}g_x = g_{x\ast y} \phi_{x,y} and \psi'_{x,y}g_y = g_{x\ast y}\psi_{x,y} are proved in [MR2155522, Theorem 2.2]. Further, since

for all w\in (R_1)_x, we have \eta'_{x}g_x = g_{\rho_X(x)} \eta_{x}. Hence, g is an (X,\rho_X)-map. A routine check shows that \mathcal{S} is a functor (see [MR2155522, Thereom 2.2]).

Let (X,\rho_X) be a symmetric rack. By Proposition 4.2 and Proposition 4.3, we have functors \mathcal{T}: \mathbf{SRMod}_{(X, \rho_X)} \to \mathbf{Ab}(\mathbf{SR}|_{(X,\rho_X)}) and \mathcal{S}: \mathbf{Ab}(\mathbf{SR}|_{(X,\rho_X)}) \to \mathbf{SRMod}_{(X, \rho_X)}. To show that the categories \mathbf{SRMod}_{(X, \rho_X)} and \mathbf{Ab}(\mathbf{SR}|_{(X,\rho_X)}) are equivalent, we show that \mathcal{T}\mathcal{S} is naturally isomorphic to \mathrm{id}_{\mathbf{Ab}(\mathbf{SR}|_{(X,\rho_X)})} and \mathcal{S}\mathcal{T} is naturally isomorphic to \mathrm{id}_{ \mathbf{SRMod}_{(X, \rho_X)}}.

Let \mathcal{O} \in \mathbf{Ab}(\mathbf{SR}|_{(X,\rho_X)}) denote the object p:(Y,\rho_Y) \rightarrow (X, \rho_X) with the multiplication map m, the inverse map i, and the section s. Then \mathcal{T}\mathcal{S} (\mathcal{O}) is the object q:(\mathscr{R} \rtimes X, \rho_{\mathscr{R} \rtimes X}) \rightarrow (X, \rho_X) in the category \mathbf{Ab}(\mathbf{SR}|_{(X,\rho_X)}), where \mathscr{R}=(R, \phi,\psi,\eta) as constructed in Proposition 4.3. Let * be the rack operation on X, and \tilde{*} the rack operation on \mathscr{R} \rtimes X. Note that, the abelian group structure on q:(\mathscr{R} \rtimes X, \rho_{\mathscr{R} \rtimes X}) \rightarrow (X, \rho_X) has the multiplication, the inverse and the section given by

• \mu ((a_1,x),(a_2,x)) = (a_1+a_2,x)=(m(a_1,a_2),x),

• \sigma (x) = (0,x)=(s(x),x),

• \nu((a,x)) = (-a,x)=(i(a),x),

for all a_1,a_2, a \in R_x and x \in X. Define \zeta_{\mathcal{O}}: (Y, \rho_Y) \rightarrow (\mathscr{R} \rtimes X, \rho_{\mathscr{R} \rtimes X}) by setting \zeta_{\mathcal{O}}(a)=(a,x) for a \in R_x. Clearly, \zeta_{\mathcal{O}} is a bijection and the diagram

commutes. Note that \zeta_{\mathcal{O}}(m(a,b))=\mu(\zeta_{\mathcal{O}}(a), \zeta_{\mathcal{O}}(b)),~ \zeta_{\mathcal{O}}(s(a))=\sigma(\zeta_{\mathcal{O}}(a)) and \zeta_{\mathcal{O}}(i(a))=\nu(\zeta_{\mathcal{O}}(a)) for all a,b \in R_x. Hence, \zeta_{\mathcal{O}} respects the abelian group structure. Now, for a \in R_x and b \in R_y, we have

where \star is the rack operation on Y and \tilde{\star} is the rack operation on Y \times Y. Thus, \zeta_{\mathcal{O}} is a rack homomorphism. Since \rho_{\mathscr{R}\rtimes X} \, \zeta_{\mathcal{O}}=\zeta_{\mathcal{O}} \, \rho_Y, it follows that \zeta_{\mathcal{O}} is a symmetric rack isomorphism, and hence \zeta_{\mathcal{O}}: \mathcal{O} \to \mathcal{T}\mathcal{S} (\mathcal{O}) is an isomorphism.

Let \mathcal{O} and \mathcal{O'} be the objects p:(Y,\rho_Y) \rightarrow (X, \rho_X) and p':(Y',\rho_{Y'}) \rightarrow (X,\rho_X) in \mathbf{Ab}(\mathbf{SR}|_{(X,\rho_X)}), respectively. Let f:\mathcal{O} \rightarrow \mathcal{O'} be a morphism. Then, by definition, f: (Y, \rho_Y) \to (Y', \rho_{Y'}) is a symmetric rack homomorphism such that p' \circ f=p and f preserves the abelian group structure. Let x \in X and a \in R_x. Then \zeta_{\mathcal{O'}} \, f(a)=(f_x(a),x), where f_x=f|_{R_x} and

Thus, the diagram

commutes, and we have shown that \mathcal{T}\mathcal{S} is naturally isomorphic to \mathrm{id}_{ \mathbf{Ab}(\mathbf{SR}|_{(X,\rho_X)})}. Next, we prove that \mathcal{S}\mathcal{T} is naturally isomorphic to \mathrm{id}_{\mathbf{SRMod}_{(X, \rho_X)}}. Let \mathcal{O} be the object \mathscr{F}=(A,\phi,\psi,\eta) in \mathbf{SRMod}_{(X, \rho_X)}. Then \mathcal{T}(\mathcal{O}) is the abelian group object p:(\mathscr{F} \rtimes X,\rho_{\mathscr{F} \rtimes X}) \rightarrow (X, \rho_X) as in Proposition 4.2. For each x \in X, since A'_{x}:=p^{-1}(x)=(A_x,x), it follows that \mathcal{S}\mathcal{T}(\mathcal{O}) is the object \mathscr{F'}:=(A',\phi',\psi',\eta'), where

for all a\in A_x, b \in A_y and x,y \in X. We now construct an (X,\rho_X)-module isomorphism \lambda_{\mathcal{O}}: \mathcal{O} \to \mathcal{S}\mathcal{T}(\mathcal{O}). For each x \in X, define (\lambda_{\mathcal{O}})_x:A_x \to A'_x as a \mapsto (a,x) for all a \in A_x. Clearly, (\lambda_O)_{x} is an abelian group isomorphism for each x \in X. Also, using (?), (?) and (?), we have

for all a\in A_x, b \in A_y and x,y \in X. Thus, (3.0.1-3.0.3) hold, and \lambda_O is an (X, \rho_X)-module isomorphism. Finally, let \mathcal{O}_1 and \mathcal{O}_2 be the objects \mathscr{F}_1=(A,\phi,\psi,\eta) and \mathscr{F}_2=(\tilde{A},\tilde{\phi},\tilde{\psi},\tilde{\eta}) in \mathbf{SRMod}_{(X, \rho_X)}, respectively. Let g:\mathcal{O}_1 \rightarrow \mathcal{O}_2 be a (X, \rho_X)-map. It is easy to see that the diagram

commutes. Thus, \mathcal{S}\mathcal{T} is naturally isomorphic to \mathrm{id}_{\mathbf{SRMod}_{(X, \rho_X)}}, and the categories \mathbf{Ab}(\mathbf{SR}|_{(X,\rho_X)}) and \mathbf{SRMod}_{(X, \rho_X)} are equivalent.

Suppose that \mathrm{(F1)-(F3)} hold. That (E(\mathscr{F},\sigma),\rho_{E(\mathscr{F},\sigma)}) is a rack is proved in [MR2155522, Proposition 3.1]. We check that \rho_{E(\mathscr{F},\sigma)} is a good involution.

• For (S1), we have

  \rho_{E(\mathscr{F},\sigma)}^2((a,x))=\rho_{E(\mathscr{F},\sigma)}(\eta_x(a), \rho_X(x))=(\eta_{\rho_X(x)}(\eta_x(a)),\rho_X^2(x))=(a,x).

• For (S2), we see that

  \begin{aligned} \big(\rho_{E(\mathscr{F},\sigma)}(a,x) \big)\tilde{\star} (b,y) &= \big(\eta_x(a), \, \rho_X(x)\big) \tilde{\star} (b,y)\\ &= \big(\phi_{\rho_X(x),y}(\eta_x(a))+\psi_{\rho_X(x),y}(b)+\sigma_{\rho_X(x),y}, \,\rho_X(x) \ast y \big)\\ &=\big( \eta_{x \ast y}(\phi_{x,y}(a)+\psi_{x,y}(b)+\sigma_{x,y}), \, \rho_X(x \ast y) \big), \quad \textrm{by (M4), (M5) and (F2)}\\ &= \rho_{E(\mathscr{F},\sigma)} \big(\phi_{x,y}(a)+\psi_{x,y}(b)+\sigma_{x,y}, \, x \ast y \big)\\ &= \rho_{E(\mathscr{F},\sigma)} \big((a,x)\tilde{\star} (b,y) \big). \end{aligned}

• For (S3), we have

  \begin{aligned} (a,x) \tilde{\star} \rho_{E(\mathscr{F},\sigma)}(b,y) &= (a,x)\tilde{\star} \big(\eta_y(b), \,\rho_X(y) \big)\\ &= \big(\phi_{x,\rho_X(y)}(a)+\psi_{x, \rho_X(y)}(\eta_y(b))+\sigma_{x, \rho_{X}(y)}, \, x \ast \rho_X(y) \big)\\ &= \big(\phi_{x \ast^{-1}y,y}^{-1}(a-\sigma_{x\ast^{-1}y,y}-\psi_{x \ast^{-1}y,y}(b)), \, x \ast^{-1}y \big), \quad \textrm{by (M6), (M8) and (F3)}\\ &= (a,x)\tilde{\star}^{-1}(b,y). \end{aligned}

Hence, (E(\mathscr{F},\sigma),\rho_{E(\mathscr{F},\sigma)}) is a symmetric rack. Define f_1:(E(\mathscr{F},\sigma),\rho_{E(\mathscr{F},\sigma)}) \rightarrow (X,\rho_X) to be the projection onto the second coordinate. Let A_x act on E(\mathscr{F},\sigma)_x =f_1^{-1}(x) by a\cdot (b,x)=(a+b,x) for each x \in X and a, b \in A_x. Note that these actions are free and transitive. Also, as proved in [MR2155522, Proposition 3.1], the actions satisfy

for all \alpha \in A_x, \beta \in A_y, (a,x) \in E(\mathscr{F},\sigma)_x and (b, y) \in E(\mathscr{F},\sigma)_x. Further, we have

Thus, (E(\mathscr{F},\sigma),\rho_{E(\mathscr{F},\sigma)}) is an extension of (X,\rho_X) by \mathscr{F}. Define s:X \rightarrow E(\mathscr{F},\sigma) as s(x )= (0,x). This is clearly a section satisfying

and

Thus, \sigma is the factor set corresponding to the section s. The converse follows by reversing the steps.

For the second assertion, let f_2:(E,\rho_E) \rightarrow (X, \rho_X) be an extension of (X,\rho_X) by \mathscr{F} with factor set \sigma corresponding to a section s:(X,\rho) \rightarrow (E,\rho_E). Define

as \Phi \big((a,x)\big)= a\cdot s(x). Since, A_x acts freely and transitively, it follows that \Phi is a bijection. Also, using (E3), we have

Hence, \Phi is a rack isomorphism. Further, we see that

for all x \in X and a \in A_x. Hence, \Phi is an isomorphism of symmetric racks. Clearly, f_2 \Phi(a, x)=f_2(a \cdot s(x))=x=f_1(a, x) for all (a, x) \in E(\mathscr{F},\sigma)_x and \Phi(\alpha \cdot (a,x))=\Phi((\alpha+a,x))=(\alpha+a)\cdot s(x)=\alpha\cdot \Phi(a,x) for all \alpha, a \in A_x and x \in X. Hence, \Phi is an equivalence of extensions. Finally, if (X, \rho_X) is a symmetric quandle, then the condition (F4) follows.

First, we prove (1) \Leftrightarrow (2). Let \Phi:(E(\mathscr{F},\sigma),\rho_{E(\mathscr{F},\sigma)}) \rightarrow (E(\mathscr{F},\tau),\rho_{E(\mathscr{F},\tau)}) be an equivalence of extensions. Since \Phi respects the projections, we obtain \Phi(0,x) = (v_x,x) for some v_x \in A_x. Further, since \Phi respects the group actions, this gives \Phi(a,x)=\Phi(a \cdot (0,x))=a \cdot \Phi(0,x)=a \cdot (v_x,x)=(a+v_x,x) for all a \in A_x. Then,

and

where \tilde{\star} and \bar{\star} are the rack operations in E(\mathscr{F},\sigma) and E(\mathscr{F},\tau), respectively. Thus, \Phi \big((a,x)\bar{\star}(b,y) \big)=\Phi(a,x)\bar{\star} \Phi(b,y) if and only if \tau_{x,y}=\sigma_{x,y}+\phi_{x,y}(v_x)+\psi_{x,y}(v_y)-v_{x\ast y} for x,y \in X. We have

and

Thus, \Phi \rho_{E(\mathscr{F},\sigma)}=\rho_{E(\mathscr{F},\tau)}\Phi if and only if \eta_x(v_x)=v_{\rho_X(x)}. Next, we prove (1) \Leftrightarrow (3). We shall use (1) \Leftrightarrow (2) proved above. Obviously, \sigma is the factor set corresponding to the section s:(X,\rho_X) \rightarrow (E(\mathscr{F},\sigma),\rho_{E(\mathscr{F},\sigma)}) given by s(x) = (0,x). Define t:(X,\rho_X) \rightarrow (E(\mathscr{F},\sigma),\rho_{E(\mathscr{F},\sigma)}) by t(x) = (v_x,x). Since

it follows that t is a section. Note that, we have \tau_{x,y}=\sigma_{x,y}+\phi_{x,y}(v_x)+\psi_{x,y}(v_y)-v_{x\ast y}. Then, we see that

Thus, \tau is the factor set corresponding to t. Conversely, suppose that \sigma and \tau are factor sets of some extension (E,\rho_E) of (X,\rho_X) by \mathscr{F} corresponding to sections s: (X,\rho_X) \rightarrow (E,\rho_E) and t:(X,\rho_X) \rightarrow (E,\rho_E), respectively. Then, for each x \in X, we have s(x)=v_x \cdot t(x) for a unique v_x \in A_x. Define \Phi:E(\mathscr{F},\sigma) \rightarrow E(\mathscr{F},\tau) as \Phi \big((a,x) \big)= (a+v_x,x). Then, a direct check implies that \Phi is an equivalence of extensions.

Let Z^2(X,\mathscr{F}) consists of all factor sets \sigma:X \times X \rightarrow \bigsqcup_{x \in X} A_x. Define the binary operation

for all x, y \in X. In view of Theorem 5.4, it follows that the conditions (F1)-(F3) are satisfied by \sigma+\tau, and hence this binary operation turns Z^2(X,\mathscr{F}) into an abelian group with the trivial factor set as the identity element. Further, the set B^2(X,\mathscr{F}) of all coboundaries forms a subgroup of Z^2(X,\mathscr{F}). Define \mathcal{H}^2_{\mathrm{SR}}(X,\mathscr{F}):=Z^2(X,\mathscr{F})/B^2(X,\mathscr{F}). Then, by Theorem 5.4 and Proposition 5.5, \mathcal{H}^2_{\mathrm{SR}}(X,\mathscr{F}) is in bijective correspondence with the set of equivalence classes of extensions of (X,\rho_X) by \mathscr{F}.

Using the fact that \rho_X is a good involution and the rack identity

for each (x_1, x_2, \ldots, x_n) \in X^n, it follows from the condition (A5) that

Similarly, it follows from (A6) that

Then, we have

Hence, \partial_n \big(U_n \big) lies in D_{n-1}^{SR}(X,\rho_X).

For a fixed 2 \le i \le n, we have

and

We see that

Then, using (Proof), the fact that \rho_X is a good involution and the condition (A2), we see that

Thus, the term 1A + 2A lies in D_{n-1}^{SR}(X,\rho_X). Next, we consider

Due to condition (A7), we have

and hence 1B + 2E =0. Consider the term

Using the condition (A2), we see that

Thus, 1C + 2C lies in D_{n-1}^{SR}(X,\rho_X). Next, we consider

Since \phi_{[x_1 \cdots \widehat{x_i} \cdots x_n],[x_i \cdots x_n]}=\phi_{[(x_1 \ast x_j) \cdots (x_{j-1} \ast x_j) x_{j+1} \cdots \widehat{x_i} \cdots x_n],[x_i \cdots x_n]}, it follows that 1D + 2D lies in D_{n-1}^{SR}(X,\rho_X). Next, we have

Consider the term

Since \phi_{[(x_1\ast x_j) \cdots \widehat{(x_i \ast x_j)} \cdots (x_{j-1}\ast x_j) x_{j+1} \cdots x_n)],[(x_i \ast x_j) \cdots (x_{j-1}\ast x_j) x_{j+1} \cdots x_n]}=\phi_{[x_1 \cdots \widehat{x_i} \cdots x_n],[x_i \cdots x_n]}, it follows that 1F + 2F lies in D_{n-1}^{SR}(X, \rho_X). Next, consider the term

If i>2, then using the condition (A3), we get

Thus, in this case, 1G + 2G lies in D_{n-1}^{SR}(X, \rho_X). If i=2, then the condition (A9) gives

Hence, 1G + 2G lies in D_{n-1}^{SR}(X, \rho_X), and we have proved that \partial_n(V_n) \subset D_{n-1}^{SR}(X,\rho_X).

In view of Lemma 6.4 and Lemma 6.5, it remains to verify that \partial_1(D_{1}^{SR}(X, \rho_X))={0}. Note that, D_{1}^{SR}(X, \rho_X) is generated by \{\eta_{x}(x)-\rho_X(x) \mid x \in X \}. It follows from the condition (A6) that

for all x \in X. Hence, \partial_n(D_{n}^{SR}(X,\rho)) \subset D_{n-1}^{SR}(X,\rho) for all n \ge 0.

It suffices to prove that \partial_n(x_1, \ldots, x_{i-1},x_i,x_i,x_{i+2},\ldots, x_n) \in D^{Q}_{n-1}(X,\rho_X) for each 1 \leq i \leq n-1. For i=1, we have

where the last equality follows due to

which itself is a consequence of the condition (A10). For i \geq 2, we have

where the i-th and the (i+1)-th terms cancel out due to the equality

Hence, \partial_n(D^{SQ}_{n}(X,\rho_X)) \subset D^{SQ}_{n-1}(X,\rho_X), and the proof is complete.

For x, y, z \in X and f \in \operatorname{Hom}(X,A), we see that

and

Thus, the left action of G_{(X,\rho_X)} on \operatorname{Hom}(X,A) is well-defined.

Let f:G_{(X,\rho_X)} \rightarrow \operatorname{Hom}(X,A) be a group 1-cocycle. Define \theta_f:X \times X \rightarrow A by

We claim that \theta_f is a symmetric rack 2-cocycle. For all x,y,z \in X, we have

Further, since f(e_y) is a symmetric rack homomorphism, we have

Since f is a group 1-cocycle, we get

and hence \theta_f is a symmetric rack 2-cocycle. This gives a map Z^1 \big(G_{(X,\rho_X)}, \operatorname{Hom}(X,A)\big) \rightarrow H^2_{SR} \big((X,\rho_X),A \big) defined by f \mapsto [\theta_f]. Since the group structure in both the domain and the codomain is the point-wise addition of maps, it follows that f \mapsto [\theta_f] is a group homomorphism. If f \in B^1(G_{(X,\rho_X)}, \operatorname{Hom}(X,A)), then for all a \in G_{(X,\rho_X)} and x\in X, we have

for some \phi \in \operatorname{Hom}(X,A). Thus, for all x,y \in X, we have

This shows that \theta_f \in B^2((X,\rho_X);A), and hence we obtain a map

For the reverse direction, let \sigma \in Z_{SR}^2((X,\rho_X),A). Then \sigma satisfies the symmetric rack 2-cocycle conditions

for all x,y,z \in X. Let us define \sigma':X \rightarrow \operatorname{Hom}(X,A) by

for all x, y \in X. Since \sigma'(x)(\rho_X(y))=\sigma(\rho_X(y),x)=-\sigma(x,y) by the cocycle condition (?), and hence \sigma'(x) \in \operatorname{Hom}(X,A). Identifying X with the set \{e_x\mid x \in X\}, we have a map \tau_{\sigma}:X \rightarrow G_{(X,\rho_X)} \ltimes \operatorname{Hom}(X,A) defined by

for all x \in X. Note that

Further, the symmetric rack 2-cocycle condition (?) gives

Thus, \sigma'(x)+e_x \cdot \sigma'(\rho_X(x)) is the trivial map, and consequently \tau_{\sigma}(e_x)\tau_{\sigma}(e_{\rho_X(x)})=(e,0), where e is the identity of G_{(X,\rho_X)}. Similarly, we can prove that \tau_{\sigma}(e_{\rho_X(x)})\tau_{\sigma}(e_x)=(e,0). Next, we consider

The symmetric rack 2-cocycle condition (?) gives

It follows that \tau_{\sigma}(e_y) \tau_{\sigma}(e_{x*y}) \tau_{\sigma}(e_y)^{-1}\tau_{\sigma}(e_x)^{-1}=(e,0) for all x, y \in X, and hence we can extend \tau_{\sigma} to a group homomorphism \tau_{\sigma}:G_{(X,\rho_X)} \rightarrow G_{(X,\rho_X)} \ltimes \operatorname{Hom}(X,A). Let us define \chi_{\sigma}:G_{(X,\rho_X)} \to \operatorname{Hom}(X,A) to be the composition

where p_2:(G_{(X.\rho_X)} \ltimes \operatorname{Hom}(X,A)) \rightarrow \operatorname{Hom}(X,A) is the projection onto the second factor. Since \tau_{\sigma} is a group homomorphism, it follows that \chi_{\sigma} is a group 1-cocycle. This gives a map Z^2_{SR}\big((X,\rho_X),A \big) \rightarrow H^1 \big(G_{(X,\rho_X)}, \operatorname{Hom}(X,A)\big) defined by \sigma \mapsto [\chi_{\sigma}]. If \pi \in B^2_{SR}((X,\rho_X),A), then \pi=\delta^1( \tilde{\pi}) for some \tilde{\pi}\in C^1_{SR}((X,\rho_X),A), that is,

for all x,y \in X. If g \in G_{(X,\rho_X)}, then g=e_{x_1}^{\epsilon_1}e_{x_2}^{\epsilon_2} \cdots e_{x_n}^{\epsilon_n} for some x_i \in X and \epsilon_i\in \{1,-1\}. Thus, for all y \in X, we have

where t(x_i)=\pi'(x_i) for \epsilon_i=1 and t(x_i)=e_{x_i}^{-1} \cdot (-\pi'(x_i)) for \epsilon_i=-1. For instance, if all \epsilon_i=1, then

The cases when some \epsilon_i=-1 can be dealt with in a similar way through a routine computation. Thus, \chi_{\pi}(g)=\tilde{\pi}-g \cdot \tilde{\pi}, which shows that \chi_{\pi} \in B^1(G_{(X,\rho_X)}, \operatorname{Hom}(X,A)). Hence, we obtain a map

and it is easy to see that \Psi and \Phi are inverses to each other, and hence H^2_{SR}\big((X,\rho_X),A\big) \cong H^1 \big(G_{(X,\rho_X)}, \operatorname{Hom}(X,A)\big), which is desired.