Fischer’s approach to deformation of coactions

Suppose that \|\cdot\|_\mu and A\rtimes_{\delta}\widehat{G}\rtimes_{\widehat\delta,\mu}G are as above. Then the double dual coaction

factors through a (double dual) coaction \widehat{\widehat{\delta}\,}_{\!\!\mu} of G on A\rtimes_{\delta}\widehat{G}\rtimes_{\widehat\delta,\mu}G which corresponds to the coaction \widetilde{\delta} on A\otimes \mathcal K via the isomorphism \Psi_\mu.

Suppose that \delta:A\to \mathcal M(A\otimes C^*(G)) is a maximal coaction. Then, for every duality crossed-product A\rtimes_\delta\widehat{G}\rtimes_{\widehat{\delta},\mu}G, there exists a unique quotient A_\mu of A such that \delta factors through a \mu-coaction \delta_\mu:A_\mu\to\mathcal M(A_\mu\otimes C^*(G)) and such that the canonical induced map A\rtimes_\delta \widehat{G}\twoheadrightarrow A_\mu\rtimes_{\delta_\mu}\widehat{G} is an isomorphism.

Suppose that i_\mathcal K:\mathcal K\to \mathcal M(E) is a nondegenerate *-homomorphism and that \epsilon: E\to\mathcal M(E\otimes C^*(G)) is a coaction such that \epsilon(i_\mathcal K(k))=i_\mathcal K(k)\otimes 1 for all k\in \mathcal K (i.e., \epsilon is trivial on i_\mathcal K(\mathcal K)). Then \epsilon restricts to a coaction \delta: A\to \mathcal M(A\otimes C^*(G)) with A=C(E,i_\mathcal K) as in (?), such that \epsilon corresponds under the isomorphism E\cong A\otimes\mathcal K to the coaction \delta\otimes_*\operatorname{\mathrm{id}}_{\mathcal K} defined by

where \sigma: C^*(G)\otimes \mathcal K\to \mathcal K\otimes C^*(G) denotes the flip map.

Let L^2(G_\sigma, \iota) denote the subspace of L^2(G_\sigma) consisting of all elements \xi\in L^2(G_\sigma) which satisfy \xi({\tilde{g}} z)=z\xi({\tilde{g}}) for all {\tilde{g}}\in G_\sigma, z\in \mathbb T. Then the pair (M^\sigma, \rho^\sigma) given by

is a covariant representation for the twisted action (\mathrm{rt},\iota^\sigma):(G_\sigma,\mathbb T)\curvearrowright C_0(G) whose integrated form induces an isomorphism

Let W^\sigma=M^\sigma\otimes\operatorname{\mathrm{id}}_G(w_G)\in U\mathcal M(\mathcal K(L^2(G_\sigma, \iota))\otimes C^*(G)). Then, the isomorphism M^\sigma\rtimes \rho^\sigma: C_0(G)\rtimes_{(\mathrm{rt},\iota)}G\to \mathcal K(L^2(G_{\sigma}, \iota)) identifies the dual coaction \widehat{(\mathrm{rt},\iota^\sigma)} with the coaction k\mapsto (W^\sigma)^*(k\otimes 1)W^\sigma of G on \mathcal K(L^2(G_\sigma,\iota)).

As a consequence, if (B,\beta,\phi) is a weak G\rtimes G-algebra, then

is a one-cocycle for the dual coaction \widehat{(\beta,\iota^\sigma)} on B\rtimes_{(\beta,\iota)}G such that the coaction \epsilon:=\textup{Ad} W_B\circ \widehat{(\beta,\iota^\sigma)} fixes the image \phi\rtimes G\big(C_0(G)\rtimes_{(\mathrm{rt},\iota^\sigma)}G)\big) in \mathcal M(B\rtimes_{(\beta,\iota^\sigma)}G).

Let \mathcal X be an A-B equivalence bimodule and suppose that S\in \mathcal M(\mathcal X) such that S^*S=1_{\mathcal M(B)} and SS^*=1_{\mathcal M(A)}. Then B\cong SBS^*=A via b\mapsto SbS^*. Here all multiplications are inside the linking algebra L(\mathcal M(\mathcal X)).

If, in addition, \delta_\mathcal X:\mathcal X\to \mathcal M(\mathcal X\otimes C^*(G)) is a coaction of G on \mathcal X which implements a Morita equivalence between the coactions (A,\delta_A) and (B,\delta_B), and such that

(using the unique extension of \delta_\mathcal X to \mathcal M(\mathcal X)) then the above isomorphism \textup{Ad} S: B\xrightarrow\sim A is \delta_B-\delta_A equivariant.

The representation (M\rtimes\rho, L^\sigma\rtimes\rho^\sigma, M^\sigma\rtimes \rho^\sigma) identifies the C_0(G)\rtimes_{\mathrm{rt}}G-C_0(G)\rtimes_{(\mathrm{rt},\iota)}G imprimitivity bimodule C_0(G_\sigma, \bar\iota)\rtimes_{\mathrm{rt}^\sigma}G with the \mathcal K(L^2(G))-\mathcal K(L^2(G_\sigma,\iota)) imprimitivity bimodule \mathcal K(L^2(G), L^2(G_\sigma,\iota)) and, therefore, it extends to an isomorphism of multiplier bimodules

Let L_\varphi: L^2(G_\sigma,\iota)\to L^2(G) and S_\varphi\in \mathcal M(C_0(G)\rtimes_{(\mathrm{rt},\iota^\sigma)}G) be as above. Then

As a consequence, the preimage S_\varphi\in \mathcal M(C_0(G_\sigma,\bar\iota)\rtimes_{\mathrm{rt}^\sigma}G) of L_\varphi under the isomorphism \mathcal M(C_0(G_\sigma,\bar\iota)\rtimes_{\mathrm{rt}^\sigma}G)\cong \mathcal B(L^2(G_\sigma,\iota), L^2(G)) is fixed by the dual coaction \widehat{\mathrm{rt}^\sigma} and the isomorphism \textup{Ad} S_\varphi: C_0(G)\rtimes_{(\mathrm{rt},\iota^\sigma)}G\xrightarrow\sim C_0(G)\rtimes_\mathrm{rt} G of Lemma 5.1 is equivariant for the dual coactions.

Let (B,\beta,\phi) be a weak G\rtimes G-algebra and let \sigma=(\mathbb T\hookrightarrow G_\sigma\twoheadrightarrow G) be a twist for G. Then the element R_\varphi\in \mathcal M(\mathcal E_\sigma(B)\rtimes_\gamma G) constructed above induces a \widehat{(\beta,\iota^\sigma)}-\widehat{\beta^\sigma} equivariant *-isomorphism

such that following diagram commutes:

Let (A,\delta) be a maximal coaction and let \sigma=(\mathbb T\hookrightarrow G_\sigma\twoheadrightarrow G) be a twist for G. Then the maximal deformation (A^\sigma, \delta^\sigma) and the maximal Fischer deformation (D^\sigma, \epsilon^\sigma) are equivariantly isomorphic.

As a consequence (using Proposition 2.6) the same holds true for the \mu-deformations (A_\mu^\sigma, \delta_\mu^\sigma) and (D_\mu^\sigma, \epsilon_\mu^\sigma) for any duality crossed-product functor \rtimes_\mu.