---
title: Packer–Raeburn Trick in Crossed Products
url: https://www.emergentmind.com/topics/packer-raeburn-trick
type: topic
---

# Packer–Raeburn Trick in Crossed Products

Searching arXiv for recent and foundational papers on the Packer–Raeburn trick and its extensions.
The Packer–Raeburn trick is a stabilization and untwisting mechanism for twisted crossed products. In its classical form for a locally compact group $G$ acting on a $C^*$-algebra $A$ by a twisted action $(\alpha,\sigma)$, the trick asserts that the cocycle $\sigma$ can be absorbed after tensoring with compact operators, so that the stabilized twisted crossed product becomes an untwisted crossed product by an ordinary action [2509.24106]. In later developments, this mechanism was extended from groups to Fell bundles over Hausdorff groupoids [1512.06046], used as a technical bridge in the analysis of reduced twisted crossed products [2109.08606], and shown to fail in general for non-Hausdorff étale groupoids [1410.2051]. A recent Banach and $L^p$-operator algebra formulation establishes an $L^p$ analogue in which $\mathcal{K}(L^p(G))$ replaces $\mathcal{K}(L^2(G))$, unitaries are replaced by invertible isometries, and the stabilized twisted $L^p$-crossed product is isometrically isomorphic to an untwisted one [2509.24106].

## 1. Classical formulation for twisted group actions

In the classical $C^*$-algebraic setting, let $G$ be a locally compact group, $A$ a $C^*$-algebra, and $(\alpha,\sigma)$ a twisted action, where $\alpha:G\to \operatorname{Aut}(A)$ is strongly continuous and $\sigma:G\times G\to UM(A)$ is a strictly continuous $2$-cocycle with values in the unitary multipliers [2509.24106]. The associated twisted crossed product $A\rtimes_{\alpha,\sigma}G$ is generated by a universal covariant representation $(\pi,u)$ satisfying
\[
u_xu_y=\widehat{\pi}(\sigma(x,y))u_{xy},\qquad
\pi(\alpha_x(a))=u_x\pi(a)u_x^{-1}.
\]

The classical stabilization statement says that there exists a strongly continuous action $\beta$ of $G$ on $\mathcal{K}(L^2(G))\otimes A$ such that
\[
\mathcal{K}(L^2(G))\otimes \big(A\rtimes_{\alpha,\sigma}G\big)
\;\cong\;
\big(\mathcal{K}(L^2(G))\otimes A\big)\rtimes_\beta G.
\]
Moreover, $\beta_x=\operatorname{Ad}(\theta_x)\circ (\operatorname{id}_{\mathcal K}\otimes \alpha_x)$, where $\theta_x$ is constructed from the left regular representation $\lambda_x^2$ and the cocycle $\sigma$ via multiplier operators [2509.24106]. In the formulation used for reduced crossed products over discrete groups, one similarly has
\[
(A\rtimes_{\alpha,u,r}\Gamma)\otimes K(\ell^2\Gamma)\cong (A\otimes K(\ell^2\Gamma))\rtimes_{\beta,r}\Gamma,
\]
with the isomorphism commuting with subgroup conditional expectations [2109.08606].

Conceptually, stabilization converts twisting into an inner perturbation on the compact-operator tensor factor. The cocycle is not removed directly on $A$ itself; rather, it becomes absorbed into a larger algebra where an exterior-equivalent untwisted action exists. This is why the result is inherently “stable” rather than literal untwisting on the original algebra [2509.24106].

## 2. Twisted actions, covariance, and exterior equivalence

The trick is formulated in terms of twisted actions and exterior equivalence. In the Banach algebra framework of "Twisted crossed products of Banach algebras" [2509.24106], a twisted action $(\alpha,\sigma)$ of a locally compact group $G$ on a nondegenerate Banach algebra $A$ with contractive approximate identity consists of a strongly continuous map $\alpha:G\to \operatorname{Aut}(A)$ and a jointly strictly continuous map $\sigma:G\times G\to \operatorname{Inv}_1(M(A))$ such that
\[
\alpha_e={\rm id}_A,\quad \sigma(x,e)=\sigma(e,x)=1,
\]
\[
\alpha_x\circ \alpha_y=\operatorname{Ad}(\sigma_{x,y})\circ \alpha_{xy},
\]
\[
\alpha_z(\sigma_{x,y})\sigma_{z,xy}=\sigma_{z,x}\sigma_{zx,y}.
\]

For such a system, the dense convolution algebra $C_c(G,A)$ carries twisted convolution
\[
(f*_{\alpha,\sigma}g)(t)=\int_G f(s)\,\alpha_s\!\big(g(s^{-1}t)\big)\,\sigma(s,s^{-1}t)\,ds.
\]
A covariant representation $(\pi,u)$ on a Banach space $E$ consists of a nondegenerate representation $\pi:A\to B(E)$ and a strongly continuous map $u:G\to \operatorname{Iso}(E)$ satisfying
\[
u_xu_y=\widehat{\pi}(\sigma_{x,y})u_{xy},\qquad
\pi(\alpha_x(a))=u_x\pi(a)u_x^{-1},
\]
with integrated form
\[
(\pi\rtimes u)(f)=\int_G \pi(f(x))u_x\,dx
\]
[2509.24106].

Exterior equivalence is the cohomological relation underlying untwisting. Two twisted actions $(\alpha,\sigma)$ and $(\beta,\omega)$ are exterior equivalent via a strictly continuous $\theta:G\to \operatorname{Inv}_1(M(A))$ if
\[
\beta_x=\operatorname{Ad}(\theta_x)\circ \alpha_x,\qquad
\omega_{x,y}\theta_{xy}=\theta_x\alpha_x(\theta_y)\sigma_{x,y}.
\]
In this situation, the corresponding twisted crossed products are isometrically isomorphic, with dense-subalgebra map
\[
f^{(\theta)}(x)=f(x)\theta_x^{-1}
\]
[2509.24106]. In the reduced $C^*$-setting, exterior equivalence is implemented by an explicit unitary on $H\otimes \ell^2(\Gamma)$, and the reduced twisted crossed products are canonically isomorphic [2109.08606].

This exterior-equivalence perspective clarifies the mechanism of the Packer–Raeburn trick: the stabilized twisted system becomes exterior equivalent to one with trivial twist. The trick is therefore not merely an isomorphism of crossed products but an equivalence of dynamical data after stabilization.

## 3. Stabilization and untwisting in the $L^p$ setting

A major recent extension replaces Hilbert-space methods by $L^p$-operator algebra techniques. Let $p\in [1,\infty)$, let $G$ be second countable, and let $A$ be a nondegenerate $L^p$-operator algebra with a contractive approximate identity. The full twisted $L^p$-crossed product is defined by
\[
F^p(G,A,\alpha,\sigma):=F_{\mathcal R^p}(G,A,\alpha,\sigma),
\]
where $\mathcal R^p$ is the class of all $\sigma$-finite, contractive covariant representations on $L^p$-spaces; for trivial $\sigma$, this recovers Phillips’s $F^p(G,A,\alpha)$ [2509.24106]. The reduced version is defined from regular covariant representations:
\[
F_r^p(G,A,\alpha,\sigma):=F_{\mathcal R_r^p}(G,A,\alpha,\sigma),
\]
where, for a nondegenerate contractive $\pi_0:A\to B(L^p(\Omega,\mu))$ and $E=L^p(G)\otimes_p L^p(\mu)$,
\[
(\pi(a)\xi)(x)=\pi_0(\alpha_x^{-1}(a))(\xi(x)),\qquad
(u_y\xi)(x)=\widehat{\pi_0}(\alpha_x^{-1}(\sigma_{y,y^{-1}x}))(\xi(y^{-1}x)).
\]
The paper defines both the full and reduced $L^p$-twisted crossed products, but the stabilization and untwisting theorem is stated for the full crossed products; an explicit reduced analogue is not claimed [2509.24106].

For $p\in (1,\infty)$, set $K:=\mathcal K(L^p(G))$. The stabilized twisted action on $K\otimes_p A$ is
\[
(\operatorname{id}_K\otimes \alpha)_x:=\operatorname{id}_K\otimes \alpha_x,\qquad
(1_K\otimes \sigma)_{x,y}:=1_K\otimes \sigma_{x,y}.
\]
One first has the stabilization isomorphism
\[
F^p\big(G,K\otimes_p A,\operatorname{id}_K\otimes \alpha,1_K\otimes \sigma\big)
\cong
K\otimes_p F^p(G,A,\alpha,\sigma).
\]
Then one defines
\[
\sigma_x^{-1}(y):=\sigma_{x,y}^{-1},
\]
lets $m(\sigma_x^{-1})$ act by coordinatewise multiplication on $L^p(G;L^p(\mu))$,
\[
\big(m(\sigma_x^{-1})\xi\big)(y)=\sigma_x^{-1}(y)\,\xi(y),
\]
writes $\Sigma_x:=m(\sigma_x^{-1})\in \operatorname{Inv}_1(M(K\otimes_p A))$, and uses the $p$-left regular representation $\lambda^p:G\to \operatorname{Inv}_1(M(K))$ to define
\[
\theta_x:=(\lambda_x^p\otimes \operatorname{id}_A)\cdot \Sigma_x.
\]
The resulting untwisted action is
\[
\beta_x:=\operatorname{Ad}(\theta_x)\circ (\operatorname{id}_K\otimes \alpha)_x.
\]
The main untwisting statement is
\[
K\otimes_p F^p(G,A,\alpha,\sigma)\cong F^p\big(G,K\otimes_p A,\beta\big),
\]
so the stabilized twisted $L^p$-crossed product is isometrically isomorphic to an untwisted crossed product [2509.24106].

This generalization preserves the structural role of stabilization while replacing $C^*$-specific ingredients by Banach-space analogues: invertible isometries replace unitaries, and multiplier theory is interpreted in the Banach sense.

## 4. Mechanism of the trick

The mechanism is explicitly described in the recent $L^p$ treatment. The cocycle is transferred into a projective representation on $L^p(G)$ and then absorbed into multiplier operators on $K\otimes_p A$ [2509.24106]. The left regular representation $\lambda_x^p$ acts by shifting the $G$-variable, while the cocycle contributes the multiplier-valued function $y\mapsto \sigma_{x,y}^{-1}$. Their product
\[
\theta_x=(\lambda_x^p\otimes \operatorname{id}_A)\Sigma_x
\]
is an invertible isometry in $M(K\otimes_p A)$, strictly continuous in $x$, and the cocycle identities imply that it implements exterior equivalence
\[
(\operatorname{id}_K\otimes \alpha,\,1_K\otimes \sigma)\overset{\theta}{\sim} (\beta,\mathbf 1),
\]
where the twist on the right-hand side is trivial [2509.24106].

At the dense-subalgebra level, the implementing transform is
\[
\Theta(F)(x):=F(x)\theta_x^{-1},\qquad F\in C_c(G,K\otimes_p A),
\]
which intertwines twisted convolution for $(\operatorname{id}_K\otimes \alpha,1_K\otimes \sigma)$ with untwisted convolution for $(\beta,1)$ [2509.24106]. The norm preservation $\|F^{(\theta)}\|=\|F\|$ allows the universal property of the full crossed product to promote this algebraic identification to an isometric isomorphism.

In the reduced $C^*$-algebra setting, Suzuki uses the same conceptual mechanism as a reduction principle rather than reproving the construction. The stabilized twisted action is exterior equivalent to a genuine action, and this yields an isomorphism of reduced crossed products commuting with subgroup conditional expectations [2109.08606]. That commuting property is central in applications to tracial weights.

A plausible implication is that the Packer–Raeburn trick is best understood as a functorial passage from cocycle-twisted covariance to ordinary covariance on a stabilised coefficient algebra, with the compact-operator factor carrying the projective defect of the original representation.

## 5. Extensions to Fell bundles and groupoids

For Hausdorff groupoids, the stabilization idea extends beyond twisted group actions to Fell bundles. "A Stabilization Theorem for Fell Bundles over groupoids" constructs, from a saturated upper-semicontinuous Fell bundle $p:B\to G$ over a second-countable Hausdorff locally compact groupoid $G$, a Hilbert module bundle $V$ over the unit space and its compact-operator bundle $K(V)$, together with an action $\alpha$ of $G$ on $K(V)$, such that the semi-direct Fell bundle $K(V)\rtimes_\alpha G$ is equivalent to $B$ [1512.06046].

For each unit $x$, the fibre $V(x)$ is the completion of $\Gamma_c(G_x;B)$ with inner product
\[
\langle \xi,\eta\rangle_{A(x)}=\int_{G_x}\xi(\gamma)^*\eta(\gamma)\,d\lambda_x(\gamma),
\]
and the resulting $K(V)$ is a $C_0(G^{(0)})$-algebra Morita equivalent to the unit-fibre algebra $A=\Gamma_0(G^{(0)};B)$ [1512.06046]. Using the bimodule structure of the Fell-bundle fibres, one defines isometric Hilbert-module isomorphisms
\[
\beta_g:V(r(g))\otimes_{A(r(g))}B(g)\to V(s(g)),
\]
from which *-isomorphisms
\[
\alpha_g:K(V(s(g)))\to K(V(r(g)))
\]
are obtained [1512.06046]. The main theorem then shows that $E=\bigsqcup_{g\in G} V(r(g))\otimes_{A(r(g))}B(g)$ implements a $(K(V)\rtimes_\alpha G)$–$B$ equivalence of Fell bundles.

Consequently,
\[
C^*(G;B)\sim_M C^*(G;K(V)\rtimes_\alpha G),\qquad
C_r^*(G;B)\sim_M C_r^*(G;K(V)\rtimes_\alpha G),
\]
and hence, by Brown–Green–Rieffel, the corresponding $C^*$-algebras are stably isomorphic [1512.06046]. The compact bundle $K(V)$ therefore plays the same role for groupoid Fell bundles that $\mathcal K(L^2(G))$ plays for twisted group actions: it absorbs the twisting or bimodule data so that the residual dynamics is encoded by an ordinary action.

This groupoid stabilization theorem was then used to transport Renault’s ideal and simplicity theory for groupoid crossed products to Fell-bundle algebras. Under the hypotheses stated in that paper, the lattice of ideals of $C^*(G;B)$ is identified with the lattice of open $G$-invariant subsets of $\operatorname{Prim}A$, and simplicity is characterized by minimality of the induced action on $\operatorname{Prim}A$ [1512.06046].

## 6. Limits of the method and non-Hausdorff obstruction

The stabilization paradigm does not extend without qualification to non-Hausdorff étale groupoids. "Inverse semigroup actions on groupoids" develops actions by partial equivalences, saturated Fell bundles, and actions on $C^*$-algebras by Hilbert bimodules, and concludes that the Packer–Raeburn Stabilisation Trick does not generalise to non-Hausdorff groupoids [1410.2051].

The main obstruction theorem states that if $G$ is an étale groupoid with Hausdorff unit space, locally Hausdorff and locally quasi-compact, but with non-Hausdorff arrow space, and if $A$ is a $C^*$-algebra with $\operatorname{Prim}(A)\cong G^{(1)}$, then there is no continuous (twisted) action of $G$ by automorphisms on $A$ that induces the left translation action of $G$ on $\operatorname{Prim}(A)=G^{(1)}$ [1410.2051]. The explanation given is that continuity of a genuine action by automorphisms forces continuity properties on transported sections over bisections, while the non-Hausdorff arrow space produces inseparable points that force discontinuities.

An explicit counterexample is constructed from a group bundle over $[0,1]$ with trivial isotropy except at $0$, where the isotropy is $\mathbb Z/2$, yielding a non-Hausdorff arrow space with inseparable points $0^+$ and $0^-$ [1410.2051]. The model algebra is
\[
A=\{f\in C([0,1],M_2): f(0)\text{ is diagonal}\},
\qquad \operatorname{Prim}(A)\cong G^{(1)},
\]
and one obtains a twisted inverse semigroup action by partial automorphisms whose section algebra is $B=C([0,1],M_2)$ [1410.2051]. However, there is no way to convert this into a genuine untwisted groupoid action by automorphisms, even after stabilization.

By contrast, the same paper notes that for Hausdorff locally compact groupoids, Fell bundles are equivalent, after stabilization, to ordinary actions by automorphisms [1410.2051]. The failure in the non-Hausdorff case therefore marks a genuine boundary of the Packer–Raeburn philosophy. A plausible implication is that in non-Hausdorff settings, Fell bundles and inverse semigroup actions are not technical substitutes for automorphic actions but the correct native framework.

## 7. Uses and conceptual significance

The Packer–Raeburn trick functions as a reduction principle. In Suzuki’s work on simplicity and tracial weights for non-unital reduced crossed products, the theorem is used to reduce twisted reduced crossed products to untwisted ones after stabilization [2109.08606]. For a twisted action $(\alpha,u):\Gamma\curvearrowright A$, one stabilizes by $K(\ell^2\Gamma)$, passes to the untwisted action $\beta$ produced by the Packer–Raeburn theorem, applies the untwisted arguments there, and transfers the conclusions back using exterior equivalence and the fact that the stabilized isomorphism commutes with subgroup conditional expectations [2109.08606].

This reduction underlies the paper’s twisted, non-unital versions of simplicity and tracial-weight results. In particular, the reduced twisted crossed product of a twisted action of a discrete $C^*$-simple group is simple if and only if $A$ has no proper $\Gamma$-invariant ideal, under the stated hypothesis that $A$ has a completely full element [2109.08606]. For tracial weights, the sets $I(\alpha,T)$ are invariant under exterior equivalence, and non-degeneracy passes to the stabilized untwisted system; this allows the classification theorem for proper tracial weights to be deduced from the untwisted case [2109.08606].

In the Banach and $L^p$ context, the trick has a structural rather than merely technical role. It shows that twisted $L^p$-crossed products are stably classified by untwisted data on $K\otimes_p A$, and that exterior equivalence yields explicit isometric isomorphisms through the maps $f\mapsto f^{(\theta)}$ [2509.24106]. This suggests a stable Morita-type perspective analogous to the $C^*$-case, though the paper phrases the conclusion in terms of stable isometric isomorphism rather than involutive Morita theory.

Across these settings, the common content of the Packer–Raeburn trick is the replacement of twisted covariance by untwisted covariance on a stabilized coefficient algebra. Where the required compact-operator stabilization and continuity theory are available, the trick transforms a cocycle problem into an ordinary dynamical system. Where they are not, as in non-Hausdorff étale groupoids, the failure of the trick identifies a substantive geometric obstruction rather than a gap in technique [1410.2051].

Source: https://www.emergentmind.com/topics/packer-raeburn-trick