---
title: G-Quasi-Invariant States Overview
url: https://www.emergentmind.com/topics/g-quasi-invariant-states
type: topic
---

# G-Quasi-Invariant States Overview

Searching arXiv for recent and foundational papers on quasi-invariant states to ground the article.
\(G\)-quasi-invariant states are noncommutative analogues of quasi-invariant measures: instead of requiring a state to be fixed by a group action, one requires each transformed state to remain comparable to the original state through a Radon–Nikodym-type defect. Across the literature, the term is used in several closely related but non-identical senses. In operator algebras, the basic pattern is that a group \(G\) acts by \(*\)-automorphisms on an algebra \(\mathcal A\), and a state \(\varphi\) is called quasi-invariant when \(\varphi\circ g\) is related to \(\varphi\) by an internal derivative or by GNS-equivalence; in measure theory, quasi-invariance is expressed by mutual absolute continuity or an explicit density cocycle; and in some specialized contexts, “\(G\)” refers not to a symmetry group but to “gauge” or “Gaussian,” producing terminological variants that are adjacent rather than identical to the general group-action notion [2209.12717].

## 1. Operator-algebraic definitions and cocycle structure

In the modern operator-algebraic formulation, a von Neumann algebra \(A\) is equipped with a group \(G\) of normal \(*\)-automorphisms, and a faithful normal state \(\varphi\) is called \(G\)-quasi invariant if for every \(g\in G\) there exists \(x_g\in A\) such that
\[
\varphi(g(a))=\varphi(x_g a), \qquad a\in A.
\]
A state is \(G\)-strongly quasi invariant if, in addition, each \(x_g\) is self-adjoint; a generalized version allows the \(x_g\) to be self-adjoint operators affiliated with \(A\) rather than bounded elements of \(A\) [2311.01481]. A closely parallel development on \(*\)-algebras and von Neumann algebras defines \((G,x)\)-quasi-invariant and \((G,x)\)-strongly quasi-invariant states in the same Radon–Nikodym form, with the strong condition again meaning hermitian cocycles [2209.12717].

These derivatives are not arbitrary. They form a normalized multiplicative left \(G\)-\(1\)-cocycle:
\[
x_e=1,\qquad x_{g_2g_1}=x_{g_1}\,g_1(x_{g_2}),
\]
and are invertible with
\[
x_g^{-1}=g^{-1}(x_{g^{-1}}).
\]
In the strongly quasi-invariant case, the cocycles are strictly positive, commute pairwise, and generate a commutative \(C^*\)-algebra lying in the centralizer of the state. This places strong quasi-invariance between arbitrary state covariance and full invariance: the state is not fixed, but its orbit under \(G\) is controlled by an internal abelian positive cocycle [2209.12717].

A recurrent distinction in the literature is between strict invariance and quasi-invariance. Strict invariance is the special case \(x_g=1\) for all \(g\). Several results characterize when strong quasi-invariance collapses to invariance. For example, if the cocycles lie in the fixed-point algebra \(F(G)=\{a\in A:g(a)=a,\ \forall g\in G\}\), then they must be trivial and the state is actually \(G\)-invariant [2311.01481]. This clarifies a common misconception: quasi-invariance is not merely “approximate invariance,” but a precise cocycle-covariance condition.

## 2. Strong quasi-invariance, modular theory, and conditional expectation

A major theme of the theory is the interaction with modular automorphism groups. For strongly quasi-invariant states, the modular group \((\sigma_t^\varphi)_{t\in\mathbb R}\) is tightly constrained by the cocycle. In a factor, strong quasi-invariance is equivalent to the existence of strictly positive \(x_g\in A\) satisfying
\[
g^{-1}\circ \sigma_t^\varphi(a)=x_g^{it}\bigl(\sigma_t^\varphi\circ g^{-1}(a)\bigr)x_g^{-it},
\]
which expresses the mismatch between \(G\) and modular dynamics by inner conjugation with the cocycle [2311.01481]. More globally, the modular group commutes with \(G\) if and only if the state is strongly quasi-invariant in the generalized sense and the cocycles are affiliated with the center \(Z=A\cap A'\) [2311.01481]. This gives a precise criterion for when quasi-invariance is compatible with modular symmetry rather than obstructing it.

For compact groups, strong quasi-invariance admits a particularly rigid description. If \(G\) is compact and \(g\mapsto x_g\) is continuous, the cocycles are uniformly bounded, and one can define
\[
K=\int_G x_g\,dg.
\]
Then \(K\) is positive and invertible, and the state can be written as a density twist of a \(G\)-invariant state:
\[
\varphi(a)=\varphi_G(K^{-1}a),
\qquad
x_g=K\,g^{-1}(K^{-1}).
\]
Thus every strongly quasi-invariant state for a compact group is cohomologically trivial at the level of bounded cocycles: it is obtained from an invariant state by conjugation with a positive invertible density [2209.12717]. A later analytic treatment with uniformly bounded cocycles reaches a closely related conclusion: quasi-invariant states are exactly bounded perturbations of invariant states, with
\[
x_g=d\,g^{-1}(d^{-1})
\]
for a bounded invertible \(d\in A\); in the strong case, \(d\) may be chosen positive and commuting suitably with its \(G\)-orbit [2412.03328].

This structure yields conditional expectations onto fixed-point algebras. Under strong quasi-invariance with uniformly bounded cocycle, one obtains a faithful normal \(G\)-invariant state and hence a unique normal faithful Umegaki conditional expectation onto the fixed-point algebra \(A^G\) [2412.03328]. For compact groups, the conditional expectation has the familiar averaging form
\[
E_G(a)=\int_G g(a)\,dg,
\]
and intertwines naturally with the modular group under additional hypotheses [2311.01481]. These results extend several classical theorems for invariant states, including Størmer-type trace decompositions on semifinite von Neumann algebras, to the strongly quasi-invariant setting [2311.01481].

## 3. GNS implementation and compact-group representation theory

Strong quasi-invariance is sufficiently rigid to produce canonical unitary implementations in the GNS representation. If \((H_\varphi,\pi,\Phi)\) is the GNS triple of a strongly quasi-invariant state, then there exists a unitary representation \(U\) of \(G\) such that
\[
U_g\,\pi(a)\Phi=\pi\bigl(g(a)x_{g^{-1}}^{1/2}\bigr)\Phi,
\qquad
U_g\,\pi(a)\,U_g^*=\pi(g^{-1}(a)).
\]
The cocycle enters through the square root \(x_{g^{-1}}^{1/2}\), exactly as a Radon–Nikodym derivative enters classical \(L^2\)-implementations of quasi-invariant measure-preserving transformations [2209.12717]. In the GNS space, the modular operators of the transformed states \(\varphi\circ g\) are related explicitly to those of \(\varphi\) by formulas involving \(\pi(\sqrt{x_g})\) and modular conjugation, giving a noncommutative modular Radon–Nikodym theory at the level of Tomita operators [2311.01481].

A recent \(C^*\)-algebraic approach for compact groups adopts a different definition: a state \(\omega\) is \(G\)-quasi-invariant if for every \(g\in G\),
\[
\pi_{\omega\circ\alpha_g}\cong \pi_\omega,
\]
that is, all transformed states have unitarily equivalent GNS representations [2507.20931]. This notion is weaker than invariance but still representation-theoretic. Under additional hypotheses—central support of \(\omega\), separability of \(\mathcal A\), and commutation of the lifted action with the modular group—the GNS representation of a \(G\)-quasi-invariant state is unitarily equivalent to that of the averaged invariant state
\[
\omega\circ E_G,\qquad E_G(a)=\int_G \alpha_g(a)\,dg.
\]
The proof uses averaging on the von Neumann algebra, modular covariance, and the Pedersen–Takesaki Radon–Nikodym theorem [2507.20931].

These two strands—internal cocycle derivatives and representation-equivalence quasi-invariance—are not identical definitions, but they are closely allied. A plausible implication is that compact-group quasi-invariance is often best understood not as a classification of new representation types, but as a way of moving within the representation class of an invariant state, provided modular obstructions are absent [2507.20931]. The literature also records genuine obstructions: without central support, quasi-invariance need not imply equivalence to an invariant-state representation, as shown by inner actions on matrix algebras [2507.20931].

## 4. Measure-theoretic analogues and quotient-space formulations

The measure-theoretic analogue is explicit in the study of double coset spaces. Let \(G\) be a locally compact Hausdorff group, \(H,K\) closed subgroups, \(N=N_G(K)\) the normalizer, and \(K\backslash G/H\) the double coset space. A positive Radon measure \(\mu\) on \(K\backslash G/H\) is \(N\)-strongly quasi-invariant if there exists a continuous positive function
\[
\lambda:N\times K\backslash G/H\to(0,\infty)
\]
such that
\[
d\mu_n(\ddot y)=\lambda(n,\ddot y)\,d\mu(\ddot y).
\]
The cocycle law
\[
\lambda(n_1n_2,p)=\lambda(n_1,n_2\cdot p)\lambda(n_2,p)
\]
follows, and the theory is organized by \(\rho\)-functions satisfying
\[
\rho(kxh)=\frac{\Delta_K(k)\Delta_H(h)}{\Delta_G(h)}\,\rho(x).
\]
Under \(K\) an IN-group and suitable openness and inclusion assumptions, every \(N\)-strongly quasi-invariant measure arises from such a \(\rho\)-function, and when \(K\triangleleft G\) the normalizer equals \(G\), giving genuine \(G\)-strongly quasi-invariant measures [1807.00132].

A different one-dimensional model concerns orientation-preserving actions of groups on \(\mathbb R\). There quasi-invariance is formulated by exact scaling:
\[
g_*\mu=\lambda_g\mu,\qquad \lambda_g>0.
\]
This is stronger than measure-class equivalence and yields a multiplicative character \(g\mapsto \lambda_g\). The associated action is semiconjugate to an affine action
\[
A_g(x)=\lambda_g x+\mu([0,g(0))),
\]
and for a broad extension-closed class \(\mathcal S\) containing all solvable groups and all groups of locally subexponential growth, the existence of a fixed-point-free element implies existence of a quasi-preserved Radon measure [1607.05310]. Integration against such measures produces quasi-invariant positive functionals on function algebras, furnishing a classical model of quasi-invariant states.

These measure-theoretic results illuminate the operator-algebraic notions rather than coincide with them. In both settings, quasi-invariance is encoded by a cocycle or density ratio; in both, strict invariance is the trivial-cocycle case; and in both, compactness or amenability often allows averaging back to invariant objects [1807.00132]. A common misconception is that operator-algebraic quasi-invariance is merely a direct translation of classical quasi-invariant measures. The analogy is structurally accurate, but the noncommutative setting introduces centralizers, modular groups, affiliated operators, and GNS-equivalence, none of which have exact classical analogues.

## 5. Rigidity of invariant states and the boundary of quasi-invariant theory

Several papers relevant to \(G\)-quasi-invariant states study only the strictly invariant endpoint. Their importance lies in showing how rigid strict invariance can become under large symmetry groups, thereby clarifying what quasi-invariance must relax.

For algebraic noncommutative tori \(\mathcal A^h_{g,\sigma}\), with irrational \(\hbar=h/2\pi\) and \(\sigma\) nondegenerate, the integral symplectic group \(\mathrm{Sp}(\mathbb Z^{2g},\sigma)\) acts by \(*\)-automorphisms
\[
\Phi_\Theta(W_m)=W_{\Theta m}.
\]
The only invariant state is the canonical trace
\[
\tau(W_m)=
\begin{cases}
1,&m=0,\\
0,&m\neq 0.
\end{cases}
\]
The proof uses orbit reduction and positivity of finite-dimensional Gram matrices; it does not introduce quasi-invariant states, but its method strongly suggests that large symplectic symmetry imposes severe positivity constraints on any state-like covariance condition [1802.02487]. A related generalization to twisted group \(*\)-algebras \(W_G\) of pre-symplectic abelian groups again shows that in irrational nondegenerate ergodic regimes, the unique invariant state is the tracial delta-at-identity state, whereas degenerate cases admit many invariant states [1906.09766].

In the setting of discrete quantum groups, a state on \(C_r(\widehat{\mathbb G})\) is \(\mathbb G\)-invariant if it is fixed under the adjoint coaction. The main theorem identifies \(\mathbb G\)-invariant states with tracial states on \(C_r(\widehat{\mathbb G})\), generalizing the unimodular case:
\[
\text{tracial states} \iff \mathbb G\text{-invariant states}.
\]
These states concentrate on the quotient by the Furstenberg boundary cokernel and are controlled by Kac-type quotient structure [2205.05176]. The paper does not define quasi-invariant states for quantum-group coactions, but it establishes the strict baseline against which any future quasi-invariant extension would be measured.

This rigidity literature corrects another frequent misunderstanding: the passage from invariance to quasi-invariance is not always a minor weakening. In many high-symmetry noncommutative systems, strict invariance leaves only a single tracial state [1802.02487]. A plausible implication is that quasi-invariant theories become interesting precisely where invariant-state theory is too rigid, not because invariance and quasi-invariance are nearly interchangeable.

## 6. Specialized usages: gauge, Gaussian, quasi-free, and anomalous relatives

The phrase “\(G\)-quasi-invariant states” is not uniform across mathematical physics. In fermionic and anyonic contexts, “\(G\)” often abbreviates “gauge,” not a symmetry group. Gauge-invariant quasi-free states on the algebra of anyon commutation relations are classified by a bounded positive operator \(T\) commuting with multiplication by functions of the first coordinate; if \(\eta=\Re q<0\), one additionally requires
\[
0\le T\le -\frac1\eta.
\]
The state is determined by its two-point function, and higher moments obey a \(Q\)-weighted Wick rule [1510.05829]. For translation-invariant lattice fermions, gauge-invariant quasi-free states with regular symbol \(\widehat C\in\mathcal W\) and \(0<\widehat C<1\) are uniquely characterized as the maximal-specific-entropy states at fixed two-point function and are weak Gibbs states for an explicit quadratic interaction [2603.14020]. These theories concern quasi-free structure and gauge symmetry, not group-action quasi-invariance in the Radon–Nikodym sense, though the terminological overlap is substantial.

A different specialized usage occurs for Gaussian continuous-variable states evolving under quadratic Hamiltonians. There the paper defines “Gaussian quasi-invariant states” as states whose covariance matrices evolve in an unusually constrained or slowly varying way under the dynamics, distinguishing them from fully invariant states and from states with invariant covariance matrices [2005.11497]. Here “\(G\)” means Gaussian. The notion is dynamical and parametric rather than cocycle-theoretic.

There are also nearby theories involving exact bulk invariance but anomalous boundary symmetry. In two-dimensional bosonic SPT systems with finite on-site symmetry group \(G\), the bulk state is exactly \(G\)-invariant, yet the boundary-localized symmetry pieces fail strict associativity and generate a \(3\)-cocycle in \(H^3(G,U(1))\) [2101.00801]. This is not a theory of quasi-invariant states, but it is directly relevant to how weakened or anomalous symmetry laws arise from exact invariant bulk states.

The resulting terminological landscape is therefore heterogeneous. In the strictest sense, \(G\)-quasi-invariant states are states on operator algebras or function spaces whose transforms are related by cocycles or GNS-equivalence [2209.12717]. In broader usage, the phrase may refer to gauge-invariant quasi-free states, Gaussian near-invariant states, or boundary-anomalous symmetry behavior [1510.05829]. Precision about the role of \(G\) is essential.

## 7. Scope, examples, and open directions

Concrete examples now span finite-dimensional matrix algebras, compact-group actions, tensor-product systems with local permutation symmetry, quantum Markov chains, double coset spaces, and dynamical systems on the line. For \(S_\infty\) acting by local permutations on an infinite tensor product \(\bigotimes_{n\in\mathbb N}\mathcal B(\mathbb C^d)\), product states with commuting one-site densities produce explicit strong quasi-invariance cocycles
\[
x_\sigma=\prod_{n\in\Lambda_\sigma} j_n(W_{\sigma(n)}/W_n),
\]
and finite-level averaging over \(S_N\) yields martingale conditional expectations converging strongly to the projection onto the full fixed-point algebra [2403.05590]. Related work shows that quantum Markov chains with commuting, invertible, hermitian conditional density amplitudes are strongly quasi-invariant under the natural action of \(\mathcal S_\infty\), again with explicit cocycle formulas [2209.12717].

Two controversies or unsettled boundaries recur across the literature. The first is definitional: some papers define quasi-invariance through internal Radon–Nikodym derivatives \(x_g\), while others use unitary equivalence of GNS representations [2507.20931]. These are related but not identical frameworks. The second concerns scope: many strong structure theorems require compactness, amenability, central support, bounded cocycles, or strong quasi-invariance rather than mere quasi-invariance [2412.03328]. Counterexamples show that without such hypotheses, conditional expectations, invariant traces, or representation-theoretic replacement by invariant states can fail.

A careful synthesis is therefore that \(G\)-quasi-invariant states form a family of covariance conditions interpolating between strict invariance and unconstrained state dynamics. Their defining feature is not small deviation from symmetry but control by a cocycle, density, or representation-equivalence class. In compact and bounded settings, this control often allows reduction to invariant-state theory; in rigid high-symmetry settings, strict invariance may be unique and tracial; and in measure-theoretic or specialized physical settings, quasi-invariance serves as the natural replacement for exact symmetry when a transformed state remains absolutely continuous, cohomologically equivalent, or dynamically near-fixed [2412.03328].

Source: https://www.emergentmind.com/topics/g-quasi-invariant-states