---
title: Ozawa's Relative Solidity Theorem
url: https://www.emergentmind.com/topics/ozawa-s-relative-solidity-theorem
type: topic
---

# Ozawa's Relative Solidity Theorem

Ozawa’s relative solidity theorem refers to a rigidity paradigm in the theory of von Neumann algebras in which geometric or \(C^*\)-algebraic properties of a group force a dichotomy for subalgebras of an associated crossed product or group factor: either the subalgebra Popa-intertwines into a specified “base” algebra, or its relative commutant or normalizer is amenable. In the literature around negatively curved groups, the baseline statement is Ozawa’s solidity theorem for group factors of i.c.c. hyperbolic groups, while the relative form appears as a crossed-product dichotomy and as a weakly compact embedding theorem that became the bridge to strong solidity [1103.4299].

## 1. From solidity to relative solidity

Ozawa’s original solidity theorem, as recalled in the structural theory of \(\mathrm{II}_1\) factors of negatively curved groups, states:

> If \(T\) is an i.c.c. Gromov hyperbolic group, then \(L T\) is solid, i.e.
> \[
> A' \cap LT \ \text{is amenable for every diffuse von Neumann subalgebra } A \subset LT.
> \]

Here “solid” means exactly that every diffuse von Neumann subalgebra has amenable relative commutant. In this form, the theorem concerns a group factor \(LT\) and gives no ambient subalgebra relative to which the diffuse algebra must be located [1103.4299].

The relative-solidity perspective replaces this absolute alternative by a structural dichotomy inside a crossed product. In the formulation used in the negatively curved setting, if
\[
M=L^\infty(X)\rtimes T,
\]
then a diffuse subalgebra \(A\subset M\) either has amenable relative commutant or intertwines into a prescribed subalgebra coming from the base algebra or from a family of subgroups. This is the sense in which “relative solidity” extends ordinary solidity: the obstruction to amenability is not arbitrary largeness, but rather concentration inside a controlled subalgebra [1103.4299].

This framework is explicitly tied to Ozawa’s use of bi-exactness and \(C^*\)-algebraic methods, but it is recast in deformation/rigidity language by combining ideas of Peterson, Ozawa, and Popa. The same viewpoint underlies later formulations for relatively hyperbolic groups, measure equivalence, generalized \(q\)-Gaussian algebras, and type III crossed products [1103.4299], [2509.19481], [2503.24167], [1509.07069], [2508.17592].

## 2. The relative theorem in crossed products

The central relative commutant dichotomy in the negatively curved-group setting is Theorem 3.2. Let \(T\) be exact and assume it admits an array into a weakly-\(\ell^2\) representation that is proper with respect to a family \(\mathcal F\) of subgroups. Let
\[
T\curvearrowright X
\]
be free, ergodic, p.m.p., and let
\[
M=L^\infty(X)\rtimes T.
\]
Then for any diffuse von Neumann subalgebra \(A\subset M\), either:

1. \(A'\cap M\) is amenable, or  
2. \(A\prec_M L^\infty(X)\rtimes \Sigma\) for some \(\Sigma\in\mathcal F\) [1103.4299].

For the case \(\mathcal F=\{\{e\}\}\), this becomes solidity. In that sense, the theorem is a genuine relative version of Ozawa’s solidity theorem: it identifies the only possible nonamenable obstruction to solidity as intertwining into a controlled crossed-product piece [1103.4299].

A second relative formulation, Theorem 4.1, is the weakly compact embedding theorem. If \(T\) is an exact group admitting a proper quasi-cocycle into a weakly-\(\ell^2\) representation, \(T\curvearrowright X\) is measure-preserving, and
\[
M=L^\infty(X)\rtimes T,
\]
then for every weakly compact embedding \(P\subset M\), one of the following holds:

1. \(P \prec_M L^\infty(X)\), or  
2. \(N_M(P)''\) is amenable [1103.4299].

This theorem is the precise relative-solidity statement used to derive strong solidity when \(X\) is a point. It isolates the normalizer, not merely the relative commutant, and it does so under the Ozawa–Popa notion of weak compactness.

The paper also formulates the underlying Popa intertwining notation explicitly:
\[
A \prec_M B,
\]
meaning that a corner of \(A\) embeds into \(B\) inside \(M\). Relative solidity is therefore not only a statement about amenability; it is a localization statement in the sense of Popa’s intertwining-by-bimodules theory [1103.4299].

## 3. Group hypotheses and the passage to strong solidity

The group-theoretic hypotheses entering the theorem are precise. The main results are stated for groups satisfying some combination of:

- **i.c.c.**,
- **exactness**,
- **weak amenability**,
- **admitting a proper quasi-\(1\)-cocycle into a weakly-\(\ell^2\) representation**, and
- for exact groups, equivalently being in the class \(QH_{\mathrm{reg}}\) [1103.4299].

Theorem A gives a unification of Ozawa’s and Peterson’s solidity results:

> Let \(T\) be an i.c.c. countable discrete group which is exact and admits a proper quasi-\(1\)-cocycle \(q:T\to H\) into a weakly-\(\ell^2\) representation. Then \(LT\) is solid.

The paper notes explicitly that for exact groups, \(QH_{\mathrm{reg}}\) is equivalent to bi-exactness, so Theorem A recovers Ozawa’s solidity theorem for hyperbolic groups [1103.4299].

The decisive strengthening is Theorem B:

> Let \(T\) be an i.c.c. countable discrete group which is weakly amenable. If \(T\) admits a proper quasi-\(1\)-cocycle into a weakly-\(\ell^2\) representation, then \(LT\) is strongly solid.

The distinction between solidity and strong solidity is exact. A \(\mathrm{II}_1\) factor \(M\) is solid if
\[
A'\cap M \ \text{is amenable for every diffuse von Neumann subalgebra } A\subset M,
\]
whereas \(M\) is strongly solid if for every diffuse amenable von Neumann subalgebra \(A\subset M\),
\[
N_M(A)'' \ \text{is amenable},
\qquad
N_M(A)=\{u\in U(M)\mid uAu^*=A\}.
\]
Theorem B upgrades control of the relative commutant to control of the whole normalizer [1103.4299].

The paper states consequences for i.c.c. hyperbolic groups, for lattices in rank-one simple Lie groups, and in particular for i.c.c. lattices in \(\mathrm{Sp}(n,1)\), \(n\ge 2\), and \(F_4(-20)\). It also records that hyperbolic groups satisfy the hypotheses via Mineyev–Monod–Shalom and Ozawa, and that \(\mathbb Z^2\rtimes SL(2,\mathbb Z)\) lies in \(QH_{\mathrm{reg}}\) [1103.4299].

## 4. Deformation/rigidity mechanism

The proof strategy is described as a hybrid of Peterson, Ozawa, and Popa. Peterson contributes the quasi-cocycle and \(L^2\)-rigidity perspective; Ozawa contributes exactness, local reflexivity, and passage from \(C_r^*(T)\) to \(LT\); Popa contributes intertwining, deformation/rigidity, and spectral gap [1103.4299].

The deformation is built from exponentiating a quasi-cocycle. For a quasi-cocycle \(q:T\to H\),
\[
\|q(gh)-q(g)-gq(h)\|\le D(q).
\]
The associated positive-definite kernel is
\[
k_t(g,h)=\exp\big(-t^2\|q(g)-q(h)\|^2\big),
\]
which gives a u.c.p. Schur multiplier \(m_t\). The paper also constructs
\[
v_t(g)(x)=\exp(it\, q(g)(x)),
\]
leading to a one-parameter family of automorphisms \(a_t\) on an extended Roe algebra \(C^*(T\ltimes Z)\) [1103.4299].

The essential analytic features are that these deformations converge pointwise to the identity on \(C_r^*(T)\), behave compactly enough on Fourier tails, and allow one to push Haagerup’s amenability criterion through. The criterion used is:

> A von Neumann subalgebra \(N\subset M\) is amenable iff for every nonzero central projection \(p\in Z(N)\) and finite \(F\subset U(Np)\),
> \[
> \left\|\sum_{u\in F} u\otimes u^*\right\| = |F|.
> \]

This is the step by which deformation estimates yield amenability of relative commutants or normalizers [1103.4299].

Weak compactness is the technical bridge from relative commutant control to normalizer control. Following Ozawa–Popa, \(P\subset M\) is weakly compact if the conjugation action of \(N_M(P)\) on \(P\) is weakly compact, witnessed by a net of unit vectors \(\eta_n\in L^2(M)\otimes L^2(M)\) satisfying approximate centrality under \(U(P)\), approximate invariance under \(N_M(P)\), and the trace marginals
\[
((x\otimes 1)\eta_n,\eta_n)=\tau(x),
\qquad
((1\otimes x)\eta_n,\eta_n)=\tau(x).
\]
This is the precise condition entering Theorem 4.1 [1103.4299].

## 5. Structural consequences and negatively curved groups

The conceptual advance of the negatively curved-group paper is stated explicitly: it turns the bi-exactness/solidity paradigm into a deformation/rigidity framework driven by quasi-cocycles and arrays. The role of “negative curvature” is interpreted cohomologically via proper quasi-cocycles or arrays, thereby unifying hyperbolic groups, rank-one lattices, and related examples under one von Neumann algebraic mechanism [1103.4299].

The significance of strong solidity is also stated concretely. Strong solidity gives finer structural control because it rules out large normalizers of amenable subalgebras, yields uniqueness of Cartan subalgebras in many crossed products, and feeds into orbit equivalence and \(W^*\)-superrigidity applications. Using the methods together with a cocycle superrigidity result of Ioana, the paper shows that profinite actions of lattices in \(\mathrm{Sp}(n,1)\), \(n>1\), are virtually \(W^*\)-superrigid [1103.4299].

A plausible implication is that the relative-solidity theorem is best understood not as an isolated commutant estimate, but as a localization principle. In the crossed-product setting, the key alternative is
\[
A'\cap M \text{ amenable }
\quad\text{or}\quad
A\prec_M L^\infty(X)\rtimes \Sigma,
\]
and in the weakly compact setting it becomes
\[
P \prec_M L^\infty(X)
\quad\text{or}\quad
N_M(P)'' \text{ amenable}.
\]
This suggests that the theorem identifies the only ways in which nonamenable algebraic structure can persist inside factors associated with negatively curved groups [1103.4299].

## 6. Later extensions and variants

Subsequent work has extended the relative-solidity paradigm in several directions.

For relatively hyperbolic groups, a 2025 paper proves that whenever \(G\) is hyperbolic relative to a finite family of exact, residually finite subgroups \(\{H_1,\dots,H_n\}\), the group von Neumann algebra \(\mathcal L(G)\) is solid relative to \(\{\mathcal L(H_1),\dots,\mathcal L(H_n)\}\) in the sense that for every nonzero projection \(p\in \mathcal L(G)\) and every \(\mathcal A\subset p\mathcal L(G)p\) whose relative commutant has no amenable direct summand, there exists \(i\) such that
\[
\mathcal A\prec_{\mathcal L(G)} \mathcal L(H_i).
\]
The paper strengthens this to a structural statement for crossed products and a control theorem for the one-sided quasi-normalizer tower [2509.19481].

In the measure-equivalence setting, Ding–Drimbe prove an analogue of relative solidity for orbit-equivalent actions of \(\Gamma\times\Sigma\), where \(\Gamma\) is nonamenable biexact and \(\Sigma\) is arbitrary infinite. If
\[
M=L^\infty(X,\mu)\rtimes(\Gamma\times\Sigma)=L^\infty(Y,\nu)\rtimes\Lambda,
\]
then for any subgroup \(\Delta<\Lambda\), either
\[
L^\infty(Y)\rtimes\Delta\prec_M L^\infty(X,\mu)\rtimes\Sigma
\]
or
\[
L^\infty(Y)\rtimes C_{\Lambda}(\Delta)
\text{ is amenable relative to }
L^\infty(X,\mu)\rtimes\Sigma
\text{ inside }M.
\]
This is described there as the measure-equivalence analogue of Ozawa–Popa relative solidity [2503.24167].

A different analogue appears for generalized \(q\)-Gaussian von Neumann algebras with coefficients. Under a subexponential growth estimate
\[
\dim_B(D_s(S))\le C d^s,
\]
the main theorem gives the dichotomy
\[
\mathcal A\prec_M B
\qquad\text{or}\qquad
\mathcal N_M(\mathcal A)''\ \text{is amenable relative to }B\text{ inside }M
\]
for diffuse \(\mathcal A\subset M=\Gamma_q(B,S\otimes H)\) that are amenable relative to \(B\). This is presented as a \(q\)-Gaussian analogue of Ozawa–Popa relative strong solidity [1509.07069].

Most recently, the theorem has been extended to the type III setting. For an action \(\alpha:\Gamma\curvearrowright B\) of a bi-exact discrete group on an amenable \(\sigma\)-finite von Neumann algebra, with \(M=B\rtimes_\alpha\Gamma\), the type III extension states that \(M\) is solid relative to \(B\): for every projection \(p\in M\) and every von Neumann subalgebra \(A\subset pMp\) with expectation such that \(A\not\preceq_M B\), the relative commutant \(A'\cap pMp\) is amenable. The new ingredient is a weak relative Dixmier theorem for inclusions equipped with a faithful normal semifinite operator-valued weight \(E_A:M\to A\) [2508.17592].

Taken together, these developments show that “Ozawa’s relative solidity theorem” now denotes a family of rigidity statements with a stable formal pattern: a geometric or exactness condition on the ambient object yields an intertwining-versus-amenability dichotomy for subalgebras, relative commutants, or normalizers. In the negatively curved \(\mathrm{II}_1\)-factor setting, the formulation in terms of proper quasi-cocycles, weakly-\(\ell^2\) representations, and weakly compact embeddings remains one of the foundational forms of that paradigm [1103.4299].

Source: https://www.emergentmind.com/topics/ozawa-s-relative-solidity-theorem