Compressible subalgebras in II$_1$ factors

Sorin Popa

REVIEW 3 major objections 4 minor 6 references

Compressible subalgebras in II$_1$ factors

T0 review · 3 major / 4 minor · reviewed 2026-08-04 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Compressible subalgebras always see the coarse part of a hyperfinite subalgebra

desk verdict Popa's compressibility is a genuinely new II1-factor notion, and the main non-quasi-regular theorem is plausible and important, but the quasi-regular conclusions rest on Lemma 2.7, whose proof is skipped—so the paper is a conditional major advance, not a finished one. read the letter →

arxiv 2510.17076 v2 pith:5ZTLQ6OK submitted 2025-10-20 math.OA

Sorin Popa This is my paper · ORCID

classification math.OA MSC 46L1046L54

keywords compressiblesubalgebraII1factorscoarsebimoduleAFDflatnessergodicinclusionfreeindependencebasicconstruction

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper introduces a new property for a subalgebra Q inside a II1 factor M: Q is compressible if averaging over unitaries of M drives arbitrary matrix amplifications of elements of Q to scalar matrices, uniformly in norm. It proves that if Q is compressible, then for every approximately finite-dimensional (AFD) subalgebra N of M, the Hilbert bimodule N L^2M_Q contains a nonzero coarse part—a subspace that behaves like the tensor product of the standard representations of N and Q. If N is quasi-regular, the whole bimodule is coarse. This rules out tight hyperfinite complements for compressible subalgebras and yields inclusions of factors that are ergodic but not AFD-ergodic, even after tensoring with the compact operators on a Hilbert space.

What carries the argument

The compressibility property: for every ε>0 there is a finite set of unitaries in M whose averaged conjugation sends every norm-one matrix amplification of elements of Q to a scalar matrix within ε in operator norm. This is used with a flatness theorem guaranteeing that for an AFD subalgebra N, the algebra generated by N and the opposite of Q acts as a spatial tensor product; compressibility then forces a normal conditional expectation from N ∨ Q^op onto N ∨ B^op and yields nonzero coarse submodules. The quasi-normalizer lemma shows the coarse part is invariant under quasi-normalizers, upgrading a nonzero coarse part to total coarseness when N is quasi-regular.

What would settle it

Find a tracial W*-inclusion Q ⊂ M with Q diffuse and compressible, and an AFD subalgebra N ⊂ M, such that N L^2M_Q has zero coarse part; equivalently, find in the hyperfinite II1 factor a diffuse compressible subalgebra in some embedding, or any diffuse quasi-regular compressible subalgebra—either would directly contradict Corollary 2.9.

Watch

Extended reading notes

Core claim

The central claim is that a compressible tracial inclusion Q ⊂ M is 'AFD-repellent'. For any AFD subalgebra N ⊂ M, the N–Q Hilbert bimodule L^2M contains a nonzero copy of the coarse N–Q bimodule L^2N ⊗ L^2Q; if N is quasi-regular in M, then the entire bimodule is coarse, i.e. a sub-bimodule of a direct sum of coarse bimodules. The proof combines compressibility—uniformly averaging matrix amplifications of Q to scalars—with a flatness theorem ensuring that over an AFD algebra the generated von Neumann algebra acts like a spatial tensor product. A density theorem for the strong operator topology is then used repeatedly to push the averaging to the level of von Neumann algebras and construct t

Load-bearing premise

The argument relies on flatness of the Hilbert bimodule N L^2M_Q for AFD N: only inside the spatial tensor product representation can compressibility be shown to produce a nonzero coarse part; if flatness fails for some N, the main theorem has no force.

Editorial extensions

If this is right

If Q ⊂ M is diffuse and compressible, it cannot be quasi-regular; in particular, the hyperfinite II1 factor has no diffuse compressible subalgebras.
A group with an infinite compressible subgroup is non-amenable, and an ergodic equivalence relation containing a type II1 compressible subequivalence relation is non-amenable.
No embedding of a free product Q ∗ Q0 into a II1 factor with Q0 ≠ C can have the hyperfinite II1 factor as a tight complement for Q.
Basic construction inclusions M ⊂ ⟨M, e_Q⟩ from ergodic compressible Q are ergodic but not AFD-ergodic, and this persists after tensoring with B(H0); such inclusions are neither MASA-ergodic nor R-ergodic.
In ultraproduct II1 factors, every separable subalgebra is compressible, so any AFD subalgebra has a nonzero coarse part relative to any separable subalgebra.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

Compressibility can be read as a uniform, matrix-stable relative Dixmier property; the theorem suggests that such uniform averaging is fundamentally incompatible with tightness, offering a negative criterion for tight decomposition problems.
Since compressibility holds automatically in free-product-like situations, the theorem gives a new obstruction to tight hyperfinite complements that may be explored through entropy or strong solidity type invariants.
A testable boundary: if one could exhibit a diffuse compressible subalgebra that is quasi-regular in some II1 factor, the main theorem would collapse; constructing such an example, or proving none exists, would sharpen the limit of the result.
The stabilization by B(H0) shows the non-AFD-ergodicity phenomenon is stable under tensoring with arbitrary type I factors, which may be relevant for transferring the obstruction to type III inclusions via continuous decomposition.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Dear colleague, the paper you asked about is a real advance, not a routine one. Popa introduces compressible W*-subalgebras and proves that if Q ⊂ M is compressible then for every AFD subalgebra N, the bimodule N L^2M_Q has a nonzero coarse part; and if N is quasi-regular, then N is coarse to Q. The corollaries—no diffuse quasi-regular compressible subalgebra, and basic-construction inclusions that are ergodic but not AFD-ergodic even after tensoring by B(l2N)—are exactly the kind of structural results people in the II1 area will care about. The notion is new, and the main implication goes beyond the Ge–Popa decomposition results in a way that looks substantive, not like repackaging.

What is good: the definition is quantitative and the examples are not afterthoughts. The proof of Theorem 2.3 is direct and convincing: compressibility plus flatness forces M^op′ ∩ (N ∨ Q^op) ⊂ N ∨ B^op, and the averaging argument is clean. Theorem 2.5's construction of the conditional expectation is also plausible. The paper cites the load-bearing norm estimate from Akemann–Ostrand and Popa–Vaes rather than proving it—that is legitimate, since those are published and independently checkable. The abstract and introduction match the content, and the final remarks about why this approach cannot solve Connes' embedding problem are honest and useful.

The soft spot is real, and the stress-test note is right. Lemma 2.7 asserts that quasi-normalizers of N and Q leave the coarse part invariant, and the proof is left as an exercise. The quasi-regular conclusions in Corollary 2.9 and Corollary 2.8 need exactly that lemma, and the reduction to a finite-index N0 is not a formality: it uses intertwining-by-bimodules in a way that is not spelled out. I do not see circularity or data-fitting here, but the claim is load-bearing and unverified as written. A referee should be told to focus there. There are two smaller presentation gaps: Proposition 1.4 states a matrix-amplified inequality but the proof writes only scalar x; the matrix extension via Lemma 1.2 is probably routine but not shown. And Corollary 2.10 cites [P19] where [P19a] is meant. Citation patterns otherwise look fine.

Who this is for: operator algebraists working on II1 factors, coarse decompositions, tightness, and Kadison-type questions. It is not an entry-level paper, but the main ideas travel well. The reader's conditional verdict is about right. I would send this to a serious referee, with the explicit instruction that Lemma 2.7 and Proposition 1.4 need full details before the quasi-regular claims can be accepted. I would not desk-reject; the paper deserves referee time even in its current conditional state.

Referee Report

3 major / 4 minor

Summary. The paper introduces a new property, compressibility, for a W*-inclusion Q⊂M in a tracial von Neumann algebra: by averaging with finitely many unitaries of M, every matrix amplification of an element of Q can be pushed uniformly close to the scalar (or relative) part. It shows that free independence of Q from a diffuse subalgebra implies compressibility, and then proves the main structural result: if Q⊂M is compressible, then for every AFD subalgebra N⊂M the Hilbert bimodule _N L^2M_Q contains a nonzero coarse part; if N is quasi-regular in M, then N is coarse to Q. Consequences include the absence of diffuse quasi-regular compressible subalgebras, non-amenability of groups with an infinite compressible subgroup, and examples of basic construction inclusions M⊂⟨M,e_Q⟩ that are ergodic but not AFD-ergodic, even after stabilization by B(ℓ^2N). The proofs combine flatness of AFD bimodules (Effros–Lance), the Kesten-type norm estimates of Akemann–Ostrand and Popa–Vaes, and compressibility arguments via Kaplansky density.

Significance. If the results are correct, compressibility is a natural and powerful obstruction to tightness and AFD-ergodicity in II_1 factors. The main theorem is a clean statement with interesting consequences for free group factors, ultraproduct factors, and Kadison-type MASA problems. The paper builds on established results ([AO77], [PV14], [EL77], [P03]) and the author's earlier work, and the definition is motivated by several concrete examples. The organization is clear, and the examples in Section 1 are useful. The central claim does not appear to reduce to a definitional identity or to a fitted quantity; it relies on genuine analytical estimates. However, as discussed below, one load-bearing lemma is left as an exercise and several matrix-extension passages are not written out.

major comments (3)

[§2.7] Lemma 2.7 is load-bearing and its proof is omitted ('easy consequence... leave the details as an exercise'). The quasi-regular clause of Corollary 2.9 and Theorem 0.1, as well as Corollary 2.8, require the invariance of the coarse part under the weak closures of the quasi-normalizers of N and Q. The suggested reduction to a finite-index subalgebra N0⊂N is not immediate: one must prove both inclusions in (N0 L^2M_Q)_co = (N L^2M_Q)_co and then justify the quasi-normalizer invariance via the intertwining-by-bimodules theorem. This is a genuine gap in the proof of the main theorem, not merely a presentation issue.
[§1.4, §2.3–2.4] The definition of compressibility is stated for all matrix amplifications Q⊗M_K, but the proof of Proposition 1.4 verifies the estimate only for a trace-zero x∈(Q)_1. The passage to x∈(Q⊗M_K)_1 is not written. Similarly, Theorems 2.3 and 2.4 apply compressibility to elements of N∨alg Q^op; the reduction to the matrix form of compressibility (block-matrix approximation plus Kaplansky density) should be made explicit. Without this, the examples in Corollary 1.5 and the subsequent theorems are not fully justified as written.
[§2.10] In the proof of Corollary 2.10, the existence of a normal conditional expectation Φ:Z(R)'→R is attributed to 'part 1◦ of Theorem 2.4', but Theorem 2.4 asserts compressibility, not the existence of Φ; the relevant statement appears to be Theorem 2.5.1. Also, the inference 'R has a type I direct summand, so there exists a non-zero projection p∈R such that pRp is abelian. Thus ... MASA-ergodic' is not justified as written: a type I direct summand may be M_n(C) with n>1, and a nonzero abelian corner does not by itself produce a unital MASA of M1 inside M. The proof needs to show that R is abelian (or otherwise construct the MASA) and to cite the correct theorem.

minor comments (4)

[§2.10] The reference '[P19]' is not in the bibliography; it should be '[P19a]' or '[P19b]' depending on the intended result.
[§2.5, part 3] The notation 'xpH0 ∈ B(H0)' should presumably be 'x|_{H0}' or 'x p_{H0}' with the projection p_{H0} defined; please clarify.
[§1.3] The numbering in Lemma 1.3 has two items marked '3◦'; renumber for clarity.
[Throughout] There are several typos, e.g. 'F actors' in the title, 'ergodc' and 'MASA-ergodc' in Section 2.10, and 'propreté' in the text. These should be corrected.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the main implication is derived from the compressibility definition, not assumed; only a load-bearing omitted proof (Lemma 2.7) is flagged as a rigor gap.

full rationale

The derivation chain is not circular. Theorem 0.1/Corollary 2.9 take compressibility as a hypothesis and, via flatness (Effros–Lance, [EL77]) and the averaging arguments of Theorems 2.3–2.5, construct a normal conditional expectation whose support projection gives a nonzero coarse submodule. The coarse part is a consequence, not part of the definition of compressibility, and no parameter is fitted to the target claim. The freeness-to-compressibility examples use [PV14], but that is an independent published theorem used only to supply examples; the main implication is conditional and does not reduce to it. The only in-scope issue flagged by the review rule is an omitted proof: Lemma 2.7 asserts invariance of the coarse part under quasi-normalizers and says 'We leave the details as an exercise.' This lemma is load-bearing for the quasi-regular clause (Corollary 2.8 and Corollary 2.9.2). That is a rigor gap/correctness risk, not a circular reduction to the paper's own inputs, so it does not raise the circularity score.

Assumptions & free parameters 0 free parameters · 7 assumptions · 0 invented entities

No free parameters fitted to data. The central theorem imports standard results in operator algebras; the most consequential are listed. None are ad hoc to this paper, and no new 'entities' in the sense of particles, forces, or dimensions are introduced.

assumptions (7)

standard math Any Hilbert N-M bimodule with N or M AFD is flat (Effros–Lance theorem).
Lemma 2.2(1); this is the premise that allows Theorems 2.3–2.5 and Cor 2.9 to convert compressibility into the existence of a coarse sub-bimodule.
standard math L-free sets in a tracial W*-algebra can be dilated to L-free unitaries and satisfy the Kesten-type norm estimate ||(1/n)Σ V_k|| ≤ 2√(n−1)/n.
Proposition 1.4 uses [PV14] (3.4) and [AO77] to prove free-product examples are compressible.
standard math For any separable Q in an ultraproduct II1 factor, there exists a diffuse abelian A free independent to Q.
Cor 1.5(4) via [P92]; this yields many compressible inclusions.
standard math Interpolated free group factors decompose as N ∗ A with A abelian diffuse (Dykema).
Cor 1.7 and Prop 1.6 use this to build exhausting chains of compressible subfactors.
standard math A MASA-ergodic inclusion of factors is R-ergodic.
Cor 2.10 uses [P19a] to convert the contradiction into non-AFD-ergodicity.
standard math Asymptotic freeness in tracial ultraproducts: A′∩M^ω is free independent to A0′∩M^ω for diffuse abelian A⊂N, A0⊂N0.
Cor 1.5(5) uses [HI23] to show relative commutants in ultraproducts are compressible.
standard math Lemma 1.2 in [P81a] allows refining finite partitions to approximate conditional expectations of a MASA.
Section 3.3.2's construction of a MASA with no free elements relies on it.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Compressible subalgebras in II$_1$ factors." pith.science (2026). https://pith.science/paper/5ZTLQ6OK

@misc{pith2026251017076,
  author       = {Pith},
  title        = {Pith review of: Compressible subalgebras in II$_1$ factors},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/5ZTLQ6OK}},
  note         = {Machine review of arXiv:2510.17076}
}

abstract

Given a II$_1$ factor $M$, a W$^*$-subalgebra $Q\subset M$ is {\it compressible} if for any $\varepsilon>0$ there exists a finite set of unitary elements $\Cal U_0\subset \Cal U(M)$ such that $\| \frac{1}{|\Cal U_0|}\sum_{u\in \Cal U_0} uxu^* -E_{1\otimes \Bbb M_K(\Bbb C)}(x)\|\leq \varepsilon$, $\forall K\geq 1$, $\forall x\in (Q\otimes \Bbb M_K(\Bbb C))_1$. Any W$^*$-subalgebra $Q$ in a II$_1$ factor $M$ which admits a diffuse W$^*$-algebra $Q_0\subset M$ that's free independent to $Q$, is compressible in $M$. We prove that if $Q\subset M$ is compressible, then $_NL^2M_Q$ contains a copy of the coarse $N-Q$ bimodule for any AFD subalgebra $N\subset M$. We use this result to provide examples of inclusions of factors $M\subset \Cal M$ that are ergodic but not AFD-ergodic, even after stabilizing by $\Cal B(\ell^2\Bbb N)$.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

6 extracted references · 3 linked inside Pith

[1]

An introduction to II 1 factors

[AO76] C.A. Akemann, P.A. Ostrand: Computing norms in group C∗-algebras. Amer. J. Math. 98 (1976), 1015-1047 [AP17] C. Anantharaman, S. Popa: “An introduction to II 1 factors”, www.math.ucla.edu/∼popa/Books/IIun-v13.pdf [BeCa22] S. Belinschi, M. Capitaine: Strong convergence of tensor products of independent G.U.E. matrices , arXiv:2205.07695 [BoCo23] C. ...

arXiv 1976
[135]

C ∗-algebras and W ∗-algebras

[EL77] E. Eﬀros, C. Lance: Tensor products of operator algebras , Advances in Math., 25 (1977), 1-34. [GP96] L. Ge, S. Popa: On some decomposition properties for factors of type II1, Duke Math. J., 94 (1998), 79-101. [Ha15] B. Hayes: 1- Bounded entropy and regularity problems in von Neumann alge bras. International Math. Res. Notices, 1 (2018), 57-137 (ar...

arXiv 1977
[1957]

Dykema: Free products of hyperﬁnite von Neumann algebras and free di men- sion, Duke Math

[Dy93] K. Dykema: Free products of hyperﬁnite von Neumann algebras and free di men- sion, Duke Math. J. 69 (1993), 97-119. [Dy94] K.J.Dykema: Interpolated free group factors , Paciﬁc J. Math. 163 (1994),123-

1993
[1971]

Theory of operator algebras

[Sc63] J. Schwartz: Two ﬁnite, non-hyperﬁnite, non-isomorphic factors , Comm. Pure App. Math. 16 (1963) 19-26. [T79] M. Takesaki: “Theory of operator algebras”, I, Spring er-Verlar, New York

1963
[1979]

Voiculescu: Circular and semicircular systems and free product factors , Prog

[Vo88] D. Voiculescu: Circular and semicircular systems and free product factors , Prog. in Math. 92, Birkhauser, Boston, 1990, pp. 45-60. Math.Dept., UCLA, LA, CA 90095-155505 E-mail address : popa@math.ucla.edu

1990
[2008]

Les alg´ ebres d’op´ erateurs sur lespace Hilbertien (Alg´ ebres de von Neumann)

[BDIP23] R. Boutonnet, D. Drimbe, A. Ioana, S. Popa: Non-isomorphism of A∗n, 2 ≤ n ≤ ∞, for a non-separable abelian von Neumann algebra A, GAF A 2024, https://doi.org/10.1007/s00039-024-00669-8, 2024 (arX iv:2308.05671). 20 SORIN POPA [C76] A. Connes: Classiﬁcation of injective factors , Ann. of Math., 104 (1976), 73-115. [D57] J. Dixmier: “Les alg´ ebres...

arXiv 2024

Pith tools

Reviewed August 4, 2026 · model on record in the stance chip above.

[1] [1]

An introduction to II 1 factors

[AO76] C.A. Akemann, P.A. Ostrand: Computing norms in group C∗-algebras. Amer. J. Math. 98 (1976), 1015-1047 [AP17] C. Anantharaman, S. Popa: “An introduction to II 1 factors”, www.math.ucla.edu/∼popa/Books/IIun-v13.pdf [BeCa22] S. Belinschi, M. Capitaine: Strong convergence of tensor products of independent G.U.E. matrices , arXiv:2205.07695 [BoCo23] C. ...

arXiv 1976

[2] [135]

C ∗-algebras and W ∗-algebras

[EL77] E. Eﬀros, C. Lance: Tensor products of operator algebras , Advances in Math., 25 (1977), 1-34. [GP96] L. Ge, S. Popa: On some decomposition properties for factors of type II1, Duke Math. J., 94 (1998), 79-101. [Ha15] B. Hayes: 1- Bounded entropy and regularity problems in von Neumann alge bras. International Math. Res. Notices, 1 (2018), 57-137 (ar...

arXiv 1977

[3] [1957]

Dykema: Free products of hyperﬁnite von Neumann algebras and free di men- sion, Duke Math

[Dy93] K. Dykema: Free products of hyperﬁnite von Neumann algebras and free di men- sion, Duke Math. J. 69 (1993), 97-119. [Dy94] K.J.Dykema: Interpolated free group factors , Paciﬁc J. Math. 163 (1994),123-

1993

[4] [1971]

Theory of operator algebras

[Sc63] J. Schwartz: Two ﬁnite, non-hyperﬁnite, non-isomorphic factors , Comm. Pure App. Math. 16 (1963) 19-26. [T79] M. Takesaki: “Theory of operator algebras”, I, Spring er-Verlar, New York

1963

[5] [1979]

Voiculescu: Circular and semicircular systems and free product factors , Prog

[Vo88] D. Voiculescu: Circular and semicircular systems and free product factors , Prog. in Math. 92, Birkhauser, Boston, 1990, pp. 45-60. Math.Dept., UCLA, LA, CA 90095-155505 E-mail address : popa@math.ucla.edu

1990

[6] [2008]

Les alg´ ebres d’op´ erateurs sur lespace Hilbertien (Alg´ ebres de von Neumann)

[BDIP23] R. Boutonnet, D. Drimbe, A. Ioana, S. Popa: Non-isomorphism of A∗n, 2 ≤ n ≤ ∞, for a non-separable abelian von Neumann algebra A, GAF A 2024, https://doi.org/10.1007/s00039-024-00669-8, 2024 (arX iv:2308.05671). 20 SORIN POPA [C76] A. Connes: Classiﬁcation of injective factors , Ann. of Math., 104 (1976), 73-115. [D57] J. Dixmier: “Les alg´ ebres...

arXiv 2024