{"id":"1008bbe9-a881-4c74-8222-5bdf65f9f5d5","arxiv_id":"2505.19508","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Amalgamated free products of weakly exact tracial von Neumann algebras over injective amalgams are relatively biexact, and injective free products over a mixing amalgam are biexact.","lead":"This paper proves that amalgamated free products of weakly exact von Neumann algebras over an injective common subalgebra are biexact relative to the factors, and that in the injective case they are biexact relative to the amalgam; if the amalgam is mixing, the free product is actually biexact. These results extend biexactness-based rigidity theorems from groups to von Neumann algebras and yield structural decomposition and Kurosh-type subalgebra results.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Corollary 1.7 depends on the new Definition 5.4 of mixing and the sketched Lemma 5.5; the non-tracial definition is not shown equivalent to standard mixing, so the upgrade to genuine biexactness is not fully verified.","rationale":"The reader identified as the weakest assumption the new Definition 5.4 of mixing and the sketched proof of Lemma 5.5, and this is also the most load-bearing concern in my read. Theorem 1.1 and the relative biexactness theorems (Corollary 1.2 and Theorem 1.3) have substantial proof infrastructure: the Toeplitz-Pimsner algebra argument in Section 4, the reduction to condition (A) for the tracial case, and the word-reduction arguments in Proposition 4.16. The paper supports these with detailed arguments, and I do not see a concrete flaw in them on a careful reading. Corollary 1.7, however, is precisely the advertised upgrade from relative biexactness to genuine biexactness, and it is the statement used in the lamplighter example. That upgrade depends on the new mixing notion and on Lemma 5.5, whose proof is compressed and whose hypothesis is not proved equivalent to the classical tracial notion. If Lemma 5.5 is incorrect, the relative biexactness results survive but the headline claim about actual biexactness for mixing amalgams does not follow. This is a genuine correctness risk, not a disagreement with the field's consensus, and it is internal to the proof structure. The concrete test of redoing Lemma 5.5 in the L(Z)∗L(Z) case would settle the core reduction identity, and the non-tracial lamplighter check would verify that the new definition is satisfied in the claimed example. The reader's CONDITIONAL verdict is therefore appropriate; I would not strengthen or weaken it without seeing the test performed.","tokens_in":46278,"tokens_out":28181,"duration_ms":231697,"concrete_test":"Specialize to the tracial free product M1 = M2 = L(Z), B = C1, so M = L(F2), where the conclusion of Lemma 5.5 is classical: L(Z) is mixing in L(F2). Reprove the lemma in this case using only the compact-operator criterion: for x, y in alternating words in L(F2) ⊖ L(Z), explicitly compute e_{L(Z)} x J y J e_{L(Z)} and verify whether the displayed reduction eM1 x1...xn J y1...ym J eM1 = eB(x1Jy1J)eB...eB(xnJynJ)eB is an operator identity. If the identity fails for any word with i1=in=j1=jm=2, the proof collapses. If it holds, perform the same verification in the non-tracial lamplighter crossed-product example of Corollary 1.7, checking Definition 5.4 directly by showing eB x J y J eB ∈ K∞,1_{K(L2M)}(M) rather than relying on compactness.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The advertised upgrade Corollary 1.7 (Theorem 5.23) rests on a chain: Theorem 1.3 gives biexactness relative to B, Corollary 5.7 produces B, M1, M2 as mixing inside M via Lemma 5.5, and Corollary 5.22 upgrades to genuine biexactness. The pivotal Lemma 5.5 is proved in a short sketch: it reduces to alternating centered products and then asserts that eB(x1Jy1J)eB(x2Jy2J)eB...eB(xnJynJ)eB lies in K∞,1(M) because B is mixing in M2 and K∞,1(M2) embeds in K∞,1(M). The weak point is Definition 5.4 itself: it defines mixing using membership in K∞,1_{K(L2M)}(M), a closure of compact operators in the M-M and M'-M' topologies, not compactness. For a tracial M this space is known to be larger than K(L2M), and the paper neither proves that Definition 5.4 restricts to the classical compactness criterion nor supplies the missing estimates in Lemma 5.5: the validity of the displayed reduction for words with i1=in=j1=jm=2, the induction that peels off pairs (xk J yk J), and the passage from K∞,1(M2) to K∞,1(M) under the natural inclusion. If the definition is subtly weaker than standard mixing, or if Lemma 5.5 has a hidden assumption, Corollary 5.23 fails even though the relative biexactness theorems survive. This is a correctness risk in the central claim that mixing amalgams yield genuinely biexact free products.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves relative biexactness results for amalgamated free product (AFP) von Neumann algebras. Its main theorems are: (1) when M1,...,Mn are weakly exact tracial von Neumann algebras with a common injective amalgam B and trace-preserving conditional expectations, the AFP is biexact relative to {M1,...,Mn} (Corollary 1.2); (2) when the Mi are injective and B is a common subalgebra with faithful normal conditional expectations, the AFP is biexact relative to B (Theorem 4.17/1.3); and (3) when the Mi are separable injective and B is mixing in each Mi, the AFP is genuinely biexact (Corollary 5.23/1.7). The paper also derives structural decomposition and absorption theorems for tracial von Neumann algebras (Theorem 1.4) and a Kurosh-type rigidity result (Corollary 1.6). The proofs use a Toeplitz-Pimsner algebra construction to factor the inclusion of the reduced AFP into an amalgamated free product of basic constructions, followed by computations of small-at-infinity boundaries.","tokens_in":46632,"tokens_out":26928,"duration_ms":261406,"significance":"If the main claims are correct, this is a substantial contribution: it extends Ozawa's and Ding--Peterson's relative biexactness technology to amalgamated free products of von Neumann algebras, works without assuming σ-finiteness, and yields new subalgebra rigidity results. The Toeplitz-Pimsner approach to nuclear embeddings in this setting is a promising and nontrivial tool. The paper is careful about many technical points, such as relative tensor products for non-σ-finite algebras and the definition of the small-at-infinity boundary. However, the advertised upgrade to genuine biexactness for mixing amalgams rests on a new definition of mixing (Definition 5.4) and on Lemma 5.5, whose proof is only sketched and contains a questionable reduction. The relative biexactness theorems themselves are largely independent of that upgrade and appear well supported. The main open issue is therefore localized but load-bearing for Corollary 1.7.","major_comments":[{"comment":"The definition of mixing in Definition 5.4 uses membership in K∞,1(M), the closure of compact operators in the M-M and M′-M′ topologies, rather than membership in K(L2M). In the tracial case K∞,1(M) is generally larger than K(L2M) (see [DEP23, Proposition 3.6]), so this definition is weaker than the classical compactness criterion quoted at the start of Section 5.2. The paper neither proves that Definition 5.4 agrees with the classical notion in the tracial case nor identifies where the difference is immaterial. Consequently Corollary 1.7 is a statement about a new notion, and the examples in the introduction do not automatically transfer to the classical notion of mixing.","section":"Section 5.2, Definition 5.4 and Lemma 5.5"},{"comment":"The proof of Lemma 5.5, which is the pivotal step for Corollary 5.23, is only sketched and contains a reduction that does not appear correct as written. After assuming i1=in=j1=jm=2, the proof asserts eM1x1⋯xnJ y1⋯ymJ(eM1−eB)=0 and then replaces eM1 by eB on the right. For a word beginning with M2, however, eM1λ(x1)⋯=0, so the displayed identity, even if it holds, does not justify the next equality; the case where the word begins with M1 is not reconciled with the reduction to i1=2. The subsequent claim that eB(x1Jy1J)eBT eB(xnJynJ)eB belongs to K∞,1(M) for all T is also justified only by an ideal-theoretic argument, without the convergence estimates needed to pass from K∞,1(M2) to K∞,1(M). Since Corollaries 5.6, 5.7, and 5.23 depend on this lemma, the upgrade from relative biexactness to genuine biexactness is not fully established.","section":"Section 5.2, Lemma 5.5"},{"comment":"The proof of Proposition 4.9 is the heart of Theorem 4.1, but the key multiplicative-domain verification is deferred: the text says one applies 'the same argument as in the proof of [Toy25, Theorem 3.2]' and then gives computations only for T2(T∗)2 and u. The displayed calculations show agreement of Φ♯∗◦π◦Ψ with the canonical inclusion on elements of Ai, but the assertion that this places the generators in the multiplicative domain of the composition is not demonstrated in the manuscript. Because all later relative biexactness results rely on Theorem 4.1, this step should be reproduced in full, or the precise theorem from [Toy25] should be stated and its hypotheses checked explicitly.","section":"Section 4.1, Proposition 4.9"},{"comment":"Theorem 5.14 is proved by saying that it 'essentially follows from the proof of [DKE24, Theorem 1.1]' and then giving a sketch. The final part of the sketch constructs projections fi and concludes that ∨i fi = f0⊥; this conclusion requires the fi to be central and mutually orthogonal, which is asserted but not shown. Similarly, the proof of Corollary 5.19 compresses a substantial rigidity argument involving the flip automorphism and [IPP08, Theorem 1.1] into a few sentences. These arguments may be correct, but the amount of omitted detail is large for statements that are advertised as applications of the main theorems.","section":"Section 5.4, Theorem 5.14 and Corollary 5.19"}],"minor_comments":[{"comment":"There is a typo: 'adapts [DP23, Theorem 5.10] to the the setting of amalgamated free product' should read 'to the setting'.","section":"Section 1, first paragraph"},{"comment":"The Jones basic construction is described as 'the von Neuamm subalgebra'; this should be 'von Neumann subalgebra'.","section":"Section 2.2, first paragraph"},{"comment":"The paper mentions that the usual requirement in the definition of mixing that un be unitary can be omitted, citing [CFM13, Theorem 3.3], but it does not explain how Definition 5.4 relates to this criterion in the tracial case; a short remark would help the reader.","section":"Section 5.2, beginning"},{"comment":"Proposition 5.15 is stated and said to follow by the same proof as [Din25, Proposition 4.2], but it is not used later. Either give the proof or state explicitly that it is included for completeness.","section":"Section 5.3, Proposition 5.15"}],"recommendation":"major_revision","confidential_remarks":"The paper contains substantial new results and a promising method. I believe the relative biexactness theorems are likely correct and the Toeplitz-Pimsner argument is a real contribution. The main concern is the genuine-biexactness upgrade: Lemma 5.5 and Definition 5.4 need a fully detailed proof and a comparison with standard mixing, respectively. This is fixable within the manuscript's scope, so I recommend major revision rather than rejection. The dependence on the first author's [Toy25] is acceptable, but the referee report should ask for explicit statements of the borrowed lemmas."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The core of this paper is solid and worth referee time, but the one headline upgrade—genuine biexactness of mixing amalgamated free products—rests on a new definition that is not yet pinned down.\n\nWhat is actually new: this is the first genuine extension of Ding–Peterson relative biexactness to amalgamated free products of von Neumann algebras. Theorem 1.1 gets an M-nuclear embedding into the free product of basic constructions via a Toeplitz–Pimsner correspondence, without assuming states or sigma-finiteness. Corollary 1.2 gives relative biexactness for weakly exact tracial algebras over injective amalgams; Theorem 1.3 gets relative-to-B biexactness for injective algebras. The applications (structural decomposition, Kurosh-type rigidity, absorption) are real and go beyond routine extension. The paper is honest about the limits of Condition (A): it is verified only in the tracial and finite-dimensional B cases.\n\nMain soft spot: Corollary 1.7 (Theorem 5.23). The upgrade from relative biexactness to biexactness uses Definition 5.4, where 'mixing' means e_N x J y J e_N ∈ K∞,1(M), a closure of compact operators in the M-M and M′-M′ topologies. That is weaker than compactness in the tracial setting, and the paper never shows that this definition agrees with the standard tracial mixing criterion. Lemma 5.5, which is the bridge that makes M1 mixing in M1∗_B M2 when B is mixing in M2, is proved in a short sketch; the reduction to alternating centered products and the induction peeling off pairs (x_k J y_k J) is asserted more than shown. If Definition 5.4 is subtly weaker than intended, or if Lemma 5.5 needs a hidden separability/faithfulness assumption, Corollary 1.7 fails while the relative biexactness theorems survive. The referee should press hard on Section 5.2.\n\nModerate caveat: Proposition 4.9 checks a multiplicative-domain identity by reference to [Toy25, Theorem 3.2]. That is a published, independently verified toolbox, so not circular, but it is load-bearing. The referee should confirm the hypotheses of that theorem are met here. Minor typos ('nomral', 'subgalgebra') don't affect the math.\n\nWho it's for: people working in deformation/rigidity of von Neumann algebras and exact groups. I'd cite Corollary 1.2 and Theorem 1.3. I'd take the main theorems as credible, but I would not rely on Corollary 1.7 until Section 5.2 is rewritten. Send it to a serious referee.","headline":"The relative biexactness theorems are real and likely correct, but the advertised upgrade to genuine biexactness under mixing rests on a new definition that needs more work before I'd trust it.","tokens_in":47201,"tokens_out":2748,"would_cite":true,"duration_ms":27406,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["46L10","46L54"],"pacs":[],"model":"deepseek-v4-flash","headline":"Amalgamated free products of injective von Neumann algebras are biexact relative to their common amalgam, and biexact outright when the amalgam is mixing.","keywords":["biexact von Neumann algebras","amalgamated free product","relative biexactness","small-at-infinity boundary","mixing subalgebras","injective von Neumann algebras","subalgebra absorption","Kurosh-type decomposition"],"falsifier":"Find two separable injective von Neumann algebras $M_1$, $M_2$ sharing a mixing amalgam $B$ for which the operator $e_B x J y J e_B$ for some $x,y\\in M$ falls outside $K^{\\infty,1}(M)$; this would violate the key step of Lemma 5.5 and thereby the upgrade to Corollary 1.7. A simpler check is whether Definition 5.4 restricted to tracial $M$ is equivalent to the standard mixing condition: any tracial pair that is mixing classically but not under Definition 5.4 would refute the paper's implicit equivalence.","tokens_in":46036,"feed_emoji":"🔗","tokens_out":8516,"duration_ms":78295,"temperature":0.7,"pith_summary":"This paper asks when an amalgamated free product of von Neumann algebras inherits biexactness, a structural finiteness property that in the group case controls how subalgebras sit inside a larger algebra. Its central result is that if several injective von Neumann algebras share an injective subalgebra $B$ with faithful normal conditional expectations, then the amalgamated free product is biexact relative to $B$, and if $B$ is “mixing” in each factor and the factors are separable, the product is biexact outright. In the tracial weakly exact case, the product is at least biexact relative to the family of factors, which yields a decomposition theorem for arbitrary subalgebras into properly proximal, amenable, and relatively amenable parts. The proof works through a Toeplitz-Pimsner algebra built from the free-product Hilbert space; this lets the paper prove the required nuclearity without any assumption on states, so the result covers non-$\\sigma$-finite algebras.","feed_headline":"Shared injective subalgebras make free products biexact","feed_subtitle":"Subalgebras either embed into the amalgam or have amenable relative commutants; free-product factors decompose uniquely.","key_machinery":"The load-bearing mechanism is the Toeplitz-Pimsner algebra $T(H)$ associated with a $C^*$-correspondence over $A_1\\oplus A_2$, where $A_i$ is an ultraweakly dense $C^*$-subalgebra of $M_i$. A creation operator $T$ on the free-product Hilbert space encodes the interleaving of the two factors, and the gauge-invariant uniqueness theorem identifies $C^*(A_1\\oplus A_2, T)$ with $T(H)$; a u.c.p. map $\\Phi$ then sends words in $T$ and the $A_i$'s into the reduced amalgamated free product. This factorization proves that the inclusion of the reduced free product into a free product of basic constructions is nuclear in the appropriate relative sense. The small-at-infinity boundary $S_X(M)$ and the “mixing” condition (Definition 5.4) are the objects that convert this nuclearity into relative and then genuine biexactness.","core_discovery":"The paper's central claim is that amalgamation preserves biexactness in the relative sense: for injective $M_1,\\dots,M_n$ with common injective $B$ and faithful normal conditional expectations, the amalgamated free product $M=M_1 *_B \\cdots *_B M_n$ is biexact relative to $B$ (Theorem 1.3). It also claims that in the weakly exact tracial case the same product is biexact relative to the collection $\\{M_1,\\dots,M_n\\}$ whenever $B$ is injective (Corollary 1.2), and that when each $M_i$ is separable injective and $B$ is mixing in each factor, $M$ itself is biexact (Corollary 1.7). On top of these biexactness statements, the paper proves structural decomposition and subalgebra absorption theorems, including a Kurosh-type uniqueness result for free products of weakly exact nonamenable non-properly proximal II$_1$ factors.","pith_inferences":["If Definition 5.4 of “mixing” is shown to match the classical tracial notion in full generality, Corollary 1.7 would extend to a wider class of non-tracial inclusions, not just separable injective ones; the paper only verifies the needed approximation in limited cases.","The Toeplitz-Pimsner mechanism is likely to prove relative biexactness for other amalgamated constructions, such as free products over finite-dimensional amalgams or crossed products by amalgamated actions, where the same creation-operator words remain tractable.","A natural test of the upgrade method is whether it can be run with a boundary piece generated by a single non-mixing subalgebra; if the small-at-infinity boundary is too large, the projection decomposition in Theorem 5.14 would fail.","The Kurosh-type result suggests that free-product factors with all factors weakly exact, nonamenable, and non-properly proximal admit a prime-like decomposition, and one could ask whether the non-proper-proximality assumption can be relaxed to mere non-biexactness."],"forward_implications":["If the main theorems hold, every finite von Neumann subalgebra $N$ of an amalgamated free product of injective algebras either embeds into the amalgam $B$ or has amenable relative commutant (Corollary 4.18).","In the tracial weakly exact case, arbitrary subalgebras decompose into a properly proximal part, an amenable part, and parts amenable relative to each factor (Theorem 1.4).","Free products of weakly exact nonamenable non-properly proximal II$_1$ factors have a unique Kurosh-type decomposition, up to unitary conjugacy and permutation of the factors (Corollary 1.6).","When $B$ is mixing in each factor, the whole amalgamated free product is biexact, so the lamplighter Bernoulli crossed product example gives a concrete family of genuine biexact algebras beyond the group case.","The relative biexactness conclusions do not require $\\sigma$-finiteness or the existence of a state, unlike earlier free-product biexactness results."],"supporting_citations":[{"why":"Introduces relative biexactness for von Neumann algebras via $M$-nuclearity of the inclusion into the small-at-infinity boundary; the theorem being adapted and the nuclearity criterion used in Section 5.5.","marker":"[DP23]"},{"why":"Supplies the Toeplitz-Pimsner C*-correspondence technique and the nuclear embedding lemma used to prove Theorem 4.1.","marker":"[Toy25]"},{"why":"Gives the group-case relative biexactness result and the creation-operator construction on the free-product space that the paper reinterprets.","marker":"[BO08]"},{"why":"Defines the small-at-infinity boundary $S_X(M)$, boundary pieces, and proper proximality, which frame the decomposition theorems.","marker":"[DEP23]"},{"why":"Provides the upgrading method and projection identities for biduals of hereditary subalgebras used in Section 5.4.","marker":"[DKE24]"},{"why":"Supplies the earlier rigidity theorems for free products that the new biexactness statements extend.","marker":"[HU16]"},{"why":"Gives the amenable-absorption theorem used to conclude $A\\subset M_1$ in Corollary 1.5.","marker":"[BH18]"},{"why":"Contains the normalizer and mixing analysis in amalgamated free products underlying Lemma 5.5.","marker":"[Vae14]"},{"why":"Provides the intertwining result for amalgamated free products used in the Kurosh-type decomposition argument.","marker":"[IPP08]"}],"fun_headline_variants":["Free products over injective amalgams stay biexact","Biexactness survives amalgamated free products","Amalgamated free products preserve relative biexactness","Injective amalgams yield biexact free products","Relative biexactness for free product von Neumann algebras"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole upgrade to genuine biexactness rests on the new definition of a mixing subalgebra for general von Neumann algebras and on the approximation argument in Lemma 5.5; if this definition does not behave like the classical tracial mixing condition, Corollary 1.7 could fail even though the relative biexactness theorems survive.","fun_headline_variants_meta":{"raw":{"variants":["Free products over injective amalgams stay biexact","Biexactness survives amalgamated free products","Amalgamated free products preserve relative biexactness","Injective amalgams yield biexact free products","Relative biexactness for free product von Neumann algebras"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000166,"raw_usage":{"total_tokens":1241,"prompt_tokens":922,"completion_tokens":319,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":538,"completion_tokens_details":{"reasoning_tokens":238}},"tokens_in":538,"tokens_out":319,"duration_ms":3719,"temperature":1.0,"reasoning_tokens":238,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T14:13:08.349713+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Find two separable injective von Neumann algebras $M_1$, $M_2$ sharing a mixing amalgam $B$ for which the operator $e_B x J y J e_B$ for some $x,y\\in M$ falls outside $K^{\\infty,1}(M)$; this would violate the key step of Lemma 5.5 and thereby the upgrade to Corollary 1.7. A simpler check is whether Definition 5.4 restricted to tracial $M$ is equivalent to the standard mixing condition: any tracial pair that is mixing classically but not under Definition 5.4 would refute the paper's implicit equivalence.","supporting_citations":[],"review_version":1}