{"id":"42797944-5e9d-41e3-9404-7f2fc923b85c","arxiv_id":"2607.09627","paper_version":3,"verdict":"ACCEPT","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"low","formal_verification":"none","parameter_count":2,"one_line_summary":"A complete ultrametric on the label set of von Neumann's incomplete tensor products is defined by the convergence exponent of the overlap series, with a gauge-invariant version serving as a decoherence exponent.","lead":"This paper introduces a distance between the sectors of von Neumann's infinite tensor product: the further apart two sectors are, the faster the series of overlaps between their defining sequences diverges. The label set of sectors becomes a complete ultrametric space, and the distance is interpreted as a decoherence rate in a toy model of Everettian branching.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No significant objection identified","rationale":"The reader's ACCEPT is justified. My independent check of the proof of Theorem 3.5—the completeness lemma in particular—found no flaw: the block construction is legitimate, the tail estimates are finite in number per stage, and the final limit satisfies the Cauchy condition. The gauge-invariant analogue in Section 4 inherits the same structure. The enumeration dependence highlighted by the reader is real but explicitly acknowledged and characterized by the invariance group G in Section 3.1; it does not affect the truth of the fixed-enumeration theorem. Similarly, Proposition 6.3's equality between dtilde and the polynomial decay exponent of the overlap requires a uniform lower bound on per-site overlaps, which is stated; without it, the physical reading is only an upper/lower bound, not the formal metric claim. These are limitations of interpretation, not of the central construction. I agree with the reader that no fitted parameters, circular steps, or omitted proofs occur in the main argument, so no verdict change is warranted.","tokens_in":29685,"tokens_out":29011,"duration_ms":283769,"concrete_test":"Independently re-derive the completeness proof for an explicit Cauchy sequence whose pairwise distances are exactly 2^{-min(n,m)}, e.g. built from unit vectors in C^2 with blockwise flip defects. Verify that the tail bounds (21) can be chosen for all k<l and that the assembled limit c* satisfies d(c*,c_n)→0. Also recompute the example after Lemma 3.1 (a_j=1/j) to confirm d=0 while φ≁ψ, and check the quotient relation d=0 does not collapse distinct weak classes.","verdict_should_be":"UNCHANGED","load_bearing_attack":"I traced the central proof chain through Lemmas 2.5, 3.2–3.4 and Theorem 4.3. The well-definedness of d under ∼ is correctly reduced to Lemma 2.5; the strong triangle inequality uses convergent weighted norm-square and imaginary-part series, and the completeness proof's block construction is internally sound: the tail bounds (21) are satisfiable because for each l only finitely many (k,p) pairs occur, and Step 3's choice l0=max(k+1,m) guarantees p′∈P_l. The quotient by d=0 is handled correctly. The two most fragile interpretive points—enumeration dependence (Section 3 remark) and the condition inf_j |⟨φ_j,ψ_j⟩|>0 in Proposition 6.3 for the exact decoherence-rate equality—are explicitly stated limitations rather than hidden gaps. I found no unsupported step in the formal construction of the complete ultrametric.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper introduces a quantitative structure on the set Γ of strong equivalence classes of C0-sequences in von Neumann's infinite tensor product theory. For two C0-sequences φ,ψ it defines d(φ,ψ) as the convergence exponent of ∑_j |⟨φ_j,ψ_j⟩−1|, proves that d is a well-defined pseudo-ultrametric on Γ, satisfies the strong triangle inequality, and is complete; after quotienting by d=0, (Γbar,d) is a complete ultrametric space (Theorems 3.5 and 4.3). A gauge-invariant variant dtilde replaces |⟨φ_j,ψ_j⟩−1| by 1−|⟨φ_j,ψ_j⟩| and is developed on weak-equivalence classes. The remainder of the paper applies this geometry: product unitaries with δ(U)>0 displace every class to distance 1 (Theorem 5.2); qubit and Bell-pair examples realize all displacements in [0,1]; in a toy Everett model, dtilde is interpreted as a decoherence exponent (Proposition 6.3); and in the ITPFI setting, the one-sided modular flow's displacement detects Connes' T(M) in the Powers case and grades the Araki–Woods tracial boundary (Theorems 7.4–7.5).","tokens_in":29944,"tokens_out":38464,"duration_ms":380428,"significance":"The formal construction is sound. The proofs of Lemmas 3.2–3.4 and Theorem 3.5 are internally consistent; the partial-sum characterization (13) is correctly used; the completeness proof's block construction is valid. The gauge-invariant variant's proofs are likewise correct, and the delegation of completeness to Lemma 3.4 is acceptable because the required ingredients are explicitly listed. The metric is parameter-free apart from the weight family j^{-p} and the fixed enumeration, both explicitly acknowledged; the enumeration-dependence caveat is not a hidden gap. The applications are suggestive and, in the case of Theorem 5.2 and Section 6, appropriately qualified as a toy model. The operator-algebra section is computationally careful. If the results hold, this enriches the previously unstructured label set of von Neumann's decomposition with a genuine geometric/completeness structure and provides a quantitative decoherence exponent. I found no circularity: the metric is defined before any application, and the Section 7 identification with Powers' criterion is a post-hoc external grounding, not an input.","major_comments":[],"minor_comments":[{"comment":"The parenthetical statement 'For type III_1 one has T(M)={0}' is not correct in this generality; for the hyperfinite III_1 factor, T(M)=R (III_1 factors can also have T(M)={0}). Since this remark motivates an open question but is not used in the proof of Theorem 7.4, this is a local correction: please qualify to the intended subclass or correct the statement.","section":"Remark after Theorem 7.4"},{"comment":"Completeness is delegated to Lemma 3.4 ('applies verbatim'). The delegation is legitimate—the proof only uses unit-norm representatives, the uniform bound on terms, (12), and the strong triangle inequality—but for self-containedness, please spell out the substitutions in a sentence or two.","section":"Theorem 4.3"},{"comment":"In the sentence 'presented in its standard form on the incomplete tensor product N_{[φ]}^j H_j ⊂ H_univ', the notation N_{[φ]}^j appears to be a typo; presumably N_{[φ]} H_j or N_j H_j is intended.","section":"§7.1"},{"comment":"The enumeration dependence of d and dtilde is stated clearly and then characterized by Condition (T). Given the word 'natural' in the abstract, a reader might expect an intrinsic metric; consider adding a sentence in the abstract or introduction that d is defined relative to a fixed enumeration (and its gauge group), so the 'natural metric' is an attribute of the enumerated family.","section":"Remark after Theorem 3.5 / §3.1"}],"recommendation":"minor_revision","confidential_remarks":"The formal core of the paper is correct and I would be happy to see it published. The only substantive concern is the peripheral statement about T(M) for type III_1 in the remark after Theorem 7.4, which should be corrected or qualified. All other comments are local and editorial. The reader's stress-test concern about enumeration dependence is, in my assessment, adequately addressed by the authors' explicit caveat and does not undermine the central theorem."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: worth a serious referee; I would accept with minor revision. The genuinely new piece is treating the convergence exponent of the series in von Neumann's equivalence relation as a distance on the sector label set. The paper proves this is a complete pseudo-ultrametric, becomes a genuine complete ultrametric after quotienting by distance zero, and characterizes exactly which reindexings preserve it (Theorem 3.7). The gauge-invariant dtilde variant is cleaner, connects to Powers' quasi-equivalence criterion, and is used to detect Connes' T(M) in the Powers case and to grade the tracial boundary. I checked the central proof chain — Lemmas 2.5, 3.2–3.4, and the block-completeness argument — and it holds. No fitted parameters, no circularity. The paper also states its own limitations candidly: enumeration dependence is flagged after Theorem 3.5, Section 6 is explicitly a caricature, and Proposition 6.3 needs the per-site overlap bounded below for the exact decoherence-rate identity.\n\nSoft spots are minor and mostly interpretive. The marquee distance depends on a fixed enumeration of N; Theorem 3.7 contains that dependence but doesn't remove it, and the physical reading adds the assumption that the enumeration encodes the environment. That is not a flaw in the formal theorems, just a boundary on what the physics section can claim. The completeness proof for dtilde is delegated ('applies verbatim') rather than written out; acceptable given the parallel with Section 3, but a referee should ask for two sentences to make it self-contained. Section 7 ends with a genuinely open question about the III_0/III_1 regime, and the paper says so — no overclaim. The one place I would push back on framing rather than math: the decoherence exponent is conditional on the overlap lower bound; without it you only get one-sided control R_N ≤ exp(−Σ_N), so the headline 'polynomial rate' is not unconditional.\n\nWho this is for: people working on infinite tensor products, quasi-equivalence of product states, or sector-based decoherence models. It is a useful, contained contribution, not a breakthrough. I would send it to a referee, and I would cite the ultrametric construction if I were working on the label set or on product-state equivalence.\n\nRecommendation: engage. Accept after minor revision; expand the delegated completeness argument slightly, and make the interpretive assumptions of Section 6 explicit where they are introduced.","headline":"Correct, clearly written, and honestly scoped: the paper puts a genuinely new quantitative layer on von Neumann's sector label set, with the main caveat being dependence on a fixed enumeration.","tokens_in":30371,"tokens_out":2665,"would_cite":true,"duration_ms":28879,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["46C05","46L10","54E35"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper establishes that the sectors of von Neumann's incomplete tensor products — the equivalence classes of C0-sequences indexing the complete infinite tensor product — carry a natural complete pseudo-ultrametric distance, the converge","keywords":["ultrametric space","von Neumann infinite tensor products","C0-sequences","convergence exponent","weak equivalence","decoherence exponent","Everett branching","ITPFI factors"],"falsifier":"For φ_j=e_1 and ψ_j=cos(θ_j)e_1+sin(θ_j)e_2 on H_j=C^2 with 1−cos θ_j=1/j, the paper predicts d=0 for two classes that are not von Neumann equivalent: the series Σ_j j^{-p}·j^{-1} converges for every p>0 and diverges at p=0. A direct numerical check of this sum — or, at the level of the physical interpretation, measuring the finite-volume overlap R_N between two branch records and verifying that −log R_N grows polynomially with exponent dtilde rather than, say, exponentially with a large constant — would settle whether the claimed zero-distance phenomenon and the decoherence-exponent reading a","tokens_in":29622,"feed_emoji":"⚛","tokens_out":12414,"duration_ms":115958,"temperature":0.7,"pith_summary":"The paper establishes that the label set of von Neumann's incomplete tensor products, long treated as an unstructured index set, carries a quantitative geometry: the distance between two sectors is the convergence exponent of the series formed from per-component overlaps, and this distance is a complete pseudo-ultrametric, becoming a complete ultrametric after quotienting by distance zero. A gauge-invariant variant, built from the moduli of the overlaps and weak equivalence, controls the polynomial rate at which branch states of a many-worlds toy model become operationally distinct, and the same exponent reappears in the operator-algebraic classification of ITPFI factors. A sympathetic reader would care because the paper converts von Neumann's yes-or-no equivalence dichotomy into a continuous scale with concrete applications to decoherence rates and factor classification.","feed_headline":"Convergence exponent makes sector labels a complete ultrametric space","feed_subtitle":"Sector distance equals the overlap-series convergence exponent — a quantitative decoherence rate for branching worlds.","key_machinery":"The central object is the convergence exponent functional E(a) = inf{p ≥ 0 : Σ_j a_j j^{-p} < ∞}, equivalently limsup_N log⁺(Σ_{j≤N} a_j)/log N, applied to the per-site overlap defects a_j = |⟨φ_j,ψ_j⟩−1| (or 1−|⟨φ_j,ψ_j⟩| for the gauge-invariant variant). This single number — the polynomial growth rate of the partial sums — replaces von Neumann's yes-or-no equivalence dichotomy with a continuous measure of inequivalence. The completeness proof assembles the limit class from blocks of components taken from progressively later Cauchy-sequence members, bypassing the failure of componentwise convergence. For dtilde, a quasi-subadditivity inequality for the deficits 1−|⟨x,z⟩| carries the argumen","core_discovery":"The central claim is that the set Γ of equivalence classes of C0-sequences, which labels the incomplete tensor products inside a complete infinite tensor product, is a complete pseudo-ultrametric space under d(c,d) = inf{ p ≥ 0 : Σ_j j^{-p} |⟨φ_j,ψ_j⟩−1| < ∞ }, the convergence exponent of the overlap series; after quotienting Γ by d=0, the pair (Γ~, d) is a complete ultrametric space. The gauge-invariant variant dtilde, defined with |⟨φ_j,ψ_j⟩| in place of the full inner product and with von Neumann's weak equivalence in place of strong equivalence, satisfies the same completeness and ultrametricity and is naturally matched to quasi-equivalence of product states. Under product unitaries, the","pith_inferences":["The same construction — measuring inequivalence by the convergence exponent of a series of nonnegative per-site defects — should port to any context where a summability dichotomy defines superselection sectors, for example infrared sectors in quantum field theory; the enumeration dependence could then encode the physical ordering of the environment, and an enumeration-invariant quantity could be d","In the branching model, the paper's formal results suggest a testable extension: from finite-N overlap data R_N for two branch records, one could estimate dtilde via limsup log⁺(−log R_N)/log N and derive finite-sample confidence intervals, turning the decoherence exponent into an empirical quantity rather than a purely structural one.","The theorem that the modular-flow displacement is an indicator of Connes' invariant T(M) in the Powers case leaves open a natural conjecture: among weight sequences with inf_j μ_j > 0, dtilde(c,V_t c)=1 for all t≠0 might characterize T(M)={0}; if true, the metric would provide a dynamical characterization of type III_1 factors."],"forward_implications":["A product unitary whose factor U satisfies inf_{∥x∥=1}|⟨x,Ux⟩−1|>0 displaces every sector to the maximal distance d=1; if no eigenvalue of U is a root of unity, the orbit {U^k c} is a 1-separated set in Γ.","The gauge-invariant displacement dtilde(c,Uc) under a product unitary is class dependent and realizes every value in [0,1], interpolating between pointer states and maximally displaced superpositions.","In the Everett toy model, dtilde is a polynomial decoherence exponent: finite-volume absolute overlaps decay like exp(−N^{dtilde}) up to subpolynomial corrections, and the binary tree of branches is organized into a nested hierarchy of ultrametric balls through the strong triangle inequality.","Under the one-sided modular flow of an ITPFI reference state, dtilde detects Connes' invariant T(M) in the Powers case (it is the indicator function of the complement of T(M), hence determines the type) and equals the convergence exponent of the Araki–Woods II_1 series in the asymptotically tracial case, grading the failure of the factor to be of type II_1.","The degeneracy locus {dtilde=0} of the pseudometric corresponds to subpolynomial separation between weakly inequivalent branches — a marginal stratum that also appears as a subpolynomial collar outside the II_1 region of the Araki–Woods classification."],"fun_headline_variants":["Complete ultrametric on tensor product sector labels","Convergence exponent defines ultrametric on incomplete tensor products","Gauge-invariant ultrametric measures decoherence exponent","Product unitaries displace sectors to maximal ultrametric distance","Branching worlds: ultrametric distance as decoherence rate"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"All completeness and application results rest on fixing the standard enumeration of the index set N and the weight family j^{-p}; a permutation outside the T-group characterized in Theorem 3.7 changes the distances between the same pair of sectors, and the physical interpretation further assumes that this fixed enumeration encodes the structure of the environment.","fun_headline_variants_meta":{"raw":{"variants":["Complete ultrametric on tensor product sector labels","Convergence exponent defines ultrametric on incomplete tensor products","Gauge-invariant ultrametric measures decoherence exponent","Product unitaries displace sectors to maximal ultrametric distance","Branching worlds: ultrametric distance as decoherence rate"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000332,"raw_usage":{"total_tokens":1767,"prompt_tokens":910,"completion_tokens":857,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":654,"completion_tokens_details":{"reasoning_tokens":774}},"tokens_in":654,"tokens_out":857,"duration_ms":9277,"temperature":1.0,"reasoning_tokens":774,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-02T07:31:20.378528+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"For φ_j=e_1 and ψ_j=cos(θ_j)e_1+sin(θ_j)e_2 on H_j=C^2 with 1−cos θ_j=1/j, the paper predicts d=0 for two classes that are not von Neumann equivalent: the series Σ_j j^{-p}·j^{-1} converges for every p>0 and diverges at p=0. A direct numerical check of this sum — or, at the level of the physical interpretation, measuring the finite-volume overlap R_N between two branch records and verifying that −log R_N grows polynomially with exponent dtilde rather than, say, exponentially with a large constant — would settle whether the claimed zero-distance phenomenon and the decoherence-exponent reading a","supporting_citations":[],"review_version":2}