{"id":"2b03f200-98b3-4781-bed4-a89725645980","arxiv_id":"2607.26123","paper_version":1,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"Under finite entropy of the manipulated system, infinite-dimensional IID decoupling and quantum state merging achieve the same optimal rates as in finite dimensions (H(A) and 1/2 I(A:R)).","lead":"This paper extends quantum decoupling and state merging from finite-dimensional to infinite-dimensional (separable) systems, proving the same optimal rates whenever the manipulated system has finite entropy. It matters because continuous-variable and field-theoretic settings are infinite-dimensional, and these results show the entropic operational rates survive without finite-dimensional truncation.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No significant objection identified: the finite-entropy assumption is explicit, sufficient, and the central proof chain is internally consistent.","rationale":"The reader's weakest_assumption correctly identifies H(A)<∞ as the operative constraint. After a careful pass through the construction, I find that this constraint is explicit, physically motivated (energy/heat-bath characterization), and does the required heavy lifting: it guarantees finite entropy of the rare-block marginal, ensures the blockwise cutoff errors vanish, and makes the conditional mutual informations finite. The central theorem's proof chain from spectral cutoffs to high-probability projections to Haar-random decoupling on the common subsystem is internally coherent, and the converse is conservative rather than circular. The paper itself flags the two places where matching converses are absent (error exponent for infinite-dimensional references, state-merging converse), so the ACCEPT verdict is not overstated relative to the stated claims. The one concern I examined most closely was the rate matching in Theorem 17 when 1/2 I(A:R) is strictly below H(A): the diagonal choice q_j = I/2 + 4ε_j, the lower bound (875), and the interval-wise convergence to q_j all check out algebraically. I also checked the δ=0 special case, the finite-dimensional maximally mixed case, and the pure-state parameterization; each is handled consistently. Thus no significant objection is warranted.","tokens_in":78763,"tokens_out":17184,"duration_ms":167419,"concrete_test":"As a sanity check, instantiate Proposition 14 with an explicit infinite-dimensional diagonal state ρA on a single bosonic mode with eigenvalues p_i ∝ i^{-2} (so H(A)<∞), a reference R correlated through a finite isometry, and q = H(A)/2. Verify numerically/analytically that N_m^{1/m} → e^{H(A)}, |M_m|^{1/m} → e^q, δ_{Π_m} → 0, and then check that the final discarded rate in Theorem 17 is (1/2)I(A:R). This tests the rate bookkeeping without relying on the full technical machinery.","verdict_should_be":"UNCHANGED","load_bearing_attack":"I find no load-bearing flaw in the central claim. The weakest premise is the finite-entropy condition H(A)_ρ < ∞ introduced at Eq. (369). This is a genuine constraint, but it is stated clearly, used exactly where needed (spectral cutoffs, weak typicality of rare blocks, well-defined conditional entropies), and the paper never claims the rates hold without it. The main construction in Theorem 15 is intricate but logically coherent: fixed common blocks give a finite-dimensional subsystem for Haar-randomization, weakly typical rare blocks are projected with finite rank, residual copies contribute subexponentially, and the block length is sent to infinity after n → ∞. The converse in Theorem 19 is stronger than needed and does not assume the projections commute with the input. Missing converses for the error exponent and for state-merging cost are explicitly acknowledged in Remarks 13 and 21, so they are not hidden gaps. The one-shot analysis also appears to repair a real issue in Ref. [19] (Remark 7) without introducing a circular step. The extra C_|A| factor is shown to be O(1) after tensor powers, and the rate identities in Theorem 17 and Theorem 20 follow algebraically from the constructed dimensions. I therefore do not see a reason to downgrade the reader's ACCEPT verdict.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops a decoupling framework for separable, possibly infinite-dimensional quantum systems. For finite-dimensional input A and arbitrary separable reference/output systems R,E, it proves one-shot relative-entropy decoupling bounds (Theorem 10, Corollary 12) in terms of sandwiched Rényi conditional entropies. For infinite-dimensional IID inputs, it assumes only H(A)_ρ < ∞ and constructs finite-rank projections whose success probability tends to one while restricting Haar-randomization to a finite-dimensional subsystem. This yields achievable IID partial-trace decoupling rates lim (1/n)log|A^n_Π| = H(A)_ρ and lim (1/n)log|M_n| = (1/2)I(A:R)_ρ (Theorem 17), with matching converse bounds for arbitrary projections and unitaries (Theorem 19). The paper then constructs an infinite-dimensional IID quantum state merging protocol with rates q = (1/2)I(A:R)_ψ and q−e = H(A|B)_ψ (Theorem 20). Missing converses for the infinite-dimensional error exponent and for state-merging cost are explicitly disclosed in Remarks 13 and 21.","tokens_in":79031,"tokens_out":5853,"duration_ms":67328,"significance":"If correct, this is a substantial advance: it replaces finite-dimensionality with the finite-entropy condition on the manipulated system as the operative hypothesis, and it extends decoupling and the fully quantum Slepian–Wolf protocol to arbitrary separable systems without imposing restrictions on the reference system. The proof chain is long and technically detailed, but it is also unusually explicit about its assumptions: the key finite-entropy condition is stated at Eq. (369), and the main limitations are flagged in Remarks 13 and 21 rather than hidden. The paper also identifies and repairs a genuine error in the prior one-shot decoupling analysis (Remark 7) and supplies a corrected Jensen-operator-inequality argument. The rate converse in Theorem 19 is stronger than strictly needed, since it does not assume the projections commute with the input marginal. The absence of a matching state-merging converse under only H(A)<∞ is acknowledged as an open problem, so the stated claims are appropriately scoped.","major_comments":[],"minor_comments":[{"comment":"The phrase 'in Theorem 15, we will control this contribution' reads as though Theorem 15 has not yet been proved; since Theorem 15 appears earlier, it should be 'Theorem 15 controls this contribution via Eqs. (657)–(658)' or similar.","section":"§IV.D, after Eq. (810)"},{"comment":"The expression 'optimal up to the known critical rate' is used in the abstract and again in Corollary 12/Remark 13. Since the critical rate is defined only through a comparison with Ref. [19], it would help the self-contained reader to state the threshold explicitly or to give the precise equation in Ref. [19] where it is defined.","section":"§I.B / Remark 13"},{"comment":"The counterexample to Eq. (273) depends on the generalized-logarithm convention log 0 = 0. This convention is stated earlier, but it would be useful to repeat it in the remark so the contradiction is immediately transparent to a reader skimming the discussion.","section":"§III.C, Remark 7"}],"recommendation":"accept","confidential_remarks":"This is a long but carefully written paper with a coherent proof chain. I found no load-bearing flaw; the main theorems are explicitly conditioned, and the most important limitations are openly discussed. The paper is a strong fit for the journal. The only caveat is length/density, but that is a feature of the subject matter, not a defect."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"You should know two things. First, this paper does what it claims: it takes one-shot relative-entropy decoupling beyond finite-dimensional references/outputs, and then pushes IID partial-trace decoupling and fully quantum state merging to separable infinite-dimensional inputs under only H(A) < ∞, recovering the known optimal rates. Second, I read through the central construction looking for a load-bearing flaw and did not find one. The common–rare pattern projection is intricate but internally consistent: fixed common blocks give the finite-dimensional Haar-randomization subsystem, rare blocks are weakly typical projected, residual copies are subexponential, and the n → ∞ then m → ∞ order is handled correctly. I agree with the reader's ACCEPT verdict at moderate confidence.\n\nWhat is genuinely new: the one-shot bound (Theorem 10/Corollary 12) for infinite-dimensional R and E, the IID achievability and converse (Theorems 17 and 19), and the state-merging protocol (Theorem 20) with q = (1/2)I(A:R) and q − e = H(A|B). The paper also earns credit for Remark 7, which points to a concrete error in Ref. [19]'s operator-concavity step and supplies a corrected argument via regularized logarithms and Jensen's trace inequality. The added C_|A| factor is shown to be O(1) after tensor powers, so the asymptotic exponent is unaffected.\n\nSoft spots, in proportion: the proof chain is long and depends on external operator-theoretic lemmas, with no machine-checked verification, so residual risk is real though not alarming. The finite-entropy assumption H(A) < ∞ at Eq. (369) is the operative restriction; it is clearly stated, physically motivated as an energy constraint, and never hidden. Missing converses for the error exponent and for state-merging cost are explicitly flagged in Remarks 13 and 21, and Remark 21's discussion of the ∞−∞ obstruction in converse arguments is honest and technically sensible. The equality-case handling in Theorem 17 (when (1/2)I(A:R) = H(A)) is a bit terse but valid. Self-citation appears only where the cited result is part of the same research program, not as padding.\n\nWho this is for: anyone working on continuous-variable or infinite-dimensional quantum information theory, decoupling-based coding theorems, or operational interpretations of entropic quantities. It deserves a serious referee. I would send it to peer review rather than desk reject, and I would cite it in my own work on infinite-dimensional entanglement and decoupling.","headline":"A serious, largely convincing extension of decoupling and state merging to separable infinite-dimensional systems under a finite-entropy condition; the long proof chain is coherent and the flagged gaps are real but honestly acknowledged.","tokens_in":79491,"tokens_out":977,"would_cite":true,"duration_ms":13816,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["81P45","81P68","94A17"],"pacs":["03.67.-a"],"model":"deepseek-v4-flash","headline":"Quantum information decoupling and state merging are established for arbitrary separable, possibly infinite-dimensional systems, with finite entropy replacing finite dimensionality as the operative condition.","keywords":["quantum decoupling","infinite-dimensional systems","separable Hilbert spaces","finite entropy","quantum state merging","sandwiched Rényi conditional entropy","mutual information","finite-rank projections"],"falsifier":"Try to construct a sequence of finite-rank projections Π_n on (ρ^A)^⊗n for a state with H(A)_ρ < ∞ such that δ_{Π_n} → 0 but liminf (1/n) log rank Π_n is strictly less than H(A)_ρ. The converse bound Proposition 18 forbids this; exhibiting such a sequence would disprove the claimed optimality of the projected-input dimension rate.","tokens_in":78632,"feed_emoji":"⚛️","tokens_out":6683,"duration_ms":70751,"temperature":0.7,"pith_summary":"The paper aims to remove finite-dimensional assumptions from quantum decoupling, a core primitive underlying communication, error correction, and information recovery. It shows that Haar-random-unitary decoupling can be made to work on infinite-dimensional systems by first projecting the many-copy input onto a finite-dimensional subspace with near-unit success probability, and by routing all auxiliary components into the discarded system at negligible cost. Under the assumption that the manipulated system has finite quantum entropy, the resulting protocol achieves the same optimal first-order rates as in finite dimensions: the projected-input dimension grows like exp(n H(A)) and the discarded dimension like exp(n I(A:R)/2). As an application, it constructs an infinite-dimensional quantum-state-merging protocol that recovers the finite-dimensional rates for quantum communication and total cost. If correct, this identifies finite entropy, not finite dimensionality, as the condition that makes these operational interpretations universal.","feed_headline":"Infinite-dimensional quantum decoupling works under finite entropy","feed_subtitle":"A protocol achieves optimal first-order rates for separable systems, with no dimension assumption on the reference.","key_machinery":"The load-bearing construction is the high-probability finite-rank projection on n-copy IID states. Fixing a block length, the paper cuts each block to a finite-dimensional 'common' eigenspace, then, instead of requiring every block to pass, retains all weakly typical common–rare patterns: about the right fraction of blocks is common and the rare blocks are themselves compressed by weak typicality into a finite-rank subspace. A fixed number of common blocks is extracted from every pattern, producing one common, pattern-independent finite-dimensional subsystem on which the same Haar-random unitary can act; all remaining blocks, rare components, and residual copies are placed in the discarded s","core_discovery":"The central discovery is a construction that lifts Haar-random-unitary decoupling to infinite-dimensional input systems despite the absence of a normalized Haar measure on the infinite unitary group. The paper builds, for every copy number n, a finite-rank projection on the n-copy input space whose success probability tends to one, isolating a finite-dimensional common subsystem on which the same random unitary acts while absorbing all other components into the discarded system. For states with H(A)_ρ < ∞, this yields an infinite-dimensional IID partial-trace decoupling protocol with projected-input dimension rate H(A)_ρ and discarded-system dimension rate 1/2 I(A:R)_ρ, and matching converse","pith_inferences":["The same high-probability projection technique may plausibly extend other finite-dimensional quantum information protocols to continuous-variable settings, with finite energy or finite entropy replacing dimension assumptions.","If mother-protocol resource reductions generalize, the state-merging result could yield infinite-dimensional versions of channel coding, channel simulation, and entanglement-assisted communication at the same first-order rates.","A practical test would be to implement the spectral-cutoff construction on a two-mode squeezed or displaced thermal state and check numerically that the dimension exponents approach H(A) and I(A:R)/2 as the copy number grows.","The repair of the operator-concavity step suggests that any finite-dimensional decoupling derivation relying on that step should be reexamined, although the repaired bound differs only by an asymptotically negligible factor in the IID regime."],"forward_implications":["Infinite-dimensional continuous-variable systems, such as bosonic modes, can be decoupled at the same first-order rates as finite-dimensional systems, provided the manipulated system has finite entropy.","Infinite-dimensional IID quantum state merging, a mother protocol for distributed quantum information processing, achieves the same communication and total-cost rates as in finite dimensions: q = I(A:R)/2 and q−e = H(A|B).","The one-shot relative-entropy decoupling bounds hold when the reference and output systems are arbitrary separable, possibly infinite-dimensional, not only when they are finite-dimensional.","For partial-trace decoupling with a finite-dimensional reference, the achievable asymptotic error exponent is optimal up to the critical rate identified in the paper.","The converse bounds imply that no protocol within the stated formulation can beat the rates H(A)_ρ and I(A:R)_ρ/2, so the achievability result is tight at first order."],"fun_headline_variants":["Quantum decoupling extended to infinite dimensions","Infinite-dim decoupling works under finite entropy","Decoupling without Haar measure: optimal rates","Universal quantum info beyond finite dimensions","Infinite-dim quantum state merging via decoupling"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The load-bearing assumption is that the manipulated system has finite von Neumann entropy, H(A)_ρ < ∞; the paper does not claim its rates hold without this condition.","fun_headline_variants_meta":{"raw":{"variants":["Quantum decoupling extended to infinite dimensions","Infinite-dim decoupling works under finite entropy","Decoupling without Haar measure: optimal rates","Universal quantum info beyond finite dimensions","Infinite-dim quantum state merging via decoupling"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000225,"raw_usage":{"total_tokens":1321,"prompt_tokens":786,"completion_tokens":535,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":530,"completion_tokens_details":{"reasoning_tokens":480}},"tokens_in":530,"tokens_out":535,"duration_ms":5586,"temperature":1.0,"reasoning_tokens":480,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-01T00:41:35.629095+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Try to construct a sequence of finite-rank projections Π_n on (ρ^A)^⊗n for a state with H(A)_ρ < ∞ such that δ_{Π_n} → 0 but liminf (1/n) log rank Π_n is strictly less than H(A)_ρ. The converse bound Proposition 18 forbids this; exhibiting such a sequence would disprove the claimed optimality of the projected-input dimension rate.","supporting_citations":[],"review_version":1}