{"id":"933a298e-ba81-4ef3-96e4-8cd08d705ce2","arxiv_id":"2607.27195","paper_version":1,"verdict":"ACCEPT","confidence":"HIGH","novelty_score":7.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"Exponential strong-converse distillable entanglement is strictly smaller than exponential strong-converse cost for explicit mixed states under non-entangling and completely PPT-preserving operations.","lead":"Entanglement stays irreversible even when large fixed errors are allowed: distillation and formation rates stay strictly apart, and the gap is exponential in the number of copies. This settles a 2023 conjecture and shows the gap is not an artifact of demanding perfect asymptotic transformations.","discovery_kind":"extension","skeptic_critique":{"model":"grok-4.5","headline":"No significant objection identified","rationale":"The central claim is a strict exponential-strong-converse gap for concrete states under NE/KP and under C-PPT-P. Both halves are supported by explicit single-letter bounds whose proofs are written out. The only non-routine step is the witness construction of Lemma 3; it is lengthy but uses only elementary binomial tail bounds and Hölder, with all constants displayed. Because the Lami–Regula conjecture is already settled by the elementary operator inequality for ω_3 (and the matching one-shot preparation), a hypothetical failure of the general lemma would at worst narrow the SDP cost bound, not the headline separation. The antisymmetric family is independent and analytically clean. Reader’s ACCEPT / HIGH / low correctness risk is therefore unchanged; the residual algebra check is worth doing but does not move the verdict.","tokens_in":21482,"tokens_out":674,"duration_ms":14248,"concrete_test":"Independently re-derive the operator lower bound (74) of Lemma 3 for the concrete feasible pair (P=P_3-Φ_3, Q=-Φ_3/2) of ω_3: fix c=1.1, choose η≤1/22 small enough that the eta-correction is <δ=0.01, and verify numerically for n=20…30 that (1/n)log L_n ≥ 1-δ and (1/n)log c_n ≤ -κ<0; if either scaling fails the general cost bound needs repair (though the ω_3 claim itself remains intact via the direct argument).","verdict_should_be":"UNCHANGED","load_bearing_attack":"The reader's weakest-assumption flag on Lemma 3 (truncated tensor-product witnesses W_n, Chernoff/KL control of off-support coefficients, and the η\to0 limit that recovers −log∥(P+Q)^Γ∥_∞−δ) is the right technical locus, but on a careful re-read the construction appears internally consistent: feasible Q with ∥Q∥_∞≤1/2 and ∥(P+Q)^Γ∥_∞=α<1, binomial truncation just above the mean, dual bounds via x=1 and the KL-tuned x, and the resulting uniform exponential decay of c_n are standard and explicitly parameterized. For the headline state ω_3 the bound is saturated by an elementary choice Q=-Φ_3/2 giving α=1/2, so the general lemma is not even required for the conjecture resolution. The independent C-PPT-P half (R_max vs additive E^{1/2}_{N,PPT_2} on antisymmetric eta_P̃) is short, SDP-exact, and does not rely on Lemma 3. No load-bearing gap that would overturn E^{exp(†)}_{d,F}<E^{exp(†)}_{c,F} was found.","agreement_with_reader":"agree"},"referee_report":{"model":"grok-4.5","summary":"The manuscript establishes exponential strong-converse irreversibility of mixed-state entanglement. For the Lami–Regula state ω₃ it proves E^{exp(†)}_{d,KP}(ω₃)=log₂(3/2)<1=E^{exp(†)}_{c,KP}(ω₃) under non-entangling and PPT-preserving maps, so any protocol beyond either threshold has fidelity decaying exponentially in the number of copies. This resolves and strengthens the finite-error conjecture of Lami and Regula. The formation lower bound is obtained from a single-copy SDP quantity (kernel-robust tempered negativity / α⋆) via an explicit truncated-tensor witness construction (Lemma 3). For completely PPT-preserving operations the authors give matching one-shot bounds in terms of R_max and the additive quantity E^{1/2}_{N,PPT₂}, and construct analytically solvable weighted antisymmetric states β_P̃ with R_max(β_P̃)=log₂(1+∥P̃∥₁)<1=E^{1/2}_{N,PPT₂}(β_P̃) whenever ∥P̃∥₁<1, yielding the same exponential separation (and, by sandwiching, under LOCC).","tokens_in":21721,"tokens_out":1030,"duration_ms":45936,"significance":"The result is a clear advance in the resource theory of entanglement. It closes a stated conjecture, upgrades irreversibility from vanishing and fixed error to the exponential strong-converse regime, and supplies efficiently computable single-copy certificates (SDP for NE/KP; closed-form R_max and E^{1/2}_{N,PPT₂} for an analytic family under C-PPT-P). The C-PPT-P families also give, to the authors’ knowledge, the first exponential strong-converse cost–distillable gap that applies even under LOCC. The contrast with reversible thermodynamic resource theories is sharpened in a technically controlled way. Explicit operator inequalities for ω₃, additivity of E^{1/2}_{N,PPT₂}, and concrete preparation protocols that saturate the cost side are genuine strengths.","major_comments":[],"minor_comments":[{"comment":"Abstract: the phrase “persists even at the level of polynomially (in the number of copies) growing error” is ambiguous relative to the exponential strong-converse theorems actually proved (fidelity ≤ 2^{-γn}). Prefer wording that matches the definitions (fixed error < 1, or exponential approach of the error to 1).","section":"Abstract"},{"comment":"Abstract and elsewhere: missing spaces in compounds such as “theexponentialstrong-converseirreversibility” and “exhibiting theexponentialstrong-converseirreversibility”; fix in copy-editing.","section":"Abstract"},{"comment":"Main text after Eq. (26): the elementary choice Q=−Φ₃/2 already saturates α⋆(ω₃)=1/2, so the general Lemma 3 machinery is not needed for the conjecture resolution. A one-sentence forward pointer that the ω₃ cost bound is elementary would help readers who only need the headline result.","section":"MAIN RESULTS"},{"comment":"Supplemental Material, Lemma 3: the free parameters c=1.1 and the restriction η≤1/22 are fixed for convenience. A brief remark that any c>1 with cη<1 and sufficiently small η works (with κ(c,η)>0) would make the robustness of the Chernoff/KL argument clearer.","section":"Supplemental Material, Lemma 3"},{"comment":"Definitions of E^{exp(†)}_{d/c,F}: the main text sketches the exponential rates; the precise inf/sup formulations appear only in the Supplement. Cross-referencing the Supplement definitions at first use in the main text would improve readability.","section":"DEFINITIONS AND BACKGROUND"},{"comment":"Eq. (19) and the distillation argument for ω₃ use the separable majorant σ₃=P₃/3 and Tr(Φ_M τ)≤1/M for τ∈K. This is standard; citing the precise Fuchs–van de Graaf form used for 1−ε_d^n≤√f_n would make the exponential decay step fully self-contained.","section":"MAIN RESULTS"}],"recommendation":"accept","confidential_remarks":"I agree with the external stress-test that Lemma 3 is the only delicate locus and that it appears internally consistent; the headline ω₃ separation does not even require the full lemma. The manuscript is appropriate in scope and novelty for a high-quality quant-ph theory venue. No citation or priority concerns stood out."},"author_rebuttal":null,"desk_editor":{"model":"grok-4.5","letter":"This paper actually closes the Lami–Regula conjecture and then strengthens it. For their two-qutrit state ω₃ they get E^{exp(†)}_{d,KP}=log₂(3/2)<1=E^{exp(†)}_{c,KP}, so any protocol past those rates has fidelity decaying exponentially in n. That is the real news: irreversibility survives fixed error and even exponentially vanishing fidelity, under the maximal non-entangling class.\n\nWhat they do well is concrete. The ω₃ cost lower bound is elementary (support projector plus Q=−Φ₃/2 gives α=1/2, saturated by the known one-ebit preparation). Distillation uses a simple majorization by a separable state. The new SDP quantity N_⊥ (kernel-robust tempered negativity) is a usable one-shot lower bound on the exponential strong-converse cost for rank-deficient states. On the C-PPT-P side they give short, additive witnesses: R_max(β_P̃)=log₂(1+∥P̃∥₁) versus E^{1/2}_{N,PPT₂}=1 for the antisymmetric mixtures, with an explicit gap whenever ∥P̃∥₁<1. Those families are analytically clean and already contain known rank-two examples. Citations are appropriate; circularity is low because upper and lower bounds come from independent protocols and operator inequalities.\n\nThe soft spot is exactly where the reader flagged: Lemma 3, the truncated tensor-product witnesses with Chernoff/KL control. It is long and the η→0 limit needs care. On a re-read it looks internally consistent and standard, and—importantly—the headline ω₃ result does not need the general lemma at all. The C-PPT-P half is independent of it. Residual algebra risk is real but not load-bearing for the main claim. LLM assistance on proof details is disclosed; treat it as transparency, not a red flag.\n\nThis is for people who work on entanglement resource theories, strong converses, or the thermodynamics analogy. A serious referee should see it. I would cite the ω₃ separation and the antisymmetric families. Send it to peer review.","headline":"Resolves Lami–Regula’s finite-error cost conjecture with an exponential strong-converse gap under NE, plus clean analytic C-PPT-P families; the math holds up.","tokens_in":22467,"tokens_out":539,"would_cite":true,"duration_ms":10665,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["03.67.Mn","03.65.Ud","03.67.-a"],"model":"grok-4.5","headline":"Entanglement stays irreversible even when large fixed error is allowed: formation and distillation thresholds separate exponentially.","keywords":["quantum entanglement","irreversibility","strong converse","entanglement cost","distillable entanglement","non-entangling operations","PPT-preserving maps","semidefinite programming"],"falsifier":"Exhibit a sequence of non-entangling formation protocols for ω₃ whose rate stays strictly below 1 yet whose output fidelity with ω₃⊗n does not decay exponentially (or stays bounded away from zero); or show that every feasible SDP witness Q for N⊥(ω₃) yields ∥(P+Q)Γ∥∞≥1.","tokens_in":22300,"feed_emoji":"ₚ","tokens_out":1003,"duration_ms":18109,"temperature":0.7,"pith_summary":"Some mixed entangled states cost pure entanglement to make, yet yield none back under local operations. This paper shows the gap survives even if one allows a fixed nonzero error that never goes to zero, and in fact any protocol that tries to beat the gap must fail with error that races exponentially to one as more copies are used. The authors settle a conjecture for a standard two-qutrit example under the largest class of operations that never create entanglement, prove an efficiently computable lower bound on the exponential formation cost, and give analytically solvable antisymmetric families under completely PPT-preserving maps that exhibit the same exponential separation. A sympathetic reader cares because the result draws a sharp line between entanglement and thermodynamics: allowing imperfect conversion does not restore a second-law-like reversibility.","feed_headline":"Entanglement stays irreversible even with large fixed error","feed_subtitle":"Formation and distillation thresholds separate; beating them fails exponentially fast in copy number","key_machinery":"A single-copy SDP quantity called the kernel-robust tempered negativity N⊥(ρ) (infimum of ∥(P+Q)Γ∥∞ over Hermitian Q supported on the kernel of ρ with ∥Q∥∞≤1/2). When N⊥(ρ)<1 it lower-bounds the exponential strong-converse entanglement cost by −log N⊥(ρ), via an asymptotic family of truncated tensor-product witnesses whose off-support mass decays exponentially by Chernoff bounds.","core_discovery":"There exist mixed entangled states, including the two-qutrit state ω₃ and analytic weighted antisymmetric families, for which the exponential strong-converse distillable entanglement is strictly smaller than the exponential strong-converse entanglement cost, both under non-entangling (and PPT-preserving) maps and under completely PPT-preserving maps. Explicitly, for ω₃ the two rates equal log₂(3/2) and 1; any protocol operating beyond either threshold has fidelity decaying exponentially in the number of copies.","pith_inferences":["The same witness technique may lift to other resource theories whose free sets admit a dual robustness formulation controlled by partial transpose or a similar cone.","Because the C-PPT-P separation already sits inside LOCC, an explicit LOCC protocol family matching the analytic rates would close the last remaining gap mentioned by the authors.","Polynomial (rather than exponential) error growth is ruled out as a route to approximate reversibility for these states; only error that stays bounded away from one could still conceivably close the gap."],"forward_implications":["Fixed nonzero error cannot restore asymptotic reversibility of mixed-state entanglement under non-entangling operations.","Any attempt to form or distill beyond the stated thresholds fails with error that approaches one exponentially fast in the number of copies.","The SDP quantity N⊥ supplies an efficiently computable certificate of exponential formation cost for rank-deficient states.","Under completely PPT-preserving maps, every weighted complete-bipartite antisymmetric state with trace-norm of the weight matrix strictly less than 1 is exponentially irreversible, with exact rates log₂(1+∥P̃∥₁) versus 1.","No comparable exponential separation was previously known even for the stricter class of LOCC."],"fun_headline_variants":["Entanglement irreversibility holds even for exponentially growing error","Distillable rate and cost separate exponentially for mixed entangled states","Strong-converse gap proves entanglement formation strictly irreversible","Polynomial error still cannot restore entanglement reversibility","Antisymmetric states exhibit exponential strong-converse irreversibility"],"cache_read_input_tokens":16512,"weakest_assumption_plain":"The formation lower bound stands or falls with a technical construction of truncated multi-copy witness operators that stay controlled under partial transpose while their weight outside the support of the state decays exponentially.","fun_headline_variants_meta":{"raw":{"variants":["Entanglement irreversibility holds even for exponentially growing error","Distillable rate and cost separate exponentially for mixed entangled states","Strong-converse gap proves entanglement formation strictly irreversible","Polynomial error still cannot restore entanglement reversibility","Antisymmetric states exhibit exponential strong-converse irreversibility"]},"model":"grok-4.5","effort":"low","cost_usd":0.005712,"raw_usage":{"total_tokens":1521,"prompt_tokens":799,"num_sources_used":0,"completion_tokens":79,"cost_in_usd_ticks":57124000,"prompt_tokens_details":{"text_tokens":799,"audio_tokens":0,"image_tokens":0,"cached_tokens":128},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":643,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":799,"tokens_out":79,"duration_ms":11450,"temperature":1.0,"reasoning_tokens":643,"cache_read_input_tokens":128,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-07-30T13:47:10.166281+00:00","model_set":{"reader":"grok-4.5"},"falsifier":"Exhibit a sequence of non-entangling formation protocols for ω₃ whose rate stays strictly below 1 yet whose output fidelity with ω₃⊗n does not decay exponentially (or stays bounded away from zero); or show that every feasible SDP witness Q for N⊥(ω₃) yields ∥(P+Q)Γ∥∞≥1.","supporting_citations":[],"review_version":1}