{"id":"b31913fd-40a3-4e70-ba35-9140d796d266","arxiv_id":"2608.08836","paper_version":1,"verdict":"ACCEPT","confidence":"HIGH","novelty_score":8.0,"correctness_risk":"low","formal_verification":"none","parameter_count":1,"one_line_summary":"A one-copy-undistillable NPT state in the canonical DiVincenzo family is two-copy distillable for every local dimension d>=3, disproving the conjecture that the entire one-copy-undistillable region stays undistillable.","lead":"The authors prove that a particular entangled quantum state, which cannot be distilled from a single copy, becomes distillable when two copies are available, and this happens in every dimension from three upward. The result overturns a 25-year-old conjecture in the standard test family for a central open problem in quantum entanglement theory.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No significant objection identified: the two-copy certificate is self-contained and exact, and the one-copy half's imported G-undistillability is sound.","rationale":"The reader accepted with high confidence and flagged the external one-copy result as the weakest assumption. I agree that this is the only non-self-contained step, but it is not a live risk: Eq. (12) is algebraically correct with the canonical-plane parametrization used in the paper, and the one-copy undistillability of G is standard and independently checkable. I also spot-checked Lemma 1's closed forms, Eqs. (24)-(27), against the explicit d=3 qutrit coefficients; they reproduce Eq. (29) = -1/882. The three-copy and neighborhood claims are explicitly labeled inner bounds, so their witness dependence is not a correctness issue. No circularity, no fitted parameters in the central result, and no inconsistency between the abstract, the theorem, and the appendices. Therefore I see no basis to change the reader's accept.","tokens_in":16246,"tokens_out":18197,"duration_ms":174860,"concrete_test":"Independently verify Eq. (12) for symbolic d and confirm that G's partial transpose 2I-Omega is nonnegative on all matrices of rank at most two via |tr M|^2 <= 2||M||_F^2; if this holds, the one-copy dependency is sound.","verdict_should_be":"UNCHANGED","load_bearing_attack":"After checking the proof chain, I find no load-bearing flaw. The strongest claim is Theorem 1, whose riskier half is one-copy undistillability of rho_{C,d}. That half is inherited from the published one-copy undistillability of the Werner point G through Eq. (12). I verified Eq. (12) symbolically with the correct parametrization of the canonical plane (coordinates in Eq. (4) are (b,c), giving a=1/[d(2d-1)] at G): inserting a=1/[d(2d-1)], b=3/[d(2d-1)], c=1/[d(2d-1)] into rho^Gamma = a*Delta + (b/2)(I-Omega) + (c/2)(I+Omega-2*Delta) reproduces exactly the right-hand side of Eq. (12). The added term sum_{i<j} Pi_{ij} is positive semidefinite, so C is one-copy undistillable iff G is; G's one-copy undistillability is standard and for d>=3 follows from |tr M|^2 <= 2||M||_F^2 for rank-<=2 matrices applied to rho^Gamma_G proportional to 2I-Omega. The two-copy half is fully self-contained: Lemma 1 is proved from the Gram operator in Appendix A and yields Eqs. (28)-(29), a closed negative expectation for every d>=3. No hidden parameter, circular step, or mis-scaling appears. The paper itself correctly labels the lobe and three-copy claims as witness-dependent inner bounds.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the canonical two-parameter family of negative-partial-transpose (NPT) states introduced by DiVincenzo et al. and proves that, for every local dimension d ≥ 3, the distinguished state ρ_{C,d} at point C is one-copy undistillable but two-copy distillable. The two-copy certificate is an explicit Schmidt-rank-two vector |Ψ_d⟩ built from an equal-norm tight frame; Lemma 1, proved in Appendix A, gives its expectation values, and Eqs. (28)–(29) combine them into the negative value −(d−2)/(d^2(d−1)(3d−2)^2). The same witness is then used to certify an open region of two-copy-distillable states around C, and separate constructions provide three-copy witnesses that enlarge the certified region in some directions. The paper also contrasts this with the recently established two-copy undistillability of the neighboring triangle BGK, which follows from Werner-state results together with the DiVincenzo et al. propagation lemma.","tokens_in":16523,"tokens_out":20803,"duration_ms":211421,"significance":"If it holds, the result is significant: it disproves the conjecture that the entire one-copy-undistillable region BCGK of the canonical family remains undistillable for arbitrarily many copies, and it does so uniformly in every local dimension. The main certificate is transparent and checkable: the Schmidt-rank-two vector is explicit, the expectation values in Lemma 1 are derived from a Gram-operator calculation in Appendix A, and no parameters are fitted to force negativity. The one-copy half of Theorem 1 is imported from the standard DiVincenzo et al. result through the decomposition in Eq. (12), which the paper states clearly; this is an external but well-established input. The paper is also careful to label the two-copy lobe and the three-copy regions as witness-dependent inner bounds, not as optimized distillability boundaries. The contrast with the rigorously two-copy-undistillable triangle BGK makes the separation within one symmetry-reduced family concrete and compelling. The contribution is a strong, explicit step in a long-standing open problem.","major_comments":[],"minor_comments":[{"comment":"The decomposition in Eq. (12) is the bridge that imports one-copy undistillability from point G; a one-line verification of the coefficients, or an explicit pointer to the corresponding derivation in Ref. [8], would make the paper more self-contained.","section":"§II.C, Eq. (12)"},{"comment":"The phrase \"open two-copy-distillable neighborhood\" around C should specify \"open in the physical parameter region,\" because C lies on the boundary c = 0 of the physical triangle and the certified lobe is not open in the full (b,c) plane.","section":"§III.B and Abstract"},{"comment":"In the sentence \"Every two-copy certificate also yields a three-copy certificate by tensoring the corresponding Schmidt-rank-two witness with a suitable product vector on the third copy,\" the word \"suitable\" should be made explicit: the product vector must be chosen so that its one-copy expectation under ρ^Γ is positive, otherwise tensoring two negative expectations would not give a negative three-copy expectation.","section":"§III.C"},{"comment":"The caption states that the panels use affine schematic coordinates and are not on a common physical scale; adding the precise defining equations for the displayed regions (e.g., Eqs. (C4), (D21), and (D26)) would help readers connect the schematic to the analytic results.","section":"Figure 2"}],"recommendation":"minor_revision","confidential_remarks":"The central two-copy certificate is self-contained and correct, and I see no load-bearing flaw. The contrast section relies on several recent arXiv preprints for Werner-state two-copy undistillability [16–19]; these do not affect the main theorem, but the editor may wish to confirm their status before publication. The AI-use disclosure is unusual but transparent and does not affect the scientific assessment."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nThe headline result is exactly what the title says: the paper constructs, for every d>=3, a canonical NPT state rho_{C,d} that is one-copy undistillable but two-copy distillable. The witness is a single Schmidt-rank-two vector built from an equal-norm tight frame, and the negative expectation is closed-form: -(d-2)/(d^2(d-1)(3d-2)^2). That kills the old conjecture that the whole BCGK region is undistillable for any finite number of copies. I checked the calculation roughly and it is right.\n\nWhat is genuinely new: previous work knew the Werner line, but not the inside of the canonical family. The construction is uniform, not a per-dimension accident, and the proof is self-contained for the two-copy half. Lemma 1 is proved from the Gram operator in an appendix, and the coefficient-matrix viewpoint is a nice touch. The paper is also honest about what is not proven: the one-copy undistillability of C is imported from DiVincenzo et al.'s decomposition and the known one-copy-undistillability of the Werner point G. That is a black box, but a well-tested one; the decomposition is quoted correctly and the positivity argument is standard. The three-copy witnesses and the lobe regions are clearly labeled as witness-dependent inner bounds, not boundaries. The figures are schematic. The AI-use disclosure is unusually transparent, and the authors say they independently verified all calculations; nothing I saw contradicts that.\n\nSoft spots, in proportion: only one, and it is minor. The one-copy half rests on an external result without re-derivation. If that prior result had a subtle flaw, the advertised one-copy-undistillable status would fall. I think that risk is small, and it is the kind of thing a referee can check quickly. The three-copy section is dense and less central; I would not hold the paper to it for acceptance. There is also a slight overstatement in the discussion: disproving the conjecture that the entire region remains undistillable is true, but the phrase 'the conjecture' might suggest the general NPT problem; the abstract is careful, the discussion is fine.\n\nBottom line: this is solid work, worthy of a serious referee. I would cite it, and I would bring it to reading group. Publish it.","headline":"A clean, explicit counterexample to the DiVincenzo et al. finite-copy conjecture: point C is one-copy undistillable but two-copy distillable in every dimension, with a uniform tight-frame witness.","tokens_in":17104,"tokens_out":1511,"would_cite":true,"duration_ms":15275,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Inside a canonical family of entangled states, a state that cannot be distilled from one copy is shown to be distillable from two copies in every local dimension d≥3.","keywords":["entanglement distillation","negative partial transpose","two-copy distillability","one-copy undistillability","Schmidt-rank-two witness","equal-norm tight frame","canonical NPT family","Werner states"],"falsifier":"At $d=3$, Eq. (29) gives $\\langle\\Psi_3|(\\rho^\\Gamma_{C,3})^{\\otimes 2}|\\Psi_3\\rangle=-1/882$ for the explicit qutrit vector in Eq. (B16); evaluating this expectation by exact arithmetic settles the theorem, and any value other than $-1/882$ would show the construction or the partial-transpose formula is wrong.","tokens_in":16010,"feed_emoji":"⚛️","tokens_out":8555,"duration_ms":83325,"temperature":0.7,"pith_summary":"Entanglement distillation asks how many noisy copies of a quantum state are needed to extract pure entanglement. A long-standing open question is whether a state with a negative partial transpose (NPT) that cannot be distilled from one copy becomes distillable from finitely many copies. This paper answers the question within the canonical two-parameter family introduced as a symmetry-reduced testbed for exactly this problem: a distinguished state in its one-copy-undistillable region is already two-copy distillable in every local dimension d≥3. The authors give an explicit Schmidt-rank-two two-copy witness, built uniformly for all dimensions from an equal-norm tight frame, with a closed-form negative expectation value. Because the canonical reduction is many-to-one, the result extends to every NPT state that stochastic local operations can map to this distinguished point and disproves the conjecture that the whole one-copy-undistillable region remains undistillable for any number of copies.","feed_headline":"An NPT state that resists one-copy distillation succumbs at two copies","feed_subtitle":"A single equal-norm tight-frame witness certifies the counterexample in every local dimension.","key_machinery":"The load-bearing object is a uniform equal-norm tight-frame construction of the two-copy witness $|\\Psi_d\\rangle$. Two vectors $|x_1\\rangle,|x_2\\rangle$ spanning the $(d-1)$-dimensional subspace orthogonal to $|0\\rangle$ are chosen so that $\\langle x_r|x_s\\rangle=\\tfrac12\\delta_{rs}$ and each coordinate has total weight $1/(d-1)$; an equally spaced real frame realises this for every $n=d-1\\ge 2$. On each local two-copy space the construction takes symmetric modes $|u_r\\rangle=|0\\rangle|x_r\\rangle+|x_r\\rangle|0\\rangle$ and antisymmetric modes $|v_r\\rangle=|0\\rangle|x_r\\rangle-|x_r\\rangle|0\\rangle$, and pairs them as $|\\Psi_d\\rangle=(|u_1\\rangle|v_1\\rangle+|u_2\\rangle|v_2\\rangle)/\\sqrt{2}$. The opposite copy parities make the coefficient matrix odd under simultaneous copy swap, forcing the total trace to vanish, while the equal-norm property minimizes the unavoidable positive diagonal contribution. The result is that the four expectation values of $\\Delta\\otimes\\Delta$, $\\Delta\\otimes(I-\\Omega)$, $(I-\\Omega)\\otimes\\Delta$, and $(I-\\Omega)\\otimes(I-\\Omega)$ take the fixed values $1/(d-1)$, $-1/4$, $-1/4$, and $-1/2$, giving the closed-form negative two-copy expectation at point C.","core_discovery":"The paper's central claim is Theorem 1: for every d≥3, the canonical state $\\rho_{C,d}$ at point C is one-copy undistillable but two-copy distillable. The one-copy part follows from a decomposition in which the partial transpose of $\\rho_{C,d}$ is a positive combination of the partial transpose of the one-copy-undistillable point G and positive semidefinite projectors. The two-copy part is certified directly: a single normalized vector $|\\Psi_d\\rangle$ of Schmidt rank exactly two across $(A_1A_2):(B_1B_2)$ satisfies $\\langle\\Psi_d|(\\rho^\\Gamma_{C,d})^{\\otimes 2}|\\Psi_d\\rangle = -\\frac{d-2}{d^2(d-1)(3d-2)^2} < 0$, which by the Schmidt-rank-two criterion proves two-copy distillability. The same witness also gives negative expectation on an open hyperbolic lobe around C, yielding explicit two-copy-distillable segments along CB and CG, and separate three-copy witnesses enlarge the certified region further. In the neighbouring triangle BGK, by contrast, recent results on Werner states combined with the propagation lemma certify two-copy undistillability, so the same one-copy-undistillable region contains states whose two-copy behaviour is provably opposite.","pith_inferences":["A testable extension: the parity-and-tight-frame mechanism may transfer to higher-copy settings, so comparable closed-form witnesses could exist for three or more copies at other stages of the canonical family.","Since the canonical reduction collapses many states to one representative, the one-copy/two-copy gap is likely to appear in a broad stochastic-LOCC equivalence class, not only at the symmetric representative.","Optimising the Schmidt-rank-two vector in the unresolved part of BCG is a concrete next step: if the negative expectation reaches the boundary BG, it would show the certified two-copy lobe connects all the way to the two-copy-undistillable triangle."],"forward_implications":["Point C becomes a certified counterexample in every dimension $d\\ge 3$ to the conjecture that the whole region BCGK remains undistillable.","Every NPT state that stochastic local operations can map to $\\rho_{C,d}$ inherits two-copy distillability, since the canonical reduction is many-to-one.","The fixed two-copy witness certifies an open lobe around C in the canonical parameter plane, including explicit finite segments along CB and CG with closed-form cutoff parameters.","Separately constructed three-copy witnesses improve the certified region for $d=3$ and $d\\ge 4$, giving analytic inner bounds on the distillable set.","The triangle BGK is two-copy undistillable, so the canonical family contains rigorously certified states of both two-copy behaviours separated by an unresolved region."],"supporting_citations":[{"why":"Defines the canonical two-parameter NPT family, proves that the quadrilateral BCGK is one-copy undistillable, supplies the propagation lemma, and states the conjecture the paper targets.","marker":"[8]"},{"why":"Establishes that Werner states are two-copy distillable exactly when they are one-copy distillable, which certifies point G and the segment GH as two-copy undistillable.","marker":"[16–19]"},{"why":"Defines the Werner family and its PPT boundary, providing the benchmark line FH and the location of point G.","marker":"[23]"},{"why":"Supplies the Schmidt-rank-two criterion for k-copy distillability, the criterion used to certify two-copy distillability from the negative expectation value.","marker":"[2,3]"}],"fun_headline_variants":["Two copies unlock distillation for a stubborn NPT state in every dimension","One-copy-undistillable NPT state turns two-copy distillable for all d≥3","Even NPT states that resist one copy fall to two in every dimension","Double copy defeats one-copy distillation barrier for NPT states","Two copies overcome one-copy blockade for NPT states in all dimensions"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The one-copy-undistillable half of Theorem 1 is not re-derived in this paper; it uses the earlier result that point G is one-copy undistillable through the decomposition in Eq. (12), so if that external result or its constants were wrong, point C would lose its advertised status.","fun_headline_variants_meta":{"raw":{"variants":["Two copies unlock distillation for a stubborn NPT state in every dimension","One-copy-undistillable NPT state turns two-copy distillable for all d≥3","Even NPT states that resist one copy fall to two in every dimension","Double copy defeats one-copy distillation barrier for NPT states","Two copies overcome one-copy blockade for NPT states in all dimensions"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001165,"raw_usage":{"total_tokens":4885,"prompt_tokens":1073,"completion_tokens":3812,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":689,"completion_tokens_details":{"reasoning_tokens":3712}},"tokens_in":689,"tokens_out":3812,"duration_ms":26196,"temperature":1.0,"reasoning_tokens":3712,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T04:23:57.311403+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"At $d=3$, Eq. (29) gives $\\langle\\Psi_3|(\\rho^\\Gamma_{C,3})^{\\otimes 2}|\\Psi_3\\rangle=-1/882$ for the explicit qutrit vector in Eq. (B16); evaluating this expectation by exact arithmetic settles the theorem, and any value other than $-1/882$ would show the construction or the partial-transpose formula is wrong.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Defines the canonical two-parameter NPT family, proves that the quadrilateral BCGK is one-copy undistillable, supplies the propagation lemma, and states the conjecture the paper targets."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Defines the Werner family and its PPT boundary, providing the benchmark line FH and the location of point G."}],"review_version":1}