{"id":"ddcc95f9-2c79-4e7c-8f0d-94031d701448","arxiv_id":"2608.03390","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"Absolute PPT spectra form a spectrahedron with exposed faces, a kernel-rank facial calculus, and an inscribed polytope that gives tight bounds on purity, entropy, and relative volume.","lead":"This paper maps the geometry of the spectra of bipartite quantum states that stay PPT or separable under every global rotation. It gives exact convex-geometric descriptions plus bounds on purity, entropy, and volume; a key abstract claim about exponential volume decay is only numerically supported.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 4.1's reduction from global unitaries to permutation-indexed LMIs is the load-bearing external step, cited from [16, Lemma 3] rather than proved.","rationale":"I read the paper in good faith and checked the main geometric chain. The spectrahedral description in Eqs. (7)–(8) is internally consistent once Theorem 4.1 is granted: a finite block-diagonal LMI over the simplex is a spectrahedron, and the subsequent face, boundary, and extreme-point analyses follow standard spectrahedral facts. I found no internal contradiction in the facial-dimension arguments, the rank lower bound, or the boundary hypersurface count. The weakest point is indeed Theorem 4.1's reliance on [16, Lemma 3] for the equivalence between all global unitaries and all permutations. This is a published external result, and the equivalence is very plausibly true, but the paper does not reproduce the step, and the sign-dependent vector μ(v) makes the reduction subtle. The reader's conditional verdict is appropriate: the main geometric results should be accepted only once this external premise is either explicitly reproved or independently verified. The volume-decay overstatement in the abstract is a separate weakness but is not load-bearing for the central spectrahedron claim; it would affect the quantitative section only.","tokens_in":45800,"tokens_out":25192,"duration_ms":322711,"concrete_test":"Independently prove the missing equivalence: for fixed ψ with Schmidt coefficients x, show from von Neumann's trace inequality that Tr(UΛU† (ψψ)^Γ) ≥ 0 for all U is equivalent to dot(λ_π, α(x)) ≥ 0 for all π, where α(x) is the eigenvalue vector (x_i^2, ±x_ix_j, 0). Then show that for any v with |v_i| = x_i, the set {dot(λ_π, μ(v)) : π ∈ S_mn} equals {dot(λ_π, α(x)) : π ∈ S_mn} because μ(v) is a fixed permutation of α(x). If this cannot be shown, run an exhaustive check for m=3, n=3: enumerate all 720 permutations and search for a probability vector λ with all L_π(λ) ≥ 0 that violates Hildebrand's original ordered criterion, or vice versa. A single mismatch refutes Theorem 4.1.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim that APPT_{m,n} is a spectrahedron rests entirely on Theorem 4.1. The critical step is the assertion that, for a fixed Schmidt vector x of |ψ⟩, the condition Tr(U diag(λ)U† (|ψ⟩⟨ψ|)^Γ) ≥ 0 for all U is equivalent to Σ_k λ_{π(k)} μ_k(v) ≥ 0 for every permutation π and every v with |v_i| = x_i. This is the bridge from a continuum of unitary rotations to the finite family of LMIs L_π(λ) ≥ 0. The paper says only 'see the proof of [16, Lemma 3]' and does not reproduce the argument. If that lemma implicitly requires λ to be non-increasing, or if μ(v) must be the sorted eigenvalue vector of (|ψ⟩⟨ψ|)^Γ rather than an arbitrary sign-dependent rearrangement, then Eqs. (7)–(8) could over- or under-constrain APPT_{m,n}. Since every downstream geometric statement — spectrahedrality, exposed faces (Cor. 5.10), boundary hypersurfaces (Thm. 4.6), and the rank-dimension formula (Prop. 5.4) — inherits Theorem 4.1, this is the single most load-bearing premise. It is legitimate external support, but it is also the least locally verified part of the argument.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the convex geometry of the sets of spectra of bipartite absolutely separable (ASEP) and absolute-PPT (APPT) states. The central technical claim is a permutation-symmetric reformulation of Hildebrand's absolute PPT criterion: for m≤n, λ∈APPT_{m,n} iff L_π(λ)≥0 for all π∈S_{mn}, where L_π is an m×m matrix whose entries are linear forms in the permuted spectrum. This yields the advertised spectrahedral representation APPT_{m,n}=Δ_{mn-1}∩{⊕_π L_π(λ)≥0}. From it the paper derives a determinantal description of the boundary, a kernel-based description of all faces, a rank formula for face dimensions, and the conclusion that every face is exposed. In the quantitative part, an inscribed polytope P_{m,n} is defined via Gershgorin-type sufficient conditions; its maximal purity, minimal von Neumann entropy and relative volume are computed. The paper also gives a Monte-Carlo study suggesting that the relative volume of APPT_{m,n} decays exponentially in n with a rate close to that of the polytope, and states a conjecture that the maximal purity of APPT_{m,n} equals that of P_{m,n} for all m,n except m=n=2.","tokens_in":46131,"tokens_out":15736,"duration_ms":174371,"significance":"If the central equivalence is correct, the paper makes a substantial contribution: it gives a finite spectrahedral description of APPT_{m,n} for all m≤n, with all faces exposed, an explicit kernel-based characterization of faces and extreme points, and a dimension formula governed by the rank of a permutation-block matrix. The use of Gershgorin arguments to obtain a concrete inscribed polytope and the exact computation of its purity, entropy and volume are valuable and in part already go beyond previous work. The numerical fits for the volume decay are interesting and provide a concrete testable conjecture. However, the paper's headline volume statement in the abstract is stronger than what is rigorously proved, and the main spectrahedral representation relies on an external lemma whose proof is not reproduced. Both points need to be addressed before the paper can be accepted as is.","major_comments":[{"comment":"The step from the continuum of unitary constraints Tr(U diag(λ) U† (|ψ⟩⟨ψ|)^Γ)≥0 to the finite family L_π(λ)≥0 for all π∈S_mn is the load-bearing point of the whole paper, but the proof only says 'see the proof of [16, Lemma 3]' and does not reproduce it. Since this equivalence is what makes APPT_{m,n} a spectrahedron and is inherited by Theorem 4.3, Proposition 5.4, Corollaries 5.7, 5.10, and 5.12, the reader cannot verify that the LMI family is exactly the absolute PPT condition, especially without any ordering assumption on λ. Please either give a self-contained proof of the equivalence, or state precisely which statement of [16] is being used and why it applies to the unordered, permutation-symmetric setting.","section":"Theorem 4.1, Eqs. (7)-(8)"},{"comment":"The abstract states that the relative spectral volume of APPT_{m,n} decays exponentially in n 'by a constant multiplicative factor of the relative volume of the inscribed polytope'. The rigorous content of Section 6.3 is Theorem 6.16, which only gives the lower bound vol_rel(APPT_{m,n})≥max{vol_rel(BALL), vol_rel(P_{m,n})}. The exponential decay statement is supported by the numerical log-linear fits in Eq. (71) and Table 1, not by a proof. No upper bound of vol_rel(APPT) relative to vol_rel(P) is established. Please either prove the claimed asymptotic relation or weaken the abstract and Section 7 to state the lower bound and the numerical evidence.","section":"Abstract and Section 6.3"},{"comment":"The derivation of the exact relative volume of P_{m,n} applies Lasserre's formula (62) to coefficients h(u_i) that are not pairwise distinct. The text says that an 'identical weights condition' from [40, Sec. 2.1] applies, but the limiting expression is not stated. Since the formula (63) is used as a rigorous lower bound and feeds into the quantitative claims, the limiting argument should be written out explicitly, or the statement should be restricted to the cases where the formula can be obtained by a limiting argument from (62).","section":"Proposition 6.14"}],"minor_comments":[{"comment":"There is a typo 'and and identify' in the first paragraph of the introduction.","section":"Introduction, p. 2"},{"comment":"The caption reads 'and its the convex hull'; this should be 'and its convex hull'.","section":"Figure 2 caption"},{"comment":"The index functions p(i,j) and q(i,j) are central but somewhat opaque; a small example or one-line enumeration for m=2 would help the reader.","section":"Eq. (5)"},{"comment":"The fit log10 vol ≈ C_m + γ_m n is presented without error bars or number of Monte Carlo samples; adding these would make the numerical comparison more reliable.","section":"Section 6.3, Eq. (71)"}],"recommendation":"major_revision","confidential_remarks":"The main risk is the external dependence of Theorem 4.1 on [16, Lemma 3]. The referees should ask the authors to spell out this step or to state the precise external theorem used. The volume claim in the abstract is also stronger than the proven results and should be corrected. If the authors can supply the missing proof, the paper is likely a strong contribution."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear Colleague,\n\nThe paper is worth a look. Its main contribution is a permutation-symmetric LMI reformulation of the absolute PPT criterion, which turns APPT_{m,n} into a spectrahedron for all m≤n. That is a genuinely new structural handle, and the paper uses it well: the kernel-rank facial description (dimension of a face determined by rank of a block matrix C_U) is clean and new, as is the count of irreducible boundary hypersurfaces and the exact volume formula for the inscribed polytope. The proof that ASEP_{m,n} is semialgebraic is straightforward but correct.\n\nThe soft spots are real. Theorem 4.1, the load-bearing step that reduces a continuum of global unitaries to finitely many permutation-indexed LMIs, is not proved in the text. It says \"see [16, Lemma 3]\" and leaves it at that. Strictly speaking, the entire spectrahedral edifice inherits that external lemma. I believe Hildebrand's lemma is right, so I'm not worried about correctness, but a referee should ask for a self-contained statement, if not the full argument. Quote the lemma and explain why it applies to unordered spectra.\n\nSecond, the abstract's exponential volume decay claim is stronger than what is proven. The paper proves vol(APPT_{m,n}) ≥ vol(P_{m,n}), and P_{m,n} has exponential volume. That is a lower bound; it does not show that APPT_{m,n} itself decays exponentially in n, nor that its volume is within a constant factor of the polytope. The numerical fits are suggestive, but they are not proofs, and no code or data are included. The claim should be softened to \"lower bound plus numerical evidence\" in the abstract, and the numerics should ship.\n\nMinor things: the claimed equality of maximal purity between APPT and the polytope is honestly labeled a conjecture with numerical support, so that is fine.\n\nOverall, this is a solid paper for quantum information mathematicians studying absolute separability. It deserves a serious referee. My recommendation: send it to peer review, but ask the authors to make the external lemma explicit, temper the volume claim, and provide reproducible numerics. If they do that, I would cite it.","headline":"Solid new geometry for APPT spectra; the main LMI theorem rests on a cited lemma and the exponential volume claim outruns the proof.","tokens_in":46621,"tokens_out":3640,"would_cite":true,"duration_ms":42977,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["52A20","14P10","90C22","81P40"],"pacs":[],"model":"deepseek-v4-flash","headline":"The spectra of absolute-PPT bipartite states form a spectrahedron for every $m\\le n$, with all faces exposed and a rank formula for face dimensions.","keywords":["absolute separability","absolute PPT","spectrahedron","convex geometry of spectra","semialgebraic set","exposed faces","quantum entanglement","von Neumann entropy"],"falsifier":"A concrete check would be to take a small case such as $m=n=3$, generate spectra that satisfy the known absolute-PPT inequalities of [16], and test whether $L_\\pi(\\lambda)\\ge 0$ for every permutation; a single violation would refute Theorem 4.1. Independently, the face-dimension formula would fail if any proper face's constraint matrix $C$ had rank outside $\\{0\\}\\cup\\{m,\\ldots,mn-1\\}$, and Conjecture 6.7 could be tested by a global SDP or branch-and-bound optimization for $m=5$ searching for an APPT spectrum with purity higher than the polytope's closed-form maximum.","tokens_in":45697,"feed_emoji":"📐","tokens_out":9517,"duration_ms":110483,"temperature":0.7,"pith_summary":"Quantum states that keep the positive-partial-transpose property under every global unitary rotation are called absolute PPT, and their reachable eigenvalue distributions form a convex body $\\mathrm{APPT}_{m,n}$ inside the probability simplex. The paper's central achievement is a permutation-symmetric reformulation of the known absolute-PPT criterion that makes $\\mathrm{APPT}_{m,n}$ a spectrahedron: a spectrum belongs to the set exactly when every matrix $L_\\pi(\\lambda)$ in a permutation-indexed family is positive semidefinite. Because spectrahedra have only exposed faces, the paper obtains a complete face and extreme-point description of $\\mathrm{APPT}_{m,n}$, with face dimension $(mn-1)-\\mathrm{rank}(C)$ and maximal proper faces of dimension $mn-m-1$. It also proves that $\\mathrm{ASEP}_{m,n}$ is semialgebraic for all $m\\le n$ and that $\\mathrm{ASEP}_{2,n}$ is itself a spectrahedron, which matters because spectral geometry may help decide whether absolute separability coincides with absolute PPT. The paper adds an inscribed polytope $\\mathcal{P}_{m,n}$ giving rigorous purity, entropy, and volume bounds, and it explicitly notes that the converse of its extreme-point projection theorem is proven only for $m=2,3$.","feed_headline":"One matrix family rules all absolute-PPT spectra","feed_subtitle":"For every $m\\le n$, the spectra that survive all global unitaries form a spectrahedron with only exposed faces.","key_machinery":"The carrying object is the permutation-indexed block matrix $L(\\lambda)=\\bigoplus_{\\pi\\in\\widetilde{S}} L_\\pi(\\lambda)$, where $\\widetilde{S}$ is a reduced set of permutations of the $mn$ eigenvalues and each $L_\\pi(\\lambda)$ is the $m\\times m$ symmetric matrix defined by the index maps $p(i,j)$ and $q(i,j)$. These maps split the $m^2$ active spectral coordinates into diagonal and off-diagonal entries, and permutations acting only on the remaining $mn-m^2$ coordinates produce redundant blocks. This object converts the quantifier over all global unitaries in the definition of absolute PPT into a finite family of linear matrix inequalities, yields the topological boundary as a union of irreduc","core_discovery":"On its own terms, the paper establishes that for every $m\\le n$ the set $\\mathrm{APPT}_{m,n}$ of absolute-PPT spectra is a spectrahedron: $$\\mathrm{APPT}_{m,n}=\\Delta_{mn-1}\\cap $L^{{-1}}$\\left(\\bigoplus_{\\pi\\in\\widetilde{S}} S_m^+\\right),$$ where $L(\\lambda)$ is the block diagonal matrix formed by the $m\\times m$ matrices $L_\\pi(\\lambda)$. Each $L_\\pi(\\lambda)$ has diagonal entries $2\\lambda_{\\pi(p(i,i))}$ and off-diagonal entries $\\lambda_{\\pi(p(i,j))}-\\lambda_{\\pi(q(i,j))}$, with explicit index maps $p,q$ selecting $m^2$ of the $mn$ spectral coordinates. The paper then uses the kernel subspaces of these matrices to characterize faces: a face is determined by demanding $U_\\pi\\subseteq\\ker L_\\p","pith_inferences":["Beyond the paper, the spectrahedral description suggests that the maximum-purity conjecture could be attacked with symmetry-reduced semidefinite programming; the numerics reported stop at $m=4$, so a certified SDP computation for $m=5$ would be a natural next check.","Inference: the rank formula for face dimensions invites a purely combinatorial enumeration of faces by assigning a kernel subspace to each active permutation, which could produce the full face lattice of $\\mathrm{APPT}_{m,n}$ for small $m,n$ without sampling spectra.","Inference: if the polytope $\\mathcal{P}_{m,n}$ is eventually shown to lie inside $\\mathrm{ASEP}_{m,n}$—a containment the paper leaves open—then its exact purity, entropy, and volume bounds immediately become bounds on absolute separability, and the comparison of the two sets reduces to their boundaries outside the polytope.","Inference: the exposedness of every APPT face gives a concrete separation strategy: search for non-exposed extreme points of $\\mathrm{ASEP}_{m,n}$. The paper's explicit face equations provide a finite algebraic system on which such a search could be based."],"forward_implications":["If the spectrahedral description is correct, membership of a spectrum in $\\mathrm{APPT}_{m,n}$ is a semidefinite feasibility problem, so optimizing linear functionals over absolute-PPT spectra becomes a conic-optimization question.","Every face of $\\mathrm{APPT}_{m,n}$ is exposed; the dimension of a face is $(mn-1)-\\mathrm{rank}(C)$, extreme points are exactly the spectra with $\\mathrm{rank}(C)=mn-1$, and maximal proper faces have dimension $mn-m-1$.","For qubit-qudit systems the maximal-face equations become explicit linear relations, recovering and organizing known extreme-point structure for $\\mathrm{ASEP}_{2,n}$.","Since $\\mathrm{ASEP}_{m,n}$ is semialgebraic while $\\mathrm{APPT}_{m,n}$ is a spectrahedron, exhibiting any non-exposed extreme point of $\\mathrm{ASEP}_{m,n}$ would prove absolute separability is strictly smaller than absolute PPT.","The inscribed polytope $\\mathcal{P}_{m,n}$ supplies closed-form lower bounds on maximum purity, upper bounds on minimum von Neumann entropy, and an exact relative volume; numerical estimates indicate the relative volume of $\\mathrm{APPT}_{m,n}$ decays exponentially in $n$ at nearly the polytope's rate."],"supporting_citations":[{"why":"Supplies the original complete characterization of absolute-PPT spectra; the equivalence in Theorem 4.1 invokes its Lemma 3.","marker":"[16]"},{"why":"Establishes that qubit-qudit absolute separability equals absolute PPT, giving $\\mathrm{ASEP}_{2,n}=\\mathrm{APPT}_{2,n}$.","marker":"[17]"},{"why":"Provides the partial-transpose eigenvalue lemma used in Theorem 4.1 and the witness-spectra descriptions used for the dual cones.","marker":"[18]"},{"why":"Shows the set of separable states has no finite semidefinite representation except in dimension $3\\times 2$, background for why $\\mathrm{ASEP}_{m,n}$ is only shown semialgebraic in general.","marker":"[22]"},{"why":"Gives prior extreme-point and maximum-purity results for qubit-qudit and qutrit-qudit absolutely separable and PPT states, used as comparisons and as evidence for the ordered-spectra extreme-point theorem.","marker":"[24]"},{"why":"Supplies the facts that spectrahedral faces are exposed and that faces of the positive semidefinite cone are described by image or kernel subspaces.","marker":"[31]"},{"why":"Provides the kernel-based description of spectrahedra and the dimension identities used in Proposition 5.4 and Lemma 5.1.","marker":"[32]"},{"why":"Yields the single linear inequality $2\\lambda_{mn}+\\sum_{k=1}^{m-1}\\lambda_{mn-k}\\ge\\sum_{k=1}^{m-1}\\lambda_k$ that defines the inscribed polytope in ordered form.","marker":"[36]"},{"why":"Supplies the closed-form simplex-section volume formula used to compute the exact relative volume of the inscribed polytope.","marker":"[40]"},{"why":"Provides the volume of the separable ball, the comparison baseline for the exponential volume decay results.","marker":"[41]"}],"fun_headline_variants":["Absolute-PPT spectra form spectrahedra for all m≤n","Every APPT matrix set is a spectrahedron with exposed faces","APPT spectra: always a spectrahedron, all faces exposed","For m≤n, APPT spectra are spectrahedra with only exposed faces","Absolute-PPT spectra: one spectrahedron for every bipartite dimension"],"cache_read_input_tokens":2816,"weakest_assumption_plain":"The load-bearing external premise is the known absolute-PPT criterion, specifically the lemma cited as [16, Lemma 3], which Theorem 4.1 invokes by reference: if that lemma, or its applicability to all $m\\le n$, is incomplete, then the spectrahedron representation of $\\mathrm{APPT}_{m,n}$, the exposed-face corollary, and the rank-based dimension formula all inherit the failure.","fun_headline_variants_meta":{"raw":{"variants":["Absolute-PPT spectra form spectrahedra for all m≤n","Every APPT matrix set is a spectrahedron with exposed faces","APPT spectra: always a spectrahedron, all faces exposed","For m≤n, APPT spectra are spectrahedra with only exposed faces","Absolute-PPT spectra: one spectrahedron for every bipartite dimension"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000765,"raw_usage":{"total_tokens":3346,"prompt_tokens":980,"completion_tokens":2366,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":724,"completion_tokens_details":{"reasoning_tokens":2284}},"tokens_in":724,"tokens_out":2366,"duration_ms":19592,"temperature":1.0,"reasoning_tokens":2284,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-05T19:42:36.846980+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A concrete check would be to take a small case such as $m=n=3$, generate spectra that satisfy the known absolute-PPT inequalities of [16], and test whether $L_\\pi(\\lambda)\\ge 0$ for every permutation; a single violation would refute Theorem 4.1. Independently, the face-dimension formula would fail if any proper face's constraint matrix $C$ had rank outside $\\{0\\}\\cup\\{m,\\ldots,mn-1\\}$, and Conjecture 6.7 could be tested by a global SDP or branch-and-bound optimization for $m=5$ searching for an APPT spectrum with purity higher than the polytope's closed-form maximum.","supporting_citations":[{"cited_title":"Positive partial transpose from spectra,","cited_arxiv_id":null,"evidence_quote":"Supplies the original complete characterization of absolute-PPT spectra; the equivalence in Theorem 4.1 invokes its Lemma 3."},{"cited_title":"Separability from spectrum for qubit-qudit states,","cited_arxiv_id":null,"evidence_quote":"Establishes that qubit-qudit absolute separability equals absolute PPT, giving $\\mathrm{ASEP}_{2,n}=\\mathrm{APPT}_{2,n}$."},{"cited_title":"The inverse eigenvalue problem for entanglement witnesses,","cited_arxiv_id":null,"evidence_quote":"Provides the partial-transpose eigenvalue lemma used in Theorem 4.1 and the witness-spectra descriptions used for the dual cones."},{"cited_title":"The set of separable states has no finite semidefinite representation except in dimension 3×2,","cited_arxiv_id":null,"evidence_quote":"Shows the set of separable states has no finite semidefinite representation except in dimension $3\\times 2$, background for why $\\mathrm{ASEP}_{m,n}$ is only shown semialgebraic in general."},{"cited_title":"Extreme points of sets of absolutely separable and positive partial transpose states,","cited_arxiv_id":null,"evidence_quote":"Gives prior extreme-point and maximum-purity results for qubit-qudit and qutrit-qudit absolutely separable and PPT states, used as comparisons and as evidence for the ordered-spectra extreme-point theorem."},{"cited_title":"The geometry of spectrahedra,","cited_arxiv_id":null,"evidence_quote":"Supplies the facts that spectrahedral faces are exposed and that faces of the positive semidefinite cone are described by image or kernel subspaces."},{"cited_title":"Extreme points of gram spectrahedra of binary forms,","cited_arxiv_id":null,"evidence_quote":"Provides the kernel-based description of spectrahedra and the dimension identities used in Proposition 5.4 and Lemma 5.1."},{"cited_title":"Positive reduction from spectra,","cited_arxiv_id":null,"evidence_quote":"Yields the single linear inequality $2\\lambda_{mn}+\\sum_{k=1}^{m-1}\\lambda_{mn-k}\\ge\\sum_{k=1}^{m-1}\\lambda_k$ that defines the inscribed polytope in ordered form."},{"cited_title":"Volume of slices and sections of the simplex in closed form,","cited_arxiv_id":null,"evidence_quote":"Supplies the closed-form simplex-section volume formula used to compute the exact relative volume of the inscribed polytope."}],"review_version":1}