{"id":"5fb6c3f6-1f70-4128-b71b-dd0569822230","arxiv_id":"2608.08346","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"low","formal_verification":"none","parameter_count":1,"one_line_summary":"A non-unitary similarity deformation of a Hermitian two-qubit Hamiltonian keeps the degenerate spectrum fixed while right eigenstates interpolate from product to maximally entangled, with Schmidt-gauged magic peaking at intermediate coupling.","lead":"A non-Hermitian Hamiltonian built by a non-unitary similarity transform keeps its eigenvalues fixed while its right eigenstates can smoothly change from product to maximally entangled, so entanglement and spectral degeneracy are logically independent. The paper gives an exactly solvable two-qubit model and closed formulas for entanglement, magic, and Bell-nonlocality diagnostics.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The central analytic result is sound, but the claimed physical interpretation of the degenerate eigenspace states and the numerical protocol rest on an unproven conventional choice.","rationale":"The reader's weakest assumption correctly identifies the physical-state interpretation and the basis dependence of the degenerate eigenspace as the principal concern. My stress test agrees: the analytic derivations are correct and internally consistent, but the paper's headline claims about monotonic entanglement and CHSH violation depend on choosing the standard-inner-product right-state density matrix in a particular basis of the degenerate subspace. The paper itself flags this in Section III.C, noting that an entropy assigned to a single vector in a degenerate eigenspace depends on the selected basis, and that only ensemble quantities can be basis-invariant. Yet the abstract and discussion present the two-qubit results without this caveat. Additionally, the numerical protocol in Section VI is described algorithmically but no code or computed results are provided, so the claimed scalability and Haar-typicality comparisons are not verified. These issues do not undermine the mathematical correctness of the model, but they mean the physical significance of the results is conditional on a stated convention and on numerical support. Hence CONDITIONAL acceptance remains the appropriate verdict, with the conditions being (i) explicit acknowledgment that the entanglement and Bell results are basis-dependent properties of a chosen eigenbasis, and (ii) shipped, runnable code demonstrating the larger-space protocol.","tokens_in":22543,"tokens_out":1508,"duration_ms":13270,"concrete_test":"Test the basis dependence and the numerical protocol: (1) Fix gamma>0 and rotate the normalized right eigenstates |R_00>, |R_01> by a unitary U within the degenerate eigenspace (e.g., U = (|R_00>+|R_01>)/sqrt(2), ...), then compute the reduced density matrix, von Neumann entropy, PPT negativity, and CHSH violation of the rotated states. If any of these quantities differs from the paper's Eqs. (22), (58), and (203), the monotonic entanglement/Bell claims are basis-dependent and not intrinsic to the non-Hermitian degeneracy. (2) Implement the Section VI protocol for a 3×3 or 4×4 bipartition with a chosen degenerate H0 and S_gamma; verify that the rank condition, the entropy deficits, and the claimed comparison with Page benchmarks match the promised behavior.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The paper's central construction is internally consistent: a Hermitian H0 with a degenerate sector is similarity-transformed by S_gamma = exp(gamma sigma_x ⊗ sigma_x), yielding a non-Hermitian, non-defective H_gamma with unchanged spectrum. The two-qubit eigenvalue problem, Schmidt probabilities p_± = (1±sech(2gamma))/2, negativity, and magic are all derived correctly. The load-bearing issue is the physical interpretation of the degenerate eigenspace states. Since the eigenvalue lambda_0 has multiplicity two, the normalized right eigenstates |R_mu> = S_gamma|mu> are only one possible basis of a two-dimensional eigenspace; a unitary rotation within that eigenspace changes entanglement and CHSH violation. The paper acknowledges this basis dependence for individual vectors and introduces the orthogonal projector Q_R0 for basis-independent diagnostics, but the claimed monotonic entanglement, PPT negativity, and CHSH violation are all properties of the specific basis {|R_00>, |R_01>}, not of the eigenspace or the Hamiltonian alone. Moreover, the biorthogonal reduction (Eq. 14) is non-positive and the paper itself shows it should not be compared with Page values. No first-principles argument selects the standard-inner-product right-state density matrix as the physical state; this is a conventional choice, and the claimed physical results depend on it. The numerical section (Section VI) promises a protocol and references Haar benchmarks, but provides no explicit code, no numerical results for larger dimensions, and no demonstration that the proposed subspace sampling reproduces the claimed diagnostics for any example beyond the analytically solved two-qubit case.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper constructs a non-Hermitian Hamiltonian H_gamma = S_gamma H_0 S_gamma^{-1} with S_gamma = exp(gamma sigma_x otimes sigma_x), starting from a Hermitian H_0 with a degenerate sector. It proves that the degenerate eigenvalue remains non-defective for all finite gamma, solves the two-qubit realization exactly, and derives closed expressions for the Schmidt probabilities, von Neumann and Rényi entropies, PPT negativity, Schmidt-gauged non-local magic, a triangle-criterion magic witness, CHSH violation, and comparisons with Page/Haar benchmarks. The central algebraic claims are internally consistent, but the physical interpretation of the results depends on a specific choice of basis inside the degenerate eigenspace and on the convention that normalized right eigenstates are the physical states.","tokens_in":22763,"tokens_out":7335,"duration_ms":67824,"significance":"The paper provides a clean, fully analytic demonstration that a non-Hermitian Hamiltonian can have an unchanged degenerate spectrum while a specially selected family of right eigenstates interpolates from product to maximally entangled states. The closed forms for the negativity, magic, and CHSH violation are useful, and the explicit warning against using the non-positive biorthogonal reduction for entropy comparisons is well taken. The main value is pedagogical and technical: it isolates the distinction between spectral degeneracy, eigenvector entanglement, and eigenvector geometry. However, the claimed physical conclusions rest on a basis-selection and inner-product convention that the manuscript does not fully justify, so the presentation needs substantial reframing before the results can be stated as physical rather than conditional.","major_comments":[{"comment":"The monotonic growth of entanglement, PPT negativity, and CHSH violation is a property of the particular basis {|R00>, |R01>} = {S_gamma|00>, S_gamma|01>}, not of the degenerate eigenspace of H_gamma or of H_gamma itself. The paper explicitly acknowledges after Eq. (9) that an entropy assigned to a single vector in a degenerate eigenspace depends on the chosen basis and offers the projector Q_R0 as the basis-independent diagnostic, yet the abstract and Section VII present the monotonic entanglement as the physical result. The Hamiltonian alone does not single out this basis, so either a physical selection rule (e.g., analytic continuation in gamma from the Hermitian eigenbasis, or a preparation protocol) must be stated and used, or the claims must be explicitly qualified as holding for a chosen similarity-selected basis. This is load-bearing because a different orthonormal basis of the same eigenspace gives different Schmidt probabilities, entropies, negativity, and CHSH values.","section":"Section III.A, Eqs. (18)-(21), and abstract"},{"comment":"Despite the title 'Numerical Implementation in Larger Dimensions,' Section VI contains no numerical results, no code, no tables, and no figures; it lists only a protocol. The ensemble diagnostics in Eqs. (127)-(131) are defined but never evaluated, even for the exactly solvable two-qubit case where a closed-form subspace average would be straightforward to compute. The abstract and Section VII state that Page and Haar benchmarks provide reference values for eigenstate typicality, but this is not demonstrated for the model. Please supply actual numerical or analytic evaluations, or explicitly state that Sections V and VI are proposed protocols rather than executed computations.","section":"Sections V and VI, Eqs. (127)-(131)"},{"comment":"The identification of the normalized right-state density matrix in Eq. (10) as the physical state is a convention rather than a consequence of the formalism. The paper correctly shows that the biorthogonal reduction is non-positive and should not be compared with Page values, but the opposite convention is logically allowed in non-Hermitian quantum mechanics; under the biorthogonal convention none of the monotonic entanglement, PPT, or CHSH results follow. The paper should state this conventional dependence explicitly at the point where Eq. (10) is introduced and in the abstract, rather than presenting the right-state results as the unique physical ones.","section":"Section III, Eqs. (10) and (14), and Section VII"}],"minor_comments":[{"comment":"The phrase 'p spherical symmetry of Haar measurement' appears to contain a typo; it should probably read 'the spherical symmetry of the Haar measure'.","section":"Eq. (41)"},{"comment":"The heading 'Numerical Implementation in Larger Dimensions' is misleading because no numerical implementation is reported; renaming the section to 'Proposed Numerical Protocol' or adding actual computational results would remove the mismatch.","section":"Section VI"},{"comment":"The branch choice of the matrix logarithm is not specified at the definition in Eq. (15); the principal branch is invoked only later in Eq. (26). State the branch convention explicitly at the definition.","section":"Eqs. (15) and (26)"},{"comment":"The notation in Eq. (4) is confusing: S_gamma^dagger = e^{gamma G^dagger} = e^{gamma G} = S_gamma should be written with the Hermiticity of G stated before the equality, to avoid the impression that S_gamma is being defined in terms of itself.","section":"Eq. (4)"}],"recommendation":"major_revision","confidential_remarks":null},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Bottom line: the central analytic model is correct and the exact two-qubit formulas are new. The paper deserves a serious referee; it also needs a tighter framing of what its entanglement claims mean physically.\n\nWhat is actually new: the two-qubit realization S_gamma = exp(gamma sigma_x⊗sigma_x) on a degenerate H0 gives an exact, isospectral, non-defective non-Hermitian family. The Schmidt probabilities p± = (1±sech(2γ))/2, the von Neumann and Rényi entropies, PPT negativity N = ½|tanh(2γ)|, and Schmidt-gauged magic M = -log2(1-u²+u⁴) all check out. I independently did the algebra; the results are internally consistent. The paper is also appropriately careful to distinguish the positive right-state reduction from the non-positive biorthogonal one, and to flag that the computational-basis IPR coincides with reduced purity only because the Schmidt and computational bases align in this model.\n\nSoft spots, in rough order of size. First, the interpretation of the degenerate eigenspace. Because λ0 has multiplicity two, the right states |R_00>, |R_01> are one choice of basis; a unitary rotation inside the eigenspace changes their entanglement, negativity, and CHSH value. The paper says this, but then proceeds to treat the monotonic entanglement and Bell violation as properties of the physical state. They are properties of a chosen basis, and the choice is only justified by the convention that normalized right states under the standard inner product are the physical states. That is a legitimate convention in non-Hermitian quantum mechanics, but it is a convention, not a derivation. The subspace projector Q_R^0 and the sampling in Eq. (127) are basis-independent, but the headline results are not. This should be stated much more prominently.\n\nSecond, Section VI promises a numerical protocol but reports no actual numbers, no code, and no demonstration beyond the analytically solved two-qubit case. For a paper that stresses reproducibility and typicality benchmarks, that's a hole. It's easy to fill; the referee should ask for it.\n\nThird, the paper carries a lot of tangential machinery—Floquet reservoir conversion, finite unitary designs, twisted Rényi correlators, CHSH—that is not needed for the core claim. Some of it is correct but reads as padding. The core contribution is the solvable two-qubit family; the extra sections mainly add length.\n\nThe self-citation to Ref. [5] is motivational and not load-bearing; I see no citation-pattern problem. The linear algebra and the main derivations are sound. This is a useful conceptual benchmark for non-Hermitian many-body diagnostics: it cleanly separates non-defective degeneracy from exceptional-point collapse, and shows spectrum freezing does not freeze eigenvector entanglement. I would send it to peer review and would probably accept after revisions requiring: (i) an explicit statement that the right-state density matrix is a physical convention, (ii) actual numerical results or shipped code for the larger-dimension protocol, and (iii) trimming of the tangential sections.\n\nWho should read it: anyone building entanglement diagnostics for non-Hermitian lattice models, especially people who care about the difference between spectral and eigenvector properties. Worth a reading group session, and I would cite it if I were writing on non-Hermitian entanglement.","headline":"A correct and mostly new two-qubit toy model separating non-defective degeneracy from eigenvector entanglement; the math holds, but the physical-state convention needs clearer framing and the numerical part is unfinished.","tokens_in":23372,"tokens_out":2721,"would_cite":true,"duration_ms":24764,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Central claim: applying an invertible non-unitary similarity to a degenerate Hermitian operator leaves the spectrum unchanged while the right eigenstates interpolate from product to maximally entangled states.","keywords":["non-Hermitian quantum mechanics","non-defective degeneracy","bipartite entanglement","Schmidt spectrum","non-local magic","stabilizer Rényi entropy","partial transpose negativity","CHSH inequality"],"falsifier":"Prepare the two-qubit state $|\\tilde R_{00}(\\gamma)\\rangle$ for a finite nonzero $\\gamma$ and measure the eigenvalues of the partial transpose and the maximal CHSH value; if the negativity is not $\\tfrac12|\\tanh 2\\gamma|$ or the CHSH value does not exceed 2, the central claim is wrong.","tokens_in":22276,"feed_emoji":"🔗","tokens_out":11196,"duration_ms":90045,"temperature":0.7,"pith_summary":"This paper proves that a non-Hermitian degeneracy need not be an exceptional point and that the spectrum does not control how entangled the degenerate eigenstates are. It builds a two-qubit Hamiltonian by conjugating a Hermitian operator with a degenerate eigenspace under the invertible non-unitary map $S_\\gamma=e^{\\gamma\\,\\sigma_x\\otimes\\sigma_x}$; the spectrum stays $\\{\\lambda_0,\\lambda_0,\\lambda_1,\\lambda_2\\}$ for every finite $\\gamma$, yet the right eigenstates interpolate continuously from product states at $\\gamma=0$ to Bell states as $|\\gamma|\\to\\infty$. The paper supplies closed formulas for the resulting entanglement, PPT negativity, CHSH violation, and Schmidt-gauged non-local magic, and it warns that the biorthogonal left-right reduction is non-positive and cannot be compared with Haar-random Page benchmarks. The payoff is a compact diagnostic tuple separating algebraic degeneracy, spectral sensitivity, entanglement typicality, and non-stabilizer structure.","feed_headline":"Entanglement can reach its maximum while the spectrum stays fixed","feed_subtitle":"Exact two-qubit model: entanglement rises to the Bell limit while magic peaks at mid-coupling.","key_machinery":"The load-bearing object is the invertible non-unitary similarity map $S_\\gamma=e^{\\gamma G}$ with $G=\\sigma_x\\otimes\\sigma_x$, applied to a Hermitian $H_0$ with an $r$-fold degenerate eigenvalue. Because similarity preserves characteristic and minimal polynomials and maps the degenerate subspace bijectively onto $\\ker(H_\\gamma-\\lambda_0 I)$, the degeneracy remains non-defective at every finite $\\gamma$; the orthogonal projector $Q_0^R$ onto the right eigenspace gives a basis-independent geometric description. On the state side, the right eigenvectors $|R\\rangle=S_\\gamma|\\mu\\rangle$ carry Schmidt probabilities controlled by $\\operatorname{sech}2\\gamma$, and all diagnostic formulas—entropies, PPT negativity, CHSH bound, and Schmidt-gauged magic—are evaluated from that positive right-state reduction, with the biorthogonal reduction kept separate.","core_discovery":"The central claim is that eigenvector entanglement can change continuously while the spectrum and the algebraic structure of a degenerate eigenvalue are held fixed. For the two-qubit realization, the normalized right eigenstates $|\\tilde R_{00}\\rangle = (\\cosh\\gamma\\,|00\\rangle+\\sinh\\gamma\\,|11\\rangle)/\\sqrt{\\cosh 2\\gamma}$ have Schmidt probabilities $p_\\pm = \\tfrac12(1\\pm \\operatorname{sech} 2\\gamma)$, so the von Neumann entropy, second Rényi entropy, PPT negativity $\\tfrac12|\\tanh 2\\gamma|$, and maximal CHSH value $2\\sqrt{1+\\tanh^2 2\\gamma}$ all increase monotonically with $|\\gamma|$ and saturate at the Bell values, while the Schmidt-gauged magic $M_2^{\\mathrm{Sch}}=-\\log_2(1-u^2+u^4)$ with $u=\\operatorname{sech}2\\gamma$ vanishes at both endpoints and peaks at $\\log_2(4/3)$ when $u^2=1/2$. The same construction separates the physical positive reduction $\\rho_A^R=\\mathrm{Tr}_B(|\\tilde R\\rangle\\langle\\tilde R|)$ from the generally non-positive biorthogonal reduction $\\rho_A^{RL}$, whose entropy can become complex.","pith_inferences":["The same isospectral mechanism should work for any Hermitian generator $G$ whose degenerate-sector eigenvectors have nonzero variance under $G$; the two-qubit choice $G=\\sigma_x\\otimes\\sigma_x$ is the simplest instance of a general 'entanglement without spectral change' phenomenon.","Because the entanglement and CHSH results follow only if the normalized right eigenstates define the physical states, a direct measurement of CHSH or reduced purity on a prepared $S_\\gamma|\\mu\\rangle$ state would empirically select the right-state convention over the biorthogonal one.","The dark-state construction in the paper turns the static family into a resource: the two-mode-squeezing Lindbladian with the stated jump operators has $|\\psi_\\gamma\\rangle$ as a steady state, suggesting a concrete dissipative preparation route.","The exact equality between the computational-basis inverse participation ratio and the reduced purity is a coincidence of basis alignment; a generic global basis rotation would break it, so numerical uses of IPR as an entanglement proxy should check the basis first."],"forward_implications":["For finite $\\gamma\\neq 0$, the normalized two-qubit right eigenstate violates the CHSH inequality ($B_{\\max}>2$) and reaches the Tsirelson bound only as $|\\gamma|\\to\\infty$, so the model provides a one-parameter isospectral family that is provably nonlocal.","Entanglement growth and proximity to a defective exceptional point are decoupled: the condition number $\\kappa_2(S_\\gamma)=e^{2|\\gamma|}$ grows without bound while the degeneracy stays non-defective, so exponential spectral sensitivity can coexist with an ordinary diagonalizable degeneracy.","Non-local magic is not inherited from entanglement: the Schmidt-gauged magic vanishes for both product and maximally entangled states and is largest at $u^2=1/2$, so entropy and non-stabilizer content answer different questions.","The two-qubit right state can be verified by two local measurements—one in the Schmidt basis and one in a mutually unbiased basis—with spectral gap $1/2$, avoiding full tomography.","In larger $N\\times M$ bipartitions, Haar-typicality benchmarks must be applied to the right eigenspace through projected sampling rather than by substituting the rank $r$ into full-space Page formulas; the resulting deficits diagnose localized or entanglement-restricted sectors."],"supporting_citations":[{"why":"Supplies the perturbation-theory context in which geometric and algebraic multiplicity agree, supporting the non-defective degeneracy statement.","marker":"[3]"},{"why":"Earlier random-matrix bipartite construction showing rank deficiency and non-defective degeneracy can coexist; this paper isolates the question analytically.","marker":"[5]"},{"why":"Provides the Page formula for the Haar-averaged von Neumann entropy used as the entanglement benchmark.","marker":"[7]"},{"why":"Supplies the positive-partial-transpose separability criterion behind the two-qubit negativity witness.","marker":"[14]"},{"why":"Defines the Schmidt-gauged non-local magic from Walsh-Hadamard autocorrelations of the Schmidt spectrum.","marker":"[15]"},{"why":"Gives the Pauli-correlator reconstruction and two-spin entanglement context for applying the PPT test to a quantum processor.","marker":"[16]"},{"why":"Supplies the triangle criterion used to certify mixed-state magic and to derive the white-noise threshold.","marker":"[20]"},{"why":"Provides the Schmidt-MUB quantum-state verification protocol with the spectral-gap bound used here.","marker":"[21]"},{"why":"Supplies the Horodecki criterion relating maximal CHSH violation to the two largest eigenvalues of $T^T T$.","marker":"[29]"}],"fun_headline_variants":["Entanglement maxes out while spectrum stays fixed","Non-Hermitian pair: spectrum static, entanglement Bell-bound","Mid-coupling peaks magic, entanglement saturates to Bell","Spectrum frozen, entanglement flies to the Bell limit","Entanglement and magic diverge in non-Hermitian qubits"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The physical-state interpretation assumes the normalized right eigenvectors $|R_\\mu\\rangle=S_\\gamma|\\mu\\rangle$ are the real states of the degenerate eigenspace; if one instead uses the biorthogonal left-right pairing, the reduced density matrix is non-positive and the monotonic entanglement and CHSH results do not follow.","fun_headline_variants_meta":{"raw":{"variants":["Entanglement maxes out while spectrum stays fixed","Non-Hermitian pair: spectrum static, entanglement Bell-bound","Mid-coupling peaks magic, entanglement saturates to Bell","Spectrum frozen, entanglement flies to the Bell limit","Entanglement and magic diverge in non-Hermitian qubits"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000203,"raw_usage":{"total_tokens":1447,"prompt_tokens":1067,"completion_tokens":380,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":683,"completion_tokens_details":{"reasoning_tokens":298}},"tokens_in":683,"tokens_out":380,"duration_ms":3771,"temperature":1.0,"reasoning_tokens":298,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T00:09:39.778715+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Prepare the two-qubit state $|\\tilde R_{00}(\\gamma)\\rangle$ for a finite nonzero $\\gamma$ and measure the eigenvalues of the partial transpose and the maximal CHSH value; if the negativity is not $\\tfrac12|\\tanh 2\\gamma|$ or the CHSH value does not exceed 2, the central claim is wrong.","supporting_citations":[{"cited_title":"Verify non-defectiveness from rank(Hγ −λ 0I) =D −r","cited_arxiv_id":null,"evidence_quote":"Supplies the perturbation-theory context in which geometric and algebraic multiplicity agree, supporting the non-defective degeneracy statement."},{"cited_title":"(127), reshape every state as anN × Mcoefficient matrixC, and computeρ A =CC †/Tr(CC †)","cited_arxiv_id":null,"evidence_quote":"Earlier random-matrix bipartite construction showing rank deficiency and non-defective degeneracy can coexist; this paper isolates the question analytically."},{"cited_title":"(61)– (62); for positive two-qubit mixed reductions, evaluate the PPT spectrum in Eq","cited_arxiv_id":null,"evidence_quote":"Provides the Page formula for the Haar-averaged von Neumann entropy used as the entanglement benchmark."},{"cited_title":"”Non-defective degeneracy in non-Hermitian bipartite system.” Physica Scripta 99.10 (2024): 105237","cited_arxiv_id":null,"evidence_quote":"Supplies the positive-partial-transpose separability criterion behind the two-qubit negativity witness."},{"cited_title":"”Biorthogonal quantum mechanics.” Journal of Physics A: Mathematical and Theoretical 47.3 (2014): 035305","cited_arxiv_id":null,"evidence_quote":"Defines the Schmidt-gauged non-local magic from Walsh-Hadamard autocorrelations of the Schmidt spectrum."},{"cited_title":"”Average entropy of a subsystem.” Physical review letters 71.9 (1993): 1291","cited_arxiv_id":null,"evidence_quote":"Gives the Pauli-correlator reconstruction and two-spin entanglement context for applying the PPT test to a quantum processor."},{"cited_title":"Short, and Andreas Winter","cited_arxiv_id":null,"evidence_quote":"Supplies the triangle criterion used to certify mixed-state magic and to derive the white-noise threshold."},{"cited_title":"”Eigenstate thermalization for degenerate observables.” Physical review letters 120.15 (2018): 150603","cited_arxiv_id":null,"evidence_quote":"Provides the Schmidt-MUB quantum-state verification protocol with the spectral-gap bound used here."},{"cited_title":"”Triangle Criterion: a mixed-state magic criterion with applications in distillation and detection.” PRX Quantum 7.3 (2026): 033016","cited_arxiv_id":null,"evidence_quote":"Supplies the Horodecki criterion relating maximal CHSH violation to the two largest eigenvalues of $T^T T$."}],"review_version":1}