{"id":"3dfde0c6-8ab2-462d-9aa4-ccfd777bdf60","arxiv_id":"1908.03017","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"A consistent perturbation-theory framework for non-Hermitian (PT-symmetric, pseudo-Hermitian) quantum systems, organized as a five-Hilbert-space scheme that reduces to the standard three-space picture with a reconstructed metric.","lead":"Non-Hermitian Hamiltonians can still describe normal, unitary quantum systems if the underlying space is measured with a special metric. This paper works out a systematic five-space scheme for applying small perturbations to such systems, and shows the only safe perturbations are those whose corrected geometry stays consistent.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Central claim is conditional on an unverified EP-free analyticity assumption: Eq. (24) plus Eqs. (34)-(36) require convergent Δλ and Tλ series, but §4.2/§7.5 only assert 'safely diagonalizable' H without a criterion for W, and [13] gives λ_max=0.","rationale":"The reader's weakest_assumption points to the same analyticity/EP-free condition; I agree with that choice. The internal algebra of Eqs. (31)-(33) is consistent, and the paper is candid in Sections 7.6 and 8 that the admissibility test is a posteriori and that the model of [13] has λ_max=0. These admissions mark the exact place where the central claim is narrower than the abstract's 'solution of the problem': the method supplies a formal perturbation expansion under analyticity and EP-freeness, not a universal or a priori criterion. Because this is a limitation rather than an inconsistency, the conditional verdict is unchanged. The proposed test is minimal: the 2×2 model already exhibits the metric ambiguity and has a known EP structure, so it directly compares the recursive coefficients with an exact solution and would expose any failure of the regularity assumption.","tokens_in":18300,"tokens_out":15016,"duration_ms":146444,"concrete_test":"Run the recursion of §6.3 on the exactly solvable 2×2 Klein-Gordon model of Eqs. (14)-(15). Fix H=H(KG)(τ), choose a one-parameter perturbation λW directed toward the EP, compute T(1), T(2) and Δ(0) from Eqs. (36)-(38), and compare with the exact Taylor expansion of the solution Tλ of Eq. (26) obtained by direct diagonalization. Locate the nearest exceptional point λ_EP and check whether the series converges for |λ|<|λ_EP| and fails at λ_EP. Agreement inside the disk validates the EP-free assumption; divergence before λ_EP, or a first-order Sylvester equation with no solution for a W whose perturbed spectrum is real, would show the conditional core of the claim is not met.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The load-bearing condition is the regularity of the whole perturbed family, not the internal algebra. Equation (24) factorizes the perturbed Dyson map as Ωλ=Ω(1+λΔλ)Jλ, and Eqs. (34)-(36) postulate convergent Taylor series for Wλ, Δλ and Tλ. Section 4.2 assumes H is 'safely diagonalizable' and far from exceptional-point pathologies; Section 7.5 repeats that EPs must stay 'sufficiently remote'. No criterion is supplied to verify, from H and W alone, that the analyticity/EP-free domain contains the λ-neighborhood of interest. The author's own reference [13] gives a crypto-Hermitian family with λ_max=0, so the condition is not automatic in the non-Hermitian regime. If an EP lies inside the radius, the factorization and the order-by-order metric reconstruction of §6.3 are not justified. This is not an internal algebraic error, but it means the central 'constructive and mathematically consistent' reformulation is established only conditionally, on a model-dependent hypothesis that the paper neither tests nor bounds.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops a formalism for treating perturbations of crypto-Hermitian (quasi-Hermitian/PT-symmetric) Hamiltonians in a unitary quantum-mechanical framework. It introduces a five-Hilbert-space structure, merges the perturbed and unperturbed textbook spaces, derives an effective metric Tλ on the unperturbed working space K, and obtains equations connecting the 'upper-case' perturbation Wλ in K with the 'lower-case' Hermitian perturbation vλ in L. The main formal results are the crypto-Hermiticity equation (26), the exact relation (32) between Vλ and Wλ, the leading-order non-perturbative difference formula (33), and an order-by-order reconstruction scheme for the metric in Section 6.3. The paper argues that the only consistent criterion for admissible perturbations is self-adjointness in the perturbed physical Hilbert space, which it acknowledges to be an a posteriori test.","tokens_in":18364,"tokens_out":7171,"duration_ms":78688,"significance":"If the formalism is accepted, it clarifies why standard Rayleigh-Schrödinger intuition about 'small' perturbations fails in the quasi-Hermitian setting: the metric changes with λ, and the difference between Vλ and Wλ acquires a λ-independent commutator term. The paper is commendably explicit about the ambiguities of the metric (Lemma 1) and about the limitations (EP-free analyticity, a posteriori criterion). The algebra in Eqs. (31)-(33) and the order-by-order equations in Section 6.3 are straightforward and appear correct. However, the paper does not provide a test of the analyticity/EP-free condition for any concrete perturbed model, and its central 'constructive' claim remains conditional on this model-dependent hypothesis. The significance is therefore primarily conceptual rather than as a ready-to-use computational method.","major_comments":[{"comment":"The entire constructive scheme relies on the factorization Ωλ = Ω(1+λΔλ)Jλ and on convergent Taylor expansions for Wλ, Δλ, and Tλ in a neighborhood of λ=0. The paper only assumes (§4.2, §7.5) that the unperturbed Hamiltonian is 'safely diagonalizable' and that exceptional points stay 'sufficiently remote,' without giving a verifiable criterion in terms of H and W. Because reference [13] itself exhibits a crypto-Hermitian family with λ_max = 0, this hypothesis is not automatic and is exactly what determines whether the proposed perturbation series is valid. Please state the precise analyticity/regularity hypotheses as a theorem, or provide a practical method to bound the EP-distance from H+λW.","section":"§5.2 and §6.3 (Eqs. (24), (34)-(36))"},{"comment":"The paper's central criterion—that admissible perturbations are those self-adjoint in the perturbed physical Hilbert space—is, as the paper states, an a posteriori self-consistency test: the metric Tλ is obtained from Eq. (26), which involves the perturbed Hamiltonian itself. Consequently, the criterion does not provide an independent way to decide, before solving the theory, whether a given λW is admissible. This limitation should be stated prominently in the abstract and introduction, since the paper describes itself as reopening the problem of the smallness of perturbations.","section":"§7.6"},{"comment":"The proof of Lemma 3 establishes only that existence of a positive-definite Tλ implies real spectrum. In Sections 6.2 and 7.2 the reality of the spectrum is used as a proxy for admissibility of perturbations; this requires the converse statement (real spectrum implies existence of a positive-definite metric), which is nontrivial and not true for arbitrary unbounded non-Hermitian operators. Please restrict the claim to the setting where the converse holds (e.g., finite-dimensional diagonalizable operators, or bounded operators with a reference), or provide a proof or a citation.","section":"§6.1, Lemma 3"}],"minor_comments":[{"comment":"The arXiv header contains 'syste ms' with a stray space in 'systems'; please fix this typographical issue.","section":"Title page"},{"comment":"The author name is given as 'Z. Znojil'; it should be 'M. Znojil'.","section":"Reference [21]"},{"comment":"The positivity condition |β| < 1 is stated but not derived; adding a sentence on the determinant 1−β² and the positive trace of Θ(KG)(τ,β) would make the example more self-contained.","section":"§3.2, Eq. (15)"},{"comment":"The flowcharts are set as ASCII text; please consider typesetting them as figures or tables to improve readability.","section":"Equations (10), (12), (20), (27)"},{"comment":"The notation ~W0 is not defined in the text; please define it explicitly as the leading-order approximation of Vλ.","section":"§6.3, Eq. (38)"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is largely self-referential and would benefit from a clearer positioning against the existing literature on perturbation theory in pseudo-Hermitian settings (e.g., works by Mostafazadeh and by Krejčiřík and collaborators). The main constructive claim is conditional on an analyticity assumption that the paper does not operationalize; this should be addressed before publication. The paper is probably better suited to a mathematical-physics venue than to a broad interdisciplinary journal."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: the paper earns its place by making one clean point — in crypto-Hermitian perturbation theory, smallness of λW in the working space does not imply smallness of the corresponding textbook perturbation, and Eq. (33) states that difference as a commutator term. The five-Hilbert-space merger is mostly bookkeeping, but the reduction to the effective metric Tλ (Eqs. 24-27) and the order-by-order recursion in §6.3 are real. I spot-checked the algebra: Eq. (32) correctly solves Eq. (31), and the leading-order commutator in Eq. (33) is right. The paper is also honest where it matters. Section 7.6 admits that the admissibility criterion is an a posteriori self-consistency test, not a predictive criterion. That does not sink the paper, but it should be reflected in the abstract: this is a consistent reformulation plus a formal recursion, not a solution to the problem of smallness.\n\nThe main soft spot is exactly the one in the stress-test note, and it is real. The scheme requires the whole perturbed family to stay analytic, diagonalizable, and away from exceptional points, but the paper states this assumption (§4.2, §7.5) without giving a checkable condition on H and W. The author's own [13] has a crypto-Hermitian family with λmax = 0, so this is not an empty caveat. The theory is therefore established conditionally, for a class of models that is not characterized. The author clearly knows this; it is in the text, but the limitation should be front and center.\n\nSecond soft spot: no worked example. A paper whose pitch is 'the metric reconstruction is doable order by order' needs at least one non-commuting 2x2 example. It would make the recursion concrete and would test Eq. (33) rather than leaving it formal. Third, Lemma 3 is a restatement of the known real-spectrum-hidden-Hermiticity theorem; the one-sentence proof is acceptable but adds nothing. Minor: the prose is repetitive and the citation pattern is heavily self-referential, though most of those citations are genuinely on-topic.\n\nWho this is for: people working on PT-symmetric or pseudo-Hermitian perturbation theory. It is a useful reference for the point that smallness does not transfer across Hilbert spaces. It is not a finished recipe for applications. I would send it to peer review and ask for two things: a concrete example, and a sharper statement of the regularity assumptions, ideally with a bound that connects H and W to the distance to the nearest exceptional point. With those, the paper would be a solid contribution to a niche literature.","headline":"The paper makes a useful formal point about smallness in crypto-Hermitian perturbation theory, but its central claim is conditional on an unchecked analyticity assumption.","tokens_in":19128,"tokens_out":3302,"would_cite":true,"duration_ms":34699,"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":"A perturbation theory for closed non-Hermitian systems is consistent once the physical Hilbert-space metric is rebuilt order by order.","keywords":["hidden Hermiticity","Hilbert space metric","perturbation theory","PT symmetry","unitary quantum evolution","quasi-Hermitian Hamiltonian","exceptional points","stability"],"falsifier":"In the paper's own two-by-two matrix example, choose a perturbation W for which the spectrum of H0+λW is real and non-degenerate for all small λ, and solve Eq. (26) exactly for Tλ. If a positive-definite solution fails to exist for arbitrarily small λ while the operator remains diagonalizable, the paper's characterization of admissible perturbations is false; if a positive-definite solution always exists in such cases, the characterization is supported.","tokens_in":17881,"feed_emoji":"⚛️","tokens_out":9531,"duration_ms":91352,"temperature":0.7,"pith_summary":"The paper argues that perturbation theory for closed quantum systems described by non-Hermitian but crypto-Hermitian Hamiltonians can be made mathematically consistent, provided one tracks the geometry of the physical Hilbert space alongside the Hamiltonian. The author shows that a perturbed Hamiltonian H+λW must be interpreted in a five-Hilbert-space structure, which then reduces to a three-space scheme in which the physical metric is reconstructed order by order. The central conclusion is that an operationally admissible perturbation is exactly one that is self-adjoint in the perturbed physical Hilbert space, and that this condition cannot be read off from the working-space operator alone. Because the metric itself changes with λ, the difference between how a perturbation looks in the working space and in the textbook space is generically non-perturbative even at small λ. This matters for stability analysis and for any model where non-Hermitian representations of unitary systems are used.","feed_headline":"Five Hilbert spaces make non-Hermitian perturbation theory consistent","feed_subtitle":"For crypto-Hermitian systems, small λ does not mean a small effect: the metric itself changes with the perturbation.","key_machinery":"The central object is the five-Hilbert-space flowchart and its reduction to a three-space scheme through the effective metric Tλ in the fixed working space K. The load-bearing identity is the hidden-Hermiticity equation (H†+λW†)Tλ=Tλ(H+λW), which defines which perturbations admit a unitary interpretation, together with the perturbation expansion of the metric, Tλ=Θ+λT(1)+..., whose first-order term obeys H†T(1)+W0†Θ=ΘW0+T(1)H. These equations convert the problem of 'which perturbations are small' into the problem of reconstructing a positive-definite metric that makes the perturbed Hamiltonian self-adjoint.","core_discovery":"The discovery is a constructive reformulation of perturbation theory for unitary non-Hermitian systems. Treating the unperturbed and perturbed systems each within the three-Hilbert-space (3HS) formalism, and identifying the two textbook Hilbert spaces, yields a five-Hilbert-space flowchart; eliminating the λ-dependent working space Kλ through the map Jλ and writing the perturbed map Ωλ=Ω(1+λΔλ)Jλ converts the hidden-Hermiticity condition into the operator equation (H†+λW†)Tλ=Tλ(H+λW), with Tλ=(1+λΔλ†)Θ(1+λΔλ) the effective metric in the fixed working space. The paper derives order-by-order equations for the metric corrections, shows that a real spectrum is necessary for a positive-definite Tλ to exist, and proves that the leading difference between the perturbation Vλ seen from the textbook space and the perturbation Wλ prescribed in the working space is the commutator Δ0H−HΔ0, which need not vanish or be small as λ→0. On this basis the author claims that the operational admissibility of a perturbation is governed by self-adjointness in the perturbed physical Hilbert space, not by Hermiticity in the auxiliary space.","pith_inferences":["The paper leaves implicit a practical criterion that is testable numerically: attempt to solve Eq. (26) for a positive-definite Tλ at small λ, and treat failure as an operational sign that the perturbation is inadmissible, even when the spectrum is still real.","The appearance of the commutator Δ0H−HΔ0 in Eq. (33) points toward a geometric interpretation: the 'extra' non-perturbative contribution looks like a generator of a unitary rotation of the perturbation, so the formalism may connect to geometric-phase effects in parameter-driven non-Hermitian devices.","A natural extension would promote λ to a time-dependent parameter; the order-by-order metric reconstruction then becomes a consistency condition for adiabatic following, a setting not addressed in the paper."],"forward_implications":["A non-Hermitian perturbation in the working space can be physically admissible even if it is not Hermitian there, as long as the perturbed metric makes it self-adjoint; the smallness of λ alone does not establish the smallness of the physical effect.","Computing observable corrections requires building the metric order by order together with the wavefunctions, so the standard perturbation series is replaced by a coupled system for energy, state, and metric corrections.","Stability of a closed crypto-Hermitian system is tied to the existence of a positive-definite solution Tλ, which in turn forces the perturbed spectrum to stay real; exceptional points inside the convergence radius invalidate the scheme.","The non-perturbative commutator difference Vλ−Wλ=Δ0H−HΔ0+O(λ) means that conclusions drawn from the working-space perturbation alone can be quantitatively wrong, and the discrepancy is model-dependent.","The inherent ambiguity of the metric can be removed by extending the dynamical input to a complete set of observables, exactly as in the unperturbed case, so the formal consistency carries over."],"supporting_citations":[{"why":"Supplies the three-Hilbert-space framework and the quasi-Hermiticity condition H†Θ=ΘH that the paper extends.","marker":"[6]"},{"why":"Introduces the class of non-Hermitian Hamiltonians with real spectra that motivates the whole analysis.","marker":"[4]"},{"why":"Supplies the non-unitary similarity-transform idea behind moving from a hard textbook Hamiltonian to a simpler isospectral partner.","marker":"[5]"},{"why":"Defines the perturbation-theory convergence radius in terms of exceptional points, the benchmark the paper reinterprets.","marker":"[1]"},{"why":"Provides the review in which the physical metric is realized as a product of parity and charge operators.","marker":"[7]"},{"why":"Provides the counterexample with λmax=0 that delimits the assumptions of analyticity and exceptional-point avoidance.","marker":"[13]"},{"why":"Raises the rigor problems for unbounded quasi-Hermitian operators that motivate boundedness and diagonalizability assumptions.","marker":"[19]"},{"why":"Shows the failure of conventional perturbation expansions near exceptional-point singularities, justifying the safe-diagonalizability assumption.","marker":"[33]"},{"why":"Documents the ambiguity of the metric reconstruction, which the present framework carries into the perturbed setting.","marker":"[25]"}],"fun_headline_variants":["Five-Hilbert-space method fixes non-Hermitian perturbation theory","Metric changes, not just λ, govern non-Hermitian perturbations","Five Hilbert spaces make perturbation theory consistent in non-Hermitian systems","Hidden metric shifts: why small λ can still break non-Hermitian systems","Perturbation theory for crypto-Hermitian systems requires five Hilbert spaces"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole construction assumes the perturbed Hamiltonian and the metric can be expanded as power series in the perturbation strength λ in a whole neighbourhood of λ=0, and that the system never meets a degeneracy of the exceptional-point type inside that neighbourhood; if such a point lies inside the radius of convergence, the scheme collapses.","fun_headline_variants_meta":{"raw":{"variants":["Five-Hilbert-space method fixes non-Hermitian perturbation theory","Metric changes, not just λ, govern non-Hermitian perturbations","Five Hilbert spaces make perturbation theory consistent in non-Hermitian systems","Hidden metric shifts: why small λ can still break non-Hermitian systems","Perturbation theory for crypto-Hermitian systems requires five Hilbert spaces"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000296,"raw_usage":{"total_tokens":1744,"prompt_tokens":994,"completion_tokens":750,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":610,"completion_tokens_details":{"reasoning_tokens":655}},"tokens_in":610,"tokens_out":750,"duration_ms":7572,"temperature":1.0,"reasoning_tokens":655,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T14:31:56.388516+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"In the paper's own two-by-two matrix example, choose a perturbation W for which the spectrum of H0+λW is real and non-degenerate for all small λ, and solve Eq. (26) exactly for Tλ. If a positive-definite solution fails to exist for arbitrarily small λ while the operator remains diagonalizable, the paper's characterization of admissible perturbations is false; if a positive-definite solution always exists in such cases, the characterization is supported.","supporting_citations":[{"cited_title":"Znojil, Phys","cited_arxiv_id":null,"evidence_quote":"Provides the counterexample with λmax=0 that delimits the assumptions of analyticity and exceptional-point avoidance."},{"cited_title":"Dieudonne, Proc","cited_arxiv_id":null,"evidence_quote":"Raises the rigor problems for unbounded quasi-Hermitian operators that motivate boundedness and diagonalizability assumptions."},{"cited_title":"Krejˇ ciˇ r ´ ık, V","cited_arxiv_id":null,"evidence_quote":"Documents the ambiguity of the metric reconstruction, which the present framework carries into the perturbed setting."}],"review_version":1}