{"id":"960ca3cf-bb18-48ce-83a3-7ece63bc8926","arxiv_id":"2506.21400","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"A non-unitary similarity transformation maps a ghostly two-dimensional oscillator with bounded but non-normalisable eigenstates to an isospectral Hermitian Hamiltonian whose eigenstates are normalisable.","lead":"The authors show that a Hamiltonian with a ghost-like negative kinetic term, whose bounded-energy states are non-normalisable, can be mapped by a non-unitary transformation to a Hermitian Hamiltonian with normalisable states and the same energies. The construction is a proof of concept on a two-dimensional oscillator related to the Pais-Uhlenbeck model, and is meant as a template for quantising higher-derivative theories without ghosts.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Unbounded η1/η2 domain and unproven positivity of ρ make the claimed Hilbert-space similarity between h0 and h3 formal; the bounded-spectrum/normalisability conclusion is not yet established as an operator statement.","rationale":"The reader's weakest_assumption correctly identifies the unbounded-operator and positivity gap as load-bearing. I reviewed the g=0 reduction and the constraints in §2.4; the algebra appears internally consistent, and the explicit Gaussian computations are plausible. However, because η1 and η2 are unbounded, the proof of concept verifies normalisability of specific Gaussian eigenstates under the new metric but does not establish that η is a legitimate similarity transformation of Hilbert-space operators or that ρ defines a positive inner product on the full sector. This does not disprove the construction, but it makes the central claim conditional on a domain and spectral analysis that the paper does not provide. The reader's CONDITIONAL verdict is therefore appropriate; if the proposed checks pass, the paper would merit acceptance, but the operator-domain verification is essential.","tokens_in":8824,"tokens_out":20464,"duration_ms":239240,"concrete_test":"Take the g=0 limit of §2.3 with ν=4, Ω=-2, δ=0, λ=2, where η1=exp(p_y^2/2). Write the explicit (1,1)-sector ground and first two excited states of h0. Compute their ρ-norms ⟨ϕ,η†ηϕ⟩ directly and verify they are finite and positive; also compute η1ϕ by Fourier transform and check membership in L2(R^2) for each state. If any excited state has infinite ρ-norm or fails to lie in D(η1), the sector-normalisability claim is false. As a second check, verify that the domain of h3 is transported by η^{-1} to a dense domain of h0 on which h0 is bounded below with respect to ρ; otherwise the isospectrality is only formal.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The construction's central claim requires that η = η2η1η0 be a well-defined invertible map on the relevant sector so that ηh0η^{-1}=h3 is a genuine spectral equivalence and ρ=η†η is a positive metric. In the proof-of-concept regime (§2.3, e.g. Ω=-2, δ=0, |λ|>√2), η1 = exp(κp_x^2/2 + ξp_y^2/2) has positive ξ; in momentum space it is multiplication by exp(ξp_y^2/2), an unbounded operator with unbounded inverse. η2 = exp(μpxpy+τxy) is likewise only formally defined. The paper never specifies domains for these operators or for h0/h3, and never proves that ρ is positive definite on a domain containing the (1,1) eigenstates. For unbounded operators, formal conjugation does not preserve spectra; the paper's check that the Gaussian ground state satisfies (2.6) after transformation tests normalisability of one wavefunction, not the operator identity, self-adjointness of h3, or positivity of ρ on the excited states. Thus the statement that the formerly unphysical sector becomes physical is supported only at the level of formal eigenfunctions, leaving the load-bearing Hilbert-space step unproven.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a method to render physical the 'ghost sector' of a higher time-derivative theory, using a two-dimensional oscillator with an indefinite kinetic term as a concrete model. Starting from a Hermitian Hamiltonian h0 whose eigensystem is taken from earlier work, the authors construct a sequence of non-unitary similarity transformations η = η2 η1 η0 designed to map h0 to a new Hermitian Hamiltonian h3. They define a new inner product metric ρ = η†η and claim that in a specific parameter region the formerly non-normalisable eigenstates of the (1,1)-sector become normalisable, while the spectrum remains bounded from below. The paper presents explicit formulas for the transformed Hamiltonian and for the ground-state wavefunction, and shows numerically that the normalisability conditions can be satisfied in a proof-of-concept regime.","tokens_in":9122,"tokens_out":4195,"duration_ms":48508,"significance":"If the construction can be made rigorous, the paper would provide a concrete example of how pseudo-Hermitian techniques can be used to reinterpret ghostly sectors of higher time-derivative theories as physical, potentially extending the toolbox for ghost-free quantum models. The explicit formulas and parameter scans are useful and the connection to the Pais-Uhlenbeck oscillator is relevant. However, the central claim rests on operator-theoretic steps that are only handled formally, so the significance is contingent on closing those gaps.","major_comments":[{"comment":"The similarity operators η1 and η2 are unbounded, yet the paper never specifies their domains or proves that they define invertible maps on L²(R²) in the parameter regions used. For example, with δ=0, Ω=-2, and |λ|>√2, ξ = λ/(λ²+Ω) is positive, so η1 in momentum space is multiplication by exp(ξ p_y²/2), an unbounded operator with unbounded inverse. Likewise η2 = exp(μ p_x p_y + τ x y) is only formally defined. For unbounded operators, formal conjugation does not preserve spectra without domain conditions. The statement that h0 and h3 are isospectral and that ρ = η†η is a positive metric is therefore not established at the operator level. This is load-bearing for the paper's central claim that the ghost sector becomes a genuine Hilbert-space sector. Please provide a careful domain analysis, prove positivity of ρ on a dense domain containing the eigenstates, and establish (essential) self-adjointness of h3, or justify the formal manipulations by an approximation argument.","section":"§2.3–2.4, Eqs. (2.7), (2.31)"},{"comment":"The normalisability check is performed only for the ground state ϕ3. The paper asserts that ground-state normalisability is sufficient because excited states inherit the Gaussian factor, but the excited states of the transformed Hamiltonian are never constructed. It is not shown that all excited states are square-integrable with respect to the new metric, nor that the metric is positive on the full sector. Since the physical interpretation requires normalisable eigenstates for the entire spectrum, this gap must be closed, for instance by writing the transformed excited states explicitly or by proving a general inheritance argument for the polynomial prefactors.","section":"§2.4, Eqs. (2.29)–(2.30) and p.2, p.7"},{"comment":"The claim that h3 'has lost its ghostly nature and possesses regions in parameter space where it is positive definite' is made without proof. Positive definiteness or at least boundedness from below of h3 is essential to the claim that the spectrum is bounded from below after the transformation. Please provide the missing derivation, or state precisely how this follows from isospectrality to h0 combined with the sector classification of [11].","section":"§2.4, Eqs. (2.21)–(2.22)"},{"comment":"The full eigensystem and the sector classification (ϵ, η) are imported from the authors' earlier work [11] without derivation or verification. The construction depends critically on the existence of a sector with bounded spectrum and non-normalisable states, and on the assertion that ground-state normalisability controls all excited states. If those results are not independently verified, the present proof of concept inherits any errors. Please either include a concise derivation of the needed parts of the eigensystem or state explicitly which results are assumed and why they are reliable.","section":"§2.1, Eqs. (2.2)–(2.5)"}],"minor_comments":[{"comment":"Typo: the adjoint action of η0 on p_y should read η0 p_y η0^{-1} = p_y - iλ y, not p_x - iλ y.","section":"Eq. (2.8)"},{"comment":"The notation for the spectrum is unclear: define the ranges of N and n precisely, and clarify whether the floor function is intended in the upper limit of n.","section":"Eqs. (2.2)–(2.3)"},{"comment":"The definitions of the δ+ and δ− branches and the reason why only the δ+ branch yields a normalisable solution should be stated more explicitly.","section":"Fig. 2 and surrounding text"},{"comment":"The statement that for g→0 the Hamiltonian becomes a sum of two harmonic oscillators in a certain parameter regime should be quantified; for example, with δ=0 and Ω=-2 the regime is |λ|>√2.","section":"§2.3, after Eq. (2.14)"},{"comment":"Reference [11] is cited in final form without an arXiv identifier; providing the arXiv number would aid verification.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"The paper is a proof of concept that relies heavily on the authors' own prior work [11] for the eigensystem and sector classification. The novelty over [11] is the similarity-transformation construction, but the missing functional-analytic justification is a substantial gap. The referee's view is that the central idea is plausible and worth publishing once the operator-theoretic issues are addressed. The paper's scope fits a journal that accepts constructive quantum-mechanics papers with formal methods, but the authors should be encouraged to add a rigorous section on domains and positivity."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Bottom line: this is a real proof-of-concept for a useful move—map a ghostly Hermitian Hamiltonian h0 through a chain of non-unitary transformations to a Hermitian isospectral h3, and redefine the metric so that the formerly non-normalisable sector becomes normalisable. The explicit formulas for H2, h3, ψ2, and φ3 are coherent, the Gauss decomposition of η2 is neat, and the normalisability regions in Figure 2 are concrete. That is genuine novelty, and it deserves a referee's time.\n\nWhat it does well: it targets the sector that actually has the bounded spectrum, shows by direct computation that the transformed ground state satisfies the normalisability inequalities in a non-empty parameter region, and is honest that this is a proof of concept with open questions. The connection to the Pais-Uhlenbeck oscillator makes the example meaningful.\n\nSoft spots, in order of size. First, the Hilbert-space step is asserted, not proven. In the regime used, η1 and η2 are formal exponentials of positive quadratic momentum and position operators; on L2(R^2) they are unbounded with unbounded inverses. The paper never specifies domains, never proves that ρ=η†η is a positive metric on the sector, and checks normalisability of one wavefunction rather than establishing self-adjointness of h3. So the claim that the sector is physical currently rests on formal eigenfunctions. This may be repairable—the PT literature routinely glosses such domain issues—but it is load-bearing and needs real work.\n\nSecond, the full eigensystem and sector classification are imported from the authors' earlier papers [11,30]. That is a legitimate division of labour, but it means the novelty is concentrated in the transformation machinery, and a referee should check that those prior results are stable. Third, positive definiteness of h3 is asserted in passing without derivation; plausible from the structure, but it should be shown. The domain/metric issue is the only one I'd flag as serious; the other two are minor.\n\nWho this is for: researchers in higher-derivative quantisation and PT-symmetric methods. A serious referee can extract value, but the paper needs a major revision before the central claim is established. My recommendation is to send it to peer review with a clear instruction that the operator-domain and metric-positivity analysis is a required addition, not a nicety.","headline":"A promising but formally unfinished proof-of-concept: the ghost-to-physical map works at the level of formal eigenfunctions, but the operator-domain and metric-positivity step needs real work before the claim is solid.","tokens_in":9622,"tokens_out":2970,"would_cite":false,"duration_ms":34850,"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 ghostly sector with bounded spectrum and non-normalisable states becomes physical under a non-unitary similarity transformation and a redefined inner product.","keywords":["higher time-derivative theories","ghost problem","non-unitary similarity transformations","pseudo-Hermitian quantum mechanics","PT symmetry","Pais-Uhlenbeck oscillator","normalisable eigenstates","bounded spectrum"],"falsifier":"Evaluate the new inner product $\\langle \\phi_0 | \\rho | \\phi_0 \\rangle$ for the parameter set in figure 2 ($\\nu=4$, $\\Omega=-2$, $g=3$, $\\lambda$ in the region $|\\lambda|>2.13$). If it diverges or is not strictly positive, or if the metric $\\rho = \\eta_0^\\dagger \\eta_1^\\dagger \\eta_2^\\dagger \\eta_2 \\eta_1 \\eta_0$ has a negative eigenvalue, the claimed normalisability fails. Alternatively, check directly whether $e^{\\xi p_y^2/2}$ maps $L^2(\\mathbb{R}^2)$ into $L^2(\\mathbb{R}^2)$ for $\\xi>0$ in that region.","tokens_in":8639,"feed_emoji":"👻","tokens_out":9973,"duration_ms":93435,"temperature":0.7,"pith_summary":"The paper tackles the long-standing ghost problem in higher time-derivative theories, where quantisation either gives unbounded spectra or non-normalisable eigenstates. It proposes a new method: map the Hermitian 'ghostly' Hamiltonian to another Hermitian Hamiltonian via a non-unitary similarity transformation, then redefine the inner product with the metric $\\rho = \\eta^\\dagger \\eta$ so that the previously non-normalisable states become normalisable while the spectrum stays bounded below. The authors demonstrate the method on a concrete model related to the Pais-Uhlenbeck oscillator, showing that in a parameter region the $(1,1)$-sector, traditionally discarded, becomes a well-defined quantum system. If correct, the construction offers a consistent re-interpretation of HTDTs and a new way to build ghost-free quantum models.","feed_headline":"Non-unitary maps turn ghostly quantum states normalisable","feed_subtitle":"The bounded-energy ghost sector gains normalisable states under a new inner product.","key_machinery":"The central object is the non-unitary similarity transformation $\\eta = \\eta_2 \\eta_1 \\eta_0$ built from three explicitly defined operators: $\\eta_0 = \\exp(-\\delta x^2/2 - \\lambda y^2/2)$, $\\eta_1 = \\exp(\\kappa p_x^2/2 + \\xi p_y^2/2)$, and $\\eta_2 = \\exp(\\mu p_x p_y + \\tau x y)$. These are exponentials of quadratic generators; their adjoint actions are computed via the Baker-Campbell-Hausdorff formula, and $\\eta_2$ is factorised into $SU(2)$ group elements through a Gauss decomposition. The transformation maps the Hermitian ghostly Hamiltonian $h_0$ to a Hermitian partner $h_3$, preserving the spectrum, while the new metric $\\rho = \\eta^\\dagger \\eta$ defines the inner product in which the formerly non-normalisable eigenstates become normalisable. The parameter choices in $\\eta_1$ and $\\eta_2$ are tuned to eliminate non-Hermitian terms and to render the transformed ground state normalisable.","core_discovery":"The central claim is that the sector of the ghostly Hamiltonian $h_0 = p_x^2 - p_y^2 + \\nu^2 x^2 + \\Omega y^2 + gxy$ with $\\epsilon = \\eta = 1$, which has a spectrum bounded from below but non-normalisable eigenstates under the standard $L^2$ inner product, becomes fully physical after the sequence of non-unitary transformations $\\eta_0 \\eta_1 \\eta_2$. Specifically, the transformed Hamiltonian $h_3 = \\eta_2 \\eta_1 \\eta_0 \\, h_0 \\, \\eta_0^{-1} \\eta_1^{-1} \\eta_2^{-1}$ is Hermitian and isospectral to $h_0$, and with the metric $\\rho = \\eta_0^\\dagger \\eta_1^\\dagger \\eta_2^\\dagger \\eta_2 \\eta_1 \\eta_0$ the eigenstates, in particular the Gaussian ground state, are normalisable. The paper exhibits explicit parameter choices (e.g., $\\nu=4$, $\\Omega=-2$, $g=3$, $|\\lambda|>2.13322$) where the normalisability condition $\\alpha>0$, $\\beta>0$, $alpha\\beta-\\gamma^2>0$ holds for the $(1,1)$-sector, while the spectrum remains bounded from below. The authors present this as a proof of concept that the ghost problem in HTDTs can be circumvented by re-interpreting the Hilbert-space structure, rather than discarding the offending sector.","pith_inferences":["If the metric $\\rho$ is indeed positive, the construction can be read as a choice of a new physical Hilbert space for the ghostly theory; the non-uniqueness of $\\eta$ might then be fixed by requiring a second observable to be Hermitian, a point the authors flag but do not resolve.","The Gaussian ansatz suggests the method could extend to HTDTs whose ground states are Gaussian, but the proof of concept does not establish how the transformation generalises to interacting field theories.","A sharper test of the method would be to examine the domain issues of the unbounded operators $\\eta_1$ and $\\eta_2$; the paper does not prove they are well-defined on $L^2(\\mathbb{R}^2)$, so the preservation of the spectrum is an assumption that could fail.","Numerically scanning the full parameter space could determine whether the normalisable region $R_1$ is as large as suggested and whether similar regions exist for other branch choices."],"forward_implications":["The $(1,1)$-sector of the ghostly model, previously deemed unphysical, admits a consistent quantum description with a positive-definite inner product and a spectrum bounded from below.","The construction yields an isospectral Hermitian partner for a Pais-Uhlenbeck-related model, so the energy spectrum of the physical theory is unchanged while the states become normalisable.","The method extends pseudo/quasi-Hermitian quantum mechanics by mapping between two Hermitian Hamiltonians, not from a non-Hermitian one, broadening the class of tractable ghost problems.","For vanishing coupling $g=0$, the transformed Hamiltonian becomes a sum of two harmonic oscillators in the relevant parameter region, making the physical interpretation explicit.","The framework suggests a route to ghost-free quantisation of more general higher time-derivative theories, including field-theoretic extensions."],"supporting_citations":[{"why":"Supplies the complete eigensystem of $h_0$, including the sector classification with bounded spectrum and non-normalisable eigenstates that the paper sets out to fix.","marker":"[11]"},{"why":"Establishes that the model can be mapped to the Pais-Uhlenbeck oscillator in parts of parameter space, connecting the construction to higher time-derivative theories.","marker":"[30]"},{"why":"Defines the prototype Pais-Uhlenbeck oscillator that the ghostly model is related to and that motivates the ghost problem.","marker":"[1]"},{"why":"Introduces PT-symmetric quantum mechanics whose non-unitary map idea inspires the present similarity-transformation method.","marker":"[31]"},{"why":"Provides the pseudo-Hermitian framework in which a similarity transformation defines a new metric and inner product, the core mechanism used here.","marker":"[32]"},{"why":"Underlies the statement that the metric is not unique and that a second observable is needed to resolve the ambiguity, which the authors apply to their construction.","marker":"[35]"},{"why":"Supplies the $SU(2)$ Gauss decomposition used to factorise $\\eta_2$ and compute the transformed wavefunctions.","marker":"[37]"}],"fun_headline_variants":["Non-unitary maps turn ghostly quantum states normalisable","Ghost-free quantisation via non-unitary similarity transformations","Bounded spectrum and normalisable states: ghost problem solved?","Non-unitary transform fixes ghost states in higher-derivative theories"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument assumes the non-unitary operators $\\eta_1$ and $\\eta_2$ are well-defined invertible maps on the quantum state space for the parameter ranges used, so that the similarity transformation truly preserves the spectrum and the new metric $\\rho = \\eta^\\dagger \\eta$ is positive definite; the paper does not prove these domain and positivity properties.","fun_headline_variants_meta":{"raw":{"variants":["Non-unitary maps turn ghostly quantum states normalisable","Ghost-free quantisation via non-unitary similarity transformations","Bounded spectrum and normalisable states: ghost problem solved?","Non-unitary transform fixes ghost states in higher-derivative theories"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000638,"raw_usage":{"total_tokens":2979,"prompt_tokens":1026,"completion_tokens":1953,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":642,"completion_tokens_details":{"reasoning_tokens":1885}},"tokens_in":642,"tokens_out":1953,"duration_ms":15868,"temperature":1.0,"reasoning_tokens":1885,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T22:26:24.062852+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Evaluate the new inner product $\\langle \\phi_0 | \\rho | \\phi_0 \\rangle$ for the parameter set in figure 2 ($\\nu=4$, $\\Omega=-2$, $g=3$, $\\lambda$ in the region $|\\lambda|>2.13$). If it diverges or is not strictly positive, or if the metric $\\rho = \\eta_0^\\dagger \\eta_1^\\dagger \\eta_2^\\dagger \\eta_2 \\eta_1 \\eta_0$ has a negative eigenvalue, the claimed normalisability fails. Alternatively, check directly whether $e^{\\xi p_y^2/2}$ maps $L^2(\\mathbb{R}^2)$ into $L^2(\\mathbb{R}^2)$ for $\\xi>0$ in that region.","supporting_citations":[{"cited_title":"Fring, T","cited_arxiv_id":null,"evidence_quote":"Supplies the complete eigensystem of $h_0$, including the sector classification with bounded spectrum and non-normalisable eigenstates that the paper sets out to fix."},{"cited_title":"Lie symmetries and ghost-free representations of the Pais-Uhlenbeck model","cited_arxiv_id":"2505.07869","evidence_quote":"Establishes that the model can be mapped to the Pais-Uhlenbeck oscillator in parts of parameter space, connecting the construction to higher time-derivative theories."},{"cited_title":"Pais and G","cited_arxiv_id":null,"evidence_quote":"Defines the prototype Pais-Uhlenbeck oscillator that the ghostly model is related to and that motivates the ghost problem."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Introduces PT-symmetric quantum mechanics whose non-unitary map idea inspires the present similarity-transformation method."},{"cited_title":"Mostafazadeh, Pseudo-Hermitian Representation of Quantum Mechanics, Int","cited_arxiv_id":null,"evidence_quote":"Provides the pseudo-Hermitian framework in which a similarity transformation defines a new metric and inner product, the core mechanism used here."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Underlies the statement that the metric is not unique and that a second observable is needed to resolve the ambiguity, which the authors apply to their construction."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the $SU(2)$ Gauss decomposition used to factorise $\\eta_2$ and compute the transformed wavefunctions."}],"review_version":1}