{"id":"294d4eba-94fb-4152-94be-405b650878f8","arxiv_id":"2505.00791","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"Two quasi-Hermitian observables are mutually compatible exactly when the matrix connecting their eigenvector frames can be scaled by positive diagonal matrices into a unitary matrix.","lead":"This paper gives a recipe for telling whether two non-Hermitian operators can share one correct inner product and both act as genuine observables. The recipe is a concrete matrix condition, and the authors exhibit a pair with real measurement values that fails it.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Lemma 2 is sound, but Section 4.2's constraint count treats complex off-diagonal equations as real ones, so the paper's N=2 'parametric freedom survives' claim is false; the practical compatibility criteria misfire.","rationale":"I focus on the practical criteria rather than Lemma 2 because Lemma 2 is mathematically correct within the paper's stated non-degenerate finite-dimensional assumptions. The paper's advertised 'criteria' are partly false: the count in Eq. (16) undercounts real constraints, and the N=2 discussion ignores positivity. The explicit counterexample shows the claimed N=2 ease is not reliable. The reader already identified this as a flaw and conditioned the verdict; my concern corroborates that. I do not see an internal flaw in Lemma 2, so the verdict should remain CONDITIONAL, not move to REJECT. The non-degeneracy restriction is a genuine boundary but is explicitly stated; the count error is an in-scope error in the criteria. Hence partial agreement with the reader's weakest_assumption.","tokens_in":11531,"tokens_out":22693,"duration_ms":237750,"concrete_test":"For the pair A1=diag(1,-1), A2=[[1,2],[0,-1]] with Ω1=I and Ω2=[[1,1],[0,1]], recompute Eq. (16): M† c1^2 M = [[x1,-x1],[-x1,x1+x2]], so the off-diagonal condition forces x1=0, excluding positive c1. Independently solve A1†Θ=ΘA1 and A2†Θ=ΘA2: the first forces Θ diagonal, the second forces Θ11=0, so no positive shared metric exists. This settles that the N=2 'straightforward/parametric freedom survives' statement is wrong; Lemma 2 itself is not the failure point.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 4.2's practical criterion, not Lemma 2 itself, is where the paper is least secure. After eliminating c2 via Eq. (15), the paper counts N(N-1)/2 remaining constraints in Eq. (16) on the N positive variables in c1. This count treats each complex off-diagonal entry as a single real equation. In fact each (M† c1^2 M)_{mn}=0 for m<n is a complex equation, hence two real conditions, and positivity of c1 is an additional inequality constraint that counting alone cannot certify. The consequence is visible at N=2: the paper says there is 'just a single constraint (16) imposed upon two real parameters' and that 'the parametric freedom survives'. That is false. Take A1=diag(1,-1), A2=[[1,2],[0,-1]], with Ω1=I and Ω2=[[1,1],[0,1]] so Ω2 A2 Ω2^{-1}=diag(1,-1). Then M=[[1,-1],[0,1]], and M† c1^2 M=[[x1,-x1],[-x1,x1+x2]]. Eq. (16) forces x1=0, so no positive diagonal c1 exists; indeed the pair is incompatible because A1 Hermitian forces a shared Θ to be diagonal and A2 then forces Θ11=0. The paper's count would suggest a one-parameter family of positive solutions. The same miscount shifts the claimed generic-incompatibility threshold: comparing N(N-1) real constraints with N variables makes generic incompatibility start already at N=3, not N≥4. Lemma 2's unitary-scaling equivalence is unaffected, but the paper's advertised criteria for when compatibility is easy or generic are not reliable.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript addresses the problem of when two bounded non-Hermitian operators A1 and A2 on a finite-dimensional Hilbert space, with real spectra, can be observables of a single quasi-Hermitian quantum theory, i.e. admit a common positive metric Θ satisfying A_j^† Θ = Θ A_j. The author's strategy is to diagonalize each A_j via a Dyson map Ω_j and to parametrize all metrics for A_j by Θ_j(c_j) = Ω_j† c_j^2 Ω_j with positive diagonal c_j. Lemma 1 states that compatibility is equivalent to the existence of c1,c2 with Θ1(c1)=Θ2(c2); Lemma 2 restates this as unitarity of U(c1,c2)=c1 Ω1 Ω2^{-1} c2^{-1}. The paper then discusses counts of constraints, claims generic incompatibility for N≥4 and surviving parametric freedom at N=2, and illustrates the results with a 3x3 pair with real spectra that is shown to be incompatible.","tokens_in":11713,"tokens_out":13518,"duration_ms":126915,"significance":"The core algebraic reformulation is attractive and, under the stated non-degeneracy assumption, correct: Lemma 1 follows from the standard characterization of metrics for diagonalizable operators, and Lemma 2's equivalence to Eq. (15) is verified by direct multiplication. The 3x3 example is a concrete demonstration of incompatibility, and the paper's Lemma 2, if properly scoped, would provide a useful criterion for simultaneous quasi-Hermiticity. However, the paper's practical counting criterion contains a load-bearing error that affects the claimed threshold and the N=2 discussion, and Lemma 2 as stated is missing the non-degeneracy hypothesis. With those fixed, the criterion would be a useful contribution.","major_comments":[{"comment":"The constraint count is incorrect. Each off-diagonal equation (M† c1² M)_{mn}=0 for m<n is one complex equation, i.e. two real equations, and the diagonal entries of M† c1² M must be positive; the paper counts N(N-1)/2 complex constraints and ignores positivity. The correct real count is N(N-1) equations for N positive variables, so generic incompatibility already starts at N=3, not N≥4. The claim that at N=2 'the parametric freedom survives' is false: with A1=diag(1,-1), A2=[[1,2],[0,-1]], Ω1=I and Ω2=[[1,1],[0,1]], one obtains M=[[1,-1],[0,1]] and M† c1² M=[[x1,-x1],[-x1,x1+x2]], so Eq. (16) forces x1=0 and no positive c1 exists. The pair is indeed incompatible, since for non-degenerate A1 any shared Θ must be diagonal and then fails for A2. Thus the practical criterion as stated misfires already in the minimal case.","section":"Section 4.2, Eq. (16) (and its use in Section 5)"},{"comment":"The statement is missing the non-degeneracy hypothesis that is used to make the c_j parametrization exhaustive in Lemma 1. For degenerate spectra, positive operators commuting with a_j are not only diagonal, so the equivalence can fail. Example: A1=I and A2=[[1,2],[0,-1]] are compatible—A1 is Hermitian and A2 is quasi-Hermitian with shared metric Θ=[[1,1],[1,2]]. Fix Ω1=I and Ω2=[[1,1],[0,1]] as above; then M=[[1,-1],[0,1]] and c2²=M†c1²M forces the off-diagonal entry -x1 to vanish, so no positive diagonal c1 exists. The shared metric is not of the form Ω1†c1²Ω1 with diagonal c1, showing that the 'only if' direction of Lemma 2 requires the non-degeneracy assumption (or a block-diagonal generalization).","section":"Lemma 2 (Section 4.2)"}],"minor_comments":[{"comment":"'c1 and c1' should read 'c1 and c2'.","section":"Section 4.1, above Lemma 1"},{"comment":"'aadditional' is a typo for 'additional'.","section":"Section 5"},{"comment":"'Ai and A2' should read 'A1 and A2'.","section":"Section 6"},{"comment":"The phrase 'Both sides of this relation are Hermitian matrices which have two parts' is vague; it would be clearer to state that Eq. (15) is equivalent to the vanishing of all off-diagonal entries of M† c1² M together with positivity of its diagonal entries.","section":"Section 4.2"},{"comment":"The abstract's unqualified 'criteria of existence' should mention the finite-dimensional, non-degenerate setting in which the criteria are proved, since Section 3.1 explicitly restricts to N<∞ and Section 6 warns that N=∞ is nontrivial.","section":"Abstract and Section 3.1"}],"recommendation":"major_revision","confidential_remarks":"The main mathematical idea is sound under the non-degeneracy assumption, and the paper is likely publishable after revision. The two substantive issues are the incorrect constraint count in Section 4.2 and the missing non-degeneracy hypothesis in Lemma 2; both are local and repairable. The paper leans heavily on the author's own prior work for motivation and notation; the editor may wish to encourage a more balanced treatment of the broader pseudo-Hermitian operator literature."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The one thing worth knowing about this paper: the exact, non-perturbative compatibility criterion in Lemma 2 is correct, and it fills a real gap left by the earlier perturbative no-go. The paper shows that two finite-dimensional quasi-Hermitian observables with real, non-degenerate spectra are compatible iff the frame-relation matrix M = Ω1 Ω2^{-1} can be diagonal-scaled to unitarity. That is a clean and useful reformulation, and the accompanying reduction to linear equations (15)–(16) is practical. The explicit 3x3 pair that fails the test, even though both operators have real spectra, is a valuable counterexample; I spot-checked the eigenvector relations and they do satisfy Eq. (19).\n\nThe problems are concentrated in Section 4.2. The count of constraints in (16) treats each complex off-diagonal equality as a single real condition. Each of those is two real conditions. That is not a pedantic distinction: it breaks the N=2 claim. The paper says a single constraint on two real parameters leaves parametric freedom. But for the counterexample with A1 = diag(1,-1), A2 = [[1,2],[0,-1]], the off-diagonal equation forces x1 = 0, so no positive diagonal c1 exists and the pair is incompatible. The same miscount shifts the generic-incompatibility threshold from N ≥ 4 to N ≥ 3. So Lemma 2 survives, but the paper's advertised guidance about when compatibility is easy or generic is unreliable.\n\nMinor points: the paper relies on non-degenerate spectra and finite dimension, and it says so, so those are limitations rather than hidden flaws. The example's decisive algebra is asserted rather than shown, but it checks out. The text is discursive and has typos, but it is readable.\n\nWho is this for? People building quasi-Hermitian models with a second observable—cosmology, PT-symmetric contexts, model builders. They should read Lemma 2 and the example, and ignore the threshold statements until corrected. The core result is serious enough to deserve referee time, but not without requiring a fix to the constraint counting and the N=2 discussion. My recommendation: send it to peer review, with the practical criteria section flagged for major revision.","headline":"The exact compatibility criterion (Lemma 2) is sound and is the real takeaway, but the paper's practical criteria in Section 4.2 mis-count constraints and the N=2 'parametric freedom survives' claim is false.","tokens_in":12453,"tokens_out":2248,"would_cite":true,"duration_ms":21460,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["81Q12","47B50"],"pacs":[],"model":"deepseek-v4-flash","headline":"Two non-Hermitian observables are compatible exactly when a diagonal rescaling makes their frame link unitary.","keywords":["quasi-Hermitian observables","shared inner-product metric","non-Hermitian quantum mechanics","Dyson map","matrix compatibility criterion","real spectra","finite-dimensional Hilbert space","positive diagonal scaling"],"falsifier":"Solve the linear system (14) for the paper's $3\\times3$ example at $s=1/2$, $a=1/2$: the unique solution forces $x_1=-\\rho/12+i\\sqrt3\\,\\rho/12$, $x_2=\\rho/9$, $x_3=\\rho$, with $\\rho$ a free complex parameter, so the $c_j$ cannot be real positive; that is direct evidence against compatibility. To refute the general criterion, one would need either a pair whose only scaling solution is non-positive yet which admits a shared metric outside the $\\Theta_j=\\Omega_j^\\dagger c_j^2\\Omega_j$ family, or a pair that passes the unitary-scaling test but still has no shared metric.","tokens_in":11112,"feed_emoji":"⚛️","tokens_out":9532,"duration_ms":88122,"temperature":0.7,"pith_summary":"This paper asks when two non-Hmitian operators with real spectra can both be observables of one quantum system. In quasi-Hermitian quantum mechanics, an observable need only be Hermitian after a change of physical inner product, but a single metric must serve every observable. The paper proves that for a pair of $N\\times N$ candidates with real non-degenerate spectra, a shared metric exists if and only if positive diagonal matrices $c_1,c_2$ can rescale the matrix $M=\\Omega_1\\Omega_2^{-1}$ into a unitary matrix $U=c_1Mc_2^{-1}$. It exhibits a three-by-three pair, one real asymmetric and one complex symmetric, whose spectra are real yet which fails the test, so no metric makes them simultaneously quasi-Hermitian. The payoff is that the result reduces a question about operators and metrics to a finite algebraic test, and it warns that real spectra alone do not make a set of observables mutually compatible.","feed_headline":"Two non-Hermitian observables: shared metric means one scaling test","feed_subtitle":"In quasi-Hermitian quantum mechanics, two real-spectrum operators co-exist only if a rescaled matrix becomes unitary.","key_machinery":"The load-bearing object is the frame-relation matrix $M=\\Omega_1\\Omega_2^{-1}$ between the two diagonalizing Dyson maps, together with the positive diagonal matrices $c_1,c_2$ that encode the arbitrary normalization of each eigenvector. Since each physical metric for $A_j$ is $\\Theta_j=\\Omega_j^\\dagger c_j^2\\Omega_j$, requiring one shared metric is the same as asking for $c_1,c_2$ with $\\Omega_1^\\dagger c_1^2\\Omega_1=\\Omega_2^\\dagger c_2^2\\Omega_2$; this equality is equivalent to $M^\\dagger c_1^2M$ being diagonal and, in turn, to $c_1Mc_2^{-1}$ being unitary. The unitary form is the paper's compact criterion and the practical route to constructing compatible pairs from a chosen unitary matrix.","core_discovery":"Let $A_1$ and $A_2$ be two $N\\times N$ matrices with real, non-degenerate spectra, and choose Dyson maps $\\Omega_1,\\Omega_2$ such that $a_j=\\Omega_j A_j\\Omega_j^{-1}$ is real diagonal. Set $M=\\Omega_1\\Omega_2^{-1}$. The paper's central claim is that $A_1$ and $A_2$ are compatible—there exists a positive-definite metric $\\Theta$ with $A_j^\\dagger\\Theta=\\Theta A_j$ for $j=1,2$—if and only if there are positive diagonal matrices $c_1,c_2$ for which $U=c_1Mc_2^{-1}$ is unitary. Equivalently, the off-diagonal entries of $M^\\dagger c_1^2 M$ must vanish, with its diagonal entries defining $c_2^2$. The proof runs through the residual normalization freedom of the eigenvector matrices: every eligible metric for $A_j$ has the form $\\Theta_j=\\Omega_j^\\dagger c_j^2\\Omega_j$, so a shared metric is exactly a choice of $c_1,c_2$ making these two expressions equal. The paper's worked $3\\times3$ example has real spectra but forces the scaling parameters to be non-real, demonstrating an incompatible pair.","pith_inferences":["A numerical route the paper does not develop: for a given $M$, search over positive diagonal $c_1,c_2$ by reducing the deviation of $c_1Mc_2^{-1}$ from unitarity; the same algorithm would serve as a practical compatibility test for larger $N$.","The parameter count suggests a rigidity principle the paper does not state: when compatibility holds for $N\\geq3$, the shared metric is generically overdetermined rather than freely chosen, so adding observables fixes the physical inner product almost uniquely.","A testable extension: for $N=\\infty$ approximations by finite truncations, the scaling parameters' behavior under refinement could indicate which infinite-dimensional pairs are compatible, although the paper warns the infinite limit is nontrivial."],"forward_implications":["Compatibility of two quasi-Hermitian observables with real non-degenerate spectra can be decided by the linear constraints (16) on $c_1^2$; no search over all metrics is needed.","At $N\\geq4$, the $N(N-1)/2$ off-diagonal constraints outnumber the $N$ free parameters, so a generic pair of non-Hermitian candidates is incompatible.","At $N=2$ there is only one constraint on two parameters, so compatible pairs remain typical within the class considered.","For $K$ observables the same test requires $K-1$ equations of type (16) on $N$ parameters, so adding observables makes shared metrics rapidly more restrictive.","The explicit $3\\times3$ example shows that each operator having a real spectrum does not imply the pair is jointly observable; hidden Hermiticity of the individual operators is insufficient."],"supporting_citations":[{"why":"It supplies the quasi-Hermitian framework and the requirement that all observables satisfy $A_j^\\dagger\\Theta=\\Theta A_j$.","marker":"[2]"},{"why":"It states the pseudo-Hermitian formulation that the paper uses to define observability and the reality of spectra.","marker":"[5]"},{"why":"It is the predecessor no-go study showing that arbitrary pairs of non-Hermitian observables need not share any metric, and it motivates the compatibility question.","marker":"[7]"},{"why":"It introduces the Dyson map and the factorization $\\Theta=\\Omega^\\dagger\\Omega$ used to represent metrics.","marker":"[21]"},{"why":"It provides the residual normalization freedom $\\Omega_j^\\dagger\\to\\Omega_j^\\dagger c_j$ and the family $\\Theta_j=\\Omega_j^\\dagger c_j^2\\Omega_j$ on which Lemma 1 is built.","marker":"[22]"},{"why":"It fixes the class of admissible metrics as positive, bounded, and with bounded inverse.","marker":"[3]"}],"fun_headline_variants":["Two observables share a metric only if scaling makes M unitary","Quasi-Hermitian pair incompatible unless scaling yields unitary matrix","Compatibility of two observables hinges on unitary rescaling test","Shared metric criterion: rescale to make the intertwiner unitary","Two non-Hermitian operators compatible iff scaling is unitary"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The if-and-only-if claim rests on treating positive rescaling of each eigenvector as the only metric freedom, which requires real, non-degenerate spectra; with degenerate eigenvalues the admissible metrics form a strictly larger family and the diagonal-scaling test stops being exhaustive.","fun_headline_variants_meta":{"raw":{"variants":["Two observables share a metric only if scaling makes M unitary","Quasi-Hermitian pair incompatible unless scaling yields unitary matrix","Compatibility of two observables hinges on unitary rescaling test","Shared metric criterion: rescale to make the intertwiner unitary","Two non-Hermitian operators compatible iff scaling is unitary"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00018,"raw_usage":{"total_tokens":1312,"prompt_tokens":960,"completion_tokens":352,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":576,"completion_tokens_details":{"reasoning_tokens":267}},"tokens_in":576,"tokens_out":352,"duration_ms":4171,"temperature":1.0,"reasoning_tokens":267,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-16T04:39:11.341910+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Solve the linear system (14) for the paper's $3\\times3$ example at $s=1/2$, $a=1/2$: the unique solution forces $x_1=-\\rho/12+i\\sqrt3\\,\\rho/12$, $x_2=\\rho/9$, $x_3=\\rho$, with $\\rho$ a free complex parameter, so the $c_j$ cannot be real positive; that is direct evidence against compatibility. To refute the general criterion, one would need either a pair whose only scaling solution is non-positive yet which admits a shared metric outside the $\\Theta_j=\\Omega_j^\\dagger c_j^2\\Omega_j$ family, or a pair that passes the unitary-scaling test but still has no shared metric.","supporting_citations":[{"cited_title":"G.; Geyer, H","cited_arxiv_id":null,"evidence_quote":"It supplies the quasi-Hermitian framework and the requirement that all observables satisfy $A_j^\\dagger\\Theta=\\Theta A_j$."},{"cited_title":"Problem of the co- existence of several non-Hermitian observables in PT-symmetric q uantum mechanics","cited_arxiv_id":null,"evidence_quote":"It is the predecessor no-go study showing that arbitrary pairs of non-Hermitian observables need not share any metric, and it motivates the compatibility question."},{"cited_title":"On the role of the normalization factors κn and of the pseudo-metric P in crypto-Hermitian quantum models","cited_arxiv_id":null,"evidence_quote":"It provides the residual normalization freedom $\\Omega_j^\\dagger\\to\\Omega_j^\\dagger c_j$ and the family $\\Theta_j=\\Omega_j^\\dagger c_j^2\\Omega_j$ on which Lemma 1 is built."},{"cited_title":"On the metric operator for the imagina ry cubic oscillator","cited_arxiv_id":null,"evidence_quote":"It fixes the class of admissible metrics as positive, bounded, and with bounded inverse."}],"review_version":1}