{"id":"dabb986c-ed30-468a-a33c-b1eb8910f861","arxiv_id":"2608.04850","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":5.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"For non-Hermitian Hamiltonians, the Arnoldi diagonal and subdiagonal coefficients are the Flaschka variables of the finite two-dimensional Toda lattice, and their squares give both the Fubini-Study metric and the Berry curvature of the Krylov subspaces.","lead":"This paper proves that the diagonal and subdiagonal entries of the Arnoldi (Hessenberg) form of a non-Hermitian quantum operator obey the equations of the two-dimensional Toda lattice, a classical integrable system. It then shows these coefficients measure the Fubini-Study metric and Berry curvature of the generated Krylov subspaces and builds an exact counterdiabatic driving term from the same data.","discovery_kind":"unification","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No significant objection identified: the Toda–Arnoldi correspondence and the geometric identifications survive cross-checking; the reader's stated conditions are minor and do not affect the central claim.","rationale":"The reader's weakest assumption is the cyclic-seed plus exponential-deformation hypothesis; I agree that this is the load-bearing premise, but it is standard, explicitly scoped, and preserved under the flow, so it does not constitute a flaw. My own check of the core algebra confirms the proof: the leading coefficients in Eq. (15) are independent of the upper Hessenberg entries because a path from site 0 to site l-1 in l steps cannot contain any downward step, which is exactly why the diagonal and subdiagonal Toda sector closes without determining h_{m,n}. The Desnanot–Jacobi identity applies to the mixed-derivative moment matrix because its entries satisfy d_z m_{i,j} = m_{i+1,j} and d_{bar z} m_{i,j} = m_{i,j+1}, with the sign factors canceling in the determinant. The counterdiabatic section is a frame-transformation identity: i d_s(U^dagger e^{-iHs} U_0) = (A+G) U^dagger e^{-iHs} U_0, and the no-transition result follows from i|dot r_n> = G|r_n>. The only remaining items are presentation-level: deriving the three-site Chern integrals explicitly and noting that Appendix E already contains the general breakdown theorem. Neither changes the verdict.","tokens_in":24853,"tokens_out":39602,"duration_ms":419749,"concrete_test":"Independently recompute the main theorem numerically: take a random 4x4 non-Hermitian H with a cyclic seed, build the Arnoldi coefficients alpha_n, beta_n on a grid of z, and compare with the Flaschka variables computed from the Gram determinants tau_n = det(K_n^dagger K_n). If the differences are below numerical precision at all grid points and beta_n^2 matches the z,bar-z derivative of ln tau_n, the central claim is confirmed. Optionally integrate b_2^2 in the three-site model over CP^1 to verify Table I.","verdict_should_be":"UNCHANGED","load_bearing_attack":"I looked for a load-bearing flaw in the main theorem (Eq. 13) and the b_n^2 geometry (Eqs. 23, 28). The construction rests on the cyclic-seed plus e^{-zH} deformation hypothesis; that hypothesis is preserved for every finite z because K_D(z) = e^{-zH} K_D(0) has full rank when K_D(0) does, so the claimed cyclic region is well defined and the scope is explicit. The determinant identities (16)–(17), the QR proof in Appendix D, and the Lax subdiagonal equation independently imply the same identification a_n = alpha_n, b_n = beta_n. The two points flagged by the reader do not rise to load-bearing: Table I follows by integrating Eqs. (34)–(35) after homogenizing the Plücker representative (degree-2 rational curve, giving C_1 = C_2 = -2), and the breakdown statement is already proven generally in Appendix E: at breakdown dimension d_K, Phi_{d_K} is constant so its metric and curvature vanish, while Phi_n for n > d_K is undefined because rank K_n < n. Thus no correction to the central claim is needed.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript proves a finite-dimensional Toda–Arnoldi correspondence. For a fixed, generally non-Hermitian Hamiltonian H and a cyclic seed |ψ(0)>, the holomorphic family |ψ(z)>=e^{-zH}|ψ(0)> is used to build Krylov Gram determinants τ_n(z,\\bar z). The authors show that these τ_n satisfy the Hirota bilinear identity of the two-dimensional Toda lattice, and prove the Main Theorem (Eq. 13): the Arnoldi diagonal entries α_n and subdiagonal entries β_n coincide with the Toda Flaschka variables a_n and b_n. They then show that b_n^2 is the Fubini–Study metric component and (up to sign) the Berry curvature of the n-th holomorphic Krylov subspace, derive the Stokes relation (26), compute Chern numbers in a three-site non-Hermitian model including Arnoldi breakdown, and construct a Hermitian tridiagonal moving-frame generator G_γ that gives exact isospectral transport and cancels transitions between instantaneous eigenspaces of the Arnoldi matrix.","tokens_in":24892,"tokens_out":43129,"duration_ms":430376,"significance":"The result, if accepted, is significant: it extends the known Toda–Lanczos relation to general non-Hermitian Arnoldi reduction, gives the diagonal and subdiagonal Arnoldi data a precise integrable-system and geometric meaning without fixing the rest of the upper Hessenberg matrix, and supplies a concrete three-site model with quantized Chern numbers and a breakdown analysis. The paper is largely self-contained: the main theorem is proved directly and again via QR factorization in Appendix D, and the key identities (16), (17), (26), (54)–(55) and the Table I entries are explicit and check out. The construction has no free parameters, and its scope (cyclic seed, exponential holomorphic deformation) is stated precisely. I found no load-bearing flaw in the central derivation.","major_comments":[],"minor_comments":[{"comment":"The typeset Table I has merged columns (e.g., \"0 2−1 0\" and \"3−2−2\"), making the entries for d_K, C1, and C2 ambiguous; please typeset the table with clear column separation and with \"undefined\" as a distinct entry.","section":"III.C, Table I"},{"comment":"Although the text states that Eq. (35) holds for 0≤θ<π/2, the right-hand side has the finite limit 1 as θ→π/2. A sentence explaining that this limit is not the value of b_2^2 because the second Arnoldi vector does not exist at θ=π/2 would remove an apparent contradiction.","section":"III.C, Eq. (35)"},{"comment":"The counterdiabatic interpretation is formulated in the moving Arnoldi frame; in the laboratory frame the evolution generated by A+G is simply free evolution under H. Please state this scope explicitly when using the term \"counterdiabatic driving,\" so that readers do not infer that a modified physical Hamiltonian is being implemented in the original Hilbert space.","section":"IV, Eqs. (44)–(46)"},{"comment":"Please fix the orientation convention for dz∧d\\bar z on Σ (e.g., dz∧d\\bar z = -2i dx∧dy) so that the nonpositivity of C_n and the numerical values in Table I are unambiguous.","section":"III.A, Eq. (24)"},{"comment":"For n=1 the product over m=1 to 0 is empty; state the empty-product convention explicitly for clarity.","section":"II.B, Eq. (16)"}],"recommendation":"minor_revision","confidential_remarks":"The manuscript is technically sound as far as I can verify; the main theorem and the geometric identifications survive cross-checking. The requested changes are editorial. The acknowledgment of AI-assisted editing is disclosed and does not affect my assessment."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Bottom line: the central claim is solid. The Toda–Arnoldi identification is proven twice, the b_n^2 = metric = curvature identifications go through, and the stress-test note holds up. The result is genuinely new: for non-Hermitian H, the diagonal and subdiagonal Arnoldi coefficients close a two-dimensional Toda sector even though the rest of the Hessenberg matrix does not participate, and the squared subdiagonal coefficient carries both Fubini–Study metric and Berry curvature content. The counterdiabatic generator in the Arnoldi frame is a useful addition, and the Hermitian limit correctly reduces to known Toda–Lanczos and Okuyama–Takahashi results, which gives me confidence in the overall construction.\n\nWhat the paper does well: it is explicit and largely self-contained. The tau functions are defined from Krylov Gram determinants, the Flaschka variables are defined from those tau functions, and the equality with the independently defined Arnoldi coefficients is shown directly and again via QR factorization. There is no circularity. Appendix E gives the geometric details, including a direct Arnoldi-basis derivation of the metric identity. The Stokes relation (26) and the circular-mean potential-theory appendix are nice. The author is not leaning on self-citation; the key references are standard and appropriate.\n\nThe soft spots are minor. The Chern numbers in Table I are asserted without derivation. The stress-test note confirms they follow by integrating (34)–(35) after homogenizing the Plücker representative, but a referee should ask the author to put that step in the text. The breakdown discussion is framed around the three-site model, which reads as an example; Appendix E actually states the general result (Phi_{d_K} is constant, Phi_n undefined for n > d_K), so the main text should point there when discussing breakdown. Finally, Sec. IV assumes a real path in the cyclic region; that is stated, but the smoothness and no-breakdown assumptions could be assembled into one explicit hypothesis before the Lax equation is used. None of these affect the load-bearing algebra.\n\nWho this is for: people working on Krylov expansions, non-Hermitian quantum dynamics, operator growth, and counterdiabatic driving. It deserves a serious referee. I would send it to review, with the request that Table I be substantiated and the breakdown section be tightened.","headline":"Solid and new: the Toda–Arnoldi identification holds up; send to review with a request to substantiate the Chern-number table and tighten the breakdown section.","tokens_in":25639,"tokens_out":3101,"would_cite":true,"duration_ms":34082,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proves that in Arnoldi reduction of a non-Hermitian Hamiltonian, the diagonal and subdiagonal entries are exactly the Flaschka variables of a two-dimensional Toda lattice.","keywords":["Toda lattice","Arnoldi method","Krylov subspaces","non-Hermitian Hamiltonian","Fubini-Study metric","Berry curvature","counterdiabatic driving","tau function"],"falsifier":"Take a non-Hermitian $H$ and a cyclic seed, and for many values of $z$ compute both the Arnoldi coefficients $\\alpha_n,\\beta_n$ and the Flaschka variables from $\\tau_n=\\det(K_n^\\dagger K_n)$; if any $\\alpha_n$ differs from $-\\partial_z\\ln(\\tau_{n+1}/\\tau_n)$ or any $\\beta_n$ differs from $\\sqrt{\\tau_{n+1}\\tau_{n-1}/\\tau_n^2}$ while all $\\tau_n>0$, the main theorem is false. A sharper check is to approach a point where Arnoldi breakdown lowers the Krylov dimension below $n$ and verify that $b_n^2$ either vanishes for the invariant subspace or that the Fubini–Study metric ceases to exist, matching the paper's three-site Chern-number prediction.","tokens_in":24458,"feed_emoji":"⚡️","tokens_out":7964,"duration_ms":77725,"temperature":0.7,"pith_summary":"The paper establishes a two-dimensional Toda–Arnoldi correspondence: for a fixed finite-dimensional, generally non-Hermitian Hamiltonian and a holomorphically deformed cyclic seed state, the Krylov Gram determinants are tau functions of the finite two-dimensional Toda lattice, and the diagonal and subdiagonal entries of the Arnoldi upper-Hessenberg matrix equal the lattice's Flaschka variables. It follows that the Toda equations close on this sector without constraining the remaining upper-Hessenberg entries. The same squared subdiagonal coefficient also gives the Fubini–Study metric and Berry curvature of the holomorphic Krylov-subspace map. Along real paths, the Arnoldi-frame connection yields a Hermitian tridiagonal generator that preserves the spectrum exactly and, in the nondegenerate diagonalizable case, cancels transitions between instantaneous eigenspaces, realizing counterdiabatic driving. A sympathetic reader would care because the result unifies Krylov reduction, integrable Toda dynamics, and holomorphic-subspace geometry for non-Hermitian systems, making the geometric meaning of Arnoldi coefficients explicit.","feed_headline":"Arnoldi coefficients are Toda variables for non-Hermitian Hamiltonians","feed_subtitle":"The same numbers fix the Fubini-Study metric, Berry curvature, and counterdiabatic driving.","key_machinery":"The central object is the Gram-determinant tau function of the finite two-dimensional Toda lattice built from holomorphically deformed Krylov vectors: $\\tau_n = \\det\\big[\\langle \\psi(\\bar z)|(H^\\dagger)^j H^i|\\psi(z)\\rangle\\big]_{i,j=0}^{n-1}$, which equals the squared volume of the exterior Krylov state $|\\Psi_n\\rangle\\rangle$. The Desnanot–Jacobi identity places these determinants into the Hirota bilinear identity, and logarithmic derivatives define the Flaschka variables $a_n$, $b_n$. The load-bearing identity is Eq. (13), $a_n=\\alpha_n$ and $b_n=\\beta_n$, where $\\alpha_n$ and $\\beta_n$ are the diagonal and subdiagonal entries of $A=U^\\dagger H U$; this identity is what transfers Toda, metric, curvature, and counterdiabatic content to the Arnoldi coefficients.","core_discovery":"At each value of $z$, with the normalized seed $|u_0\\rangle = |\\psi(z)\\rangle/\\sqrt{\\tau_1}$, the diagonal entries $\\alpha_n(z,\\bar z)$ and subdiagonal entries $\\beta_n(z,\\bar z)$ of the Arnoldi matrix $A(z,\\bar z) = U^\\dagger H U$ coincide with the Flaschka variables $a_n(z,\\bar z)$, $b_n(z,\\bar z)$ defined from the Krylov Gram determinants $\\tau_n$; this is Eq. (13) of the paper. The proof expresses the tau functions in the Arnoldi basis and compares logarithmic derivatives and ratios with the Flaschka formulas, with an independent QR-factorization proof in the appendix. Consequently $b_n^2$ equals both the Fubini–Study metric component $g^{(n)}_{z\\bar z}$ and, up to sign, the Berry curvature $F_n$ of the $n$-th holomorphic Krylov subspace. The Toda equations close on these coefficients without determining the remaining upper-Hessenberg entries, and the frame connection supplies a Hermitian tridiagonal counterdiabatic generator when the Arnoldi matrix is diagonalizable with nondegenerate spectrum.","pith_inferences":["The electrostatic reading in Appendix F suggests a measurable diagnostic: the circular mean of $\\ln \\tau_n$ is nondecreasing with radius, so probing $b_n^2$ along nested disks gives a monotone geometric charge that should be visible in any finite-dimensional non-Hermitian Krylov evolution.","The Lax-pair form $(A,-iG_\\gamma)$ indicates that exact counterdiabatic transport along closed paths in the cyclic region is an isospectral, time-parameterized Toda flow; protocols could be constructed by QR-based or inverse-scattering methods for the Arnoldi matrix rather than by diagonalizing $H$.","If the correspondence extends to superoperator or Lindblad dynamics as the conclusion suggests, the $b_n^2$ identities could supply exact counterdiabatic terms for open-system state preparation where bi-Lanczos methods are currently approximate."],"forward_implications":["The Toda equations $\\partial_{\\bar z} a_n = b_n^2 - b_{n+1}^2$ and $\\partial_z b_n^2 = (a_{n-1}-a_n)b_n^2$ hold as exact identities for the Arnoldi coefficients throughout the cyclic region.","The squared subdiagonal coefficient $b_n^2$ is the Fubini–Study speed of the $n$-dimensional Krylov subspace along the holomorphic parameter direction.","Along any smooth real path in the cyclic region, $W(s)=U^\\dagger(s)U(0)$ generates an exact isospectral evolution $i\\dot A = [G_\\gamma,A]$, preserving the spectrum and Jordan structure even when $A$ is nondiagonalizable.","In the nondegenerate, diagonalizable case, the Arnoldi-frame generator $G_\\gamma$ cancels transitions between instantaneous eigenspaces, so adding it to $A$ realizes exact counterdiabatic driving.","In the Hermitian limit the construction reduces to the known Toda–Lanczos correspondence and to the standard Hermitian counterdiabatic term."],"supporting_citations":[{"why":"Defines the two-dimensional Toda lattice, tau-function identities, and Flaschka variables used in the main theorem.","marker":"[14, 15]"},{"why":"Defines the Arnoldi iteration and the upper-Hessenberg form $A=U^\\dagger H U$ for non-Hermitian matrices.","marker":"[9, 10]"},{"why":"Supplies the known Toda–Lanczos correspondence for Hermitian Hamiltonians that the present result generalizes and reduces to.","marker":"[8]"},{"why":"Provides the Berry connection formula for a normalized holomorphic state, used to obtain $A_n$ and $F_n$.","marker":"[18]"},{"why":"Supplies the pullback Fubini–Study metric expression used to equate the metric component with $b_n^2$.","marker":"[19]"},{"why":"Gives the non-Hermitian counterdiabatic term with oblique projectors that $G_\\gamma$ generalizes and is compared against.","marker":"[21, 22]"},{"why":"Provides the Hermitian-limit counterdiabatic term $iM$ to which Eq. (55) reduces.","marker":"[23]"}],"fun_headline_variants":["Arnoldi coefficients are Toda variables","Toda lattice from Arnoldi reduction","Arnoldi encodes metric, Berry curvature, and driving","Counterdiabatic driving from Arnoldi coefficients","Non-Hermitian systems reveal Toda dynamics"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The seed state must be cyclic for every $z$, meaning that $|\\psi(z)\\rangle, H|\\psi(z)\\rangle, \\ldots, H^{D-1}|\\psi(z)\\rangle$ span the whole space, and the deformation must have the exact exponential form $|\\psi(z)\\rangle = e^{-zH}|\\psi(0)\\rangle$; without cyclicity some $\\tau_n$ vanish and some Arnoldi vectors do not exist, and without the exponential form the Gram determinants do not satisfy the two-dimensional Toda bilinear identity.","fun_headline_variants_meta":{"raw":{"variants":["Arnoldi coefficients are Toda variables","Toda lattice from Arnoldi reduction","Arnoldi encodes metric, Berry curvature, and driving","Counterdiabatic driving from Arnoldi coefficients","Non-Hermitian systems reveal Toda dynamics"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000302,"raw_usage":{"total_tokens":1779,"prompt_tokens":1025,"completion_tokens":754,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":641,"completion_tokens_details":{"reasoning_tokens":684}},"tokens_in":641,"tokens_out":754,"duration_ms":7578,"temperature":1.0,"reasoning_tokens":684,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T15:03:24.062812+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a non-Hermitian $H$ and a cyclic seed, and for many values of $z$ compute both the Arnoldi coefficients $\\alpha_n,\\beta_n$ and the Flaschka variables from $\\tau_n=\\det(K_n^\\dagger K_n)$; if any $\\alpha_n$ differs from $-\\partial_z\\ln(\\tau_{n+1}/\\tau_n)$ or any $\\beta_n$ differs from $\\sqrt{\\tau_{n+1}\\tau_{n-1}/\\tau_n^2}$ while all $\\tau_n>0$, the main theorem is false. A sharper check is to approach a point where Arnoldi breakdown lowers the Krylov dimension below $n$ and verify that $b_n^2$ either vanishes for the invariant subspace or that the Fubini–Study metric ceases to exist, matching the paper's three-site Chern-number prediction.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the known Toda–Lanczos correspondence for Hermitian Hamiltonians that the present result generalizes and reduces to."},{"cited_title":"A holomorphic and invertible change of basisV7→VG adds onlyln |det (G)|2 toK Gr, so the metric is inde- pendent of the choice of basis","cited_arxiv_id":null,"evidence_quote":"Provides the Berry connection formula for a normalized holomorphic state, used to obtain $A_n$ and $F_n$."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the pullback Fubini–Study metric expression used to equate the metric component with $b_n^2$."},{"cited_title":"Hochbruck and C","cited_arxiv_id":null,"evidence_quote":"Provides the Hermitian-limit counterdiabatic term $iM$ to which Eq. (55) reduces."}],"review_version":1}