{"id":"d4b70908-911b-4f8b-8b9a-b05eb374902c","arxiv_id":"2607.29582","paper_version":1,"verdict":"REJECT","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"high","formal_verification":"none","parameter_count":4,"one_line_summary":"A hybrid quantum-classical solver using RVVQE reproduces classical resonance-mode predictions on a 5-bus grid, but without a proof that the method finds eigenvalues of general non-Hermitian admittance matrices.","lead":"This paper maps the electrical equations of a power grid onto a small quantum circuit to find dangerous resonance frequencies. The authors show a five-bus test case matches classical results, but the claimed advantage for continental-scale grids rests on assumptions the paper does not prove.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Eq. (7)'s cost is zero only on simultaneous eigenstates of H and K, which exist only if [H,K]=0 (Y normal); for the non-normal Ybus it cannot select eigenvectors of Y, so the central 5-bus validation is not explained by the stated math.","rationale":"The reader's weakest assumption identifies the same load-bearing concern: the RVVQE cost function in Eq. (7) is minimized to zero only by simultaneous eigenstates of H and K, and such states form a complete eigenbasis only when H and K commute. For the non-Hermitian, non-normal admittance matrices that arise in power grids, this condition is not met, and the true eigenvectors of Y are not in the zero set of Cvar. The paper provides no proof or numerical evidence that its 5-bus Ybus commutes with its conjugate transpose at every frequency; therefore the central claim that RVVQE accurately extracts complex eigenvalues is unsupported. The additional deflation step enforcing orthogonality between found states is also invalid for non-normal eigenvectors, which reinforces the rejection. I find no reason to soften the reader's REJECT verdict; the concern is mathematical and would apply even if Fig. 4 were perfectly reproduced, unless the specific test matrices are shown to be normal.","tokens_in":7778,"tokens_out":5549,"duration_ms":60218,"concrete_test":"At a frequency near a |Z_cm| peak (e.g., in the 100–3000 Hz sweep), reconstruct the 5-bus Ybus, form H=(Y+Y^dagger)/2 and K=(Y−Y^dagger)/(2i), and compute ||[H,K]||_F. If nonzero, compute Cvar on each exact eigenvector of Y obtained from SciPy; verify whether any has Cvar ≤ epsilon. A positive result (nonzero commutator and/or positive Cvar on all Y-eigenvectors) falsifies the convergence criterion of Algorithm 1 and explains the discrepancy. A negative result (commutator exactly zero across the sweep) would support the authors and require checking whether the test grid is atypical.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Algorithm 1 and Eq. (7) minimize Cvar = Var(H)+Var(K) with Y = H + iK. Since H and K are Hermitian, Cvar(theta)=0 iff |psi(theta)> is an eigenstate of both H and K. For a complete set of such states to coincide with the eigenstates of Y, H and K must be simultaneously diagonalizable, i.e. [H,K]=0, which is equivalent to Y being normal (YY^dagger = Y^dagger Y). The grid admittance matrix is complex symmetric but not generally normal; for Ybus = sum_k y_k c_k c_k^T with nonuniform r/x ratios, G = Re Y and B = Im Y do not commute. For such a non-normal Y, no eigenvector of Y is an eigenstate of H or K, so Cvar is strictly positive on every exact eigenvector of Y. Thus the convergence criterion Cvar<epsilon cannot be satisfied by the very states whose eigenvalues are wanted; a state found by the optimizer with Cvar≈0 would be a simultaneous eigenstate of H and K, and lambda = <H> + i<K> is not generally an eigenvalue of Y. The sequential 'orthogonality penalty' aggravates this: eigenvectors of a non-normal matrix are not mutually orthogonal, so enforcing orthogonality to previously found states prevents reaching them. The 5-bus agreement in Fig. 4 is therefore not accounted for; either Ybus there is accidentally normal (which would need to be shown for all swept frequencies) or the plotted red markers were obtained by a different procedure than Algorithm 1.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript proposes a hybrid quantum-classical framework for Resonance Mode Analysis (RMA) of power transmission grids. The non-Hermitian bus admittance matrix Y_B,f is decomposed into Hermitian parts H_f and K_f, and the Real Variance-based Variational Quantum Eigensolver (RVVQE) is used to minimize C_var = Var(H_f) + Var(K_f) as a proxy for finding eigenstates. The authors claim that this yields the complex eigenvalues of Y_B,f with a logarithmic number of qubits, and they validate the approach on a 5-bus system padded to 8×8, comparing the critical modal impedance |Z_cm| against exact classical diagonalization over 100 Hz to 3000 Hz.","tokens_in":8260,"tokens_out":10557,"duration_ms":112114,"significance":"The application of quantum eigensolvers to power-system resonance analysis is timely, and the authors are right that standard Hermitian VQE is inapplicable to non-Hermitian admittance matrices. The paper deserves credit for benchmarking against exact SciPy diagonalization rather than fitting to the target, and for clearly identifying the non-Hermitian challenge. However, the central algorithmic claim is not supported by the stated mathematics: the cost function in Eq. (7) vanishes only on simultaneous eigenstates of H_f and K_f, which generally do not coincide with the eigenstates of a non-normal Y_B,f. This is not a minor technical gap but a load-bearing flaw affecting Algorithm 1, the convergence criterion, and the interpretation of the reported numerical agreement. The scalability argument also omits the cost of Pauli decomposition and measurement. If the central claim were correct, the paper would be a valuable bridge between quantum algorithms and power-system stability, but in its present form the proposed method does not demonstrably compute the desired eigenvalues.","major_comments":[{"comment":"The deflation/orthogonality penalty described in §III is incompatible with non-Hermitian eigenproblems. Eigenvectors of a non-normal matrix are not mutually orthogonal in the Hilbert-space inner product ⟨·|·⟩. For a complex symmetric Y, the appropriate biorthogonality relation is v_i^T v_j = δ_ij after suitable scaling, not ⟨v_i|v_j⟩=0. Enforcing ⟨ψ_current|ψ_found⟩=0 as a penalty can move the optimizer away from true eigenvectors and generally prevents extraction of the full spectrum. Since the manuscript claims that the sequential deflation technique yields all eigenvalues, this is an independent correctness issue for Algorithm 1's stated output.","section":"§II-B, Eq. (7); Algorithm 1, steps 9–15"},{"comment":"The deflation/orthogonality penalty described in §III is incompatible with non-Hermitian eigenproblems. Eigenvectors of a non-normal matrix are not mutually orthogonal in the Hilbert-space inner product ⟨·|·⟩. For a complex symmetric Y, the appropriate biorthogonality relation is v_i^T v_j = δ_ij after suitable scaling, not ⟨v_i|v_j⟩=0. Enforcing ⟨ψ_current|ψ_found⟩=0 as a penalty can move the optimizer away from true eigenvectors and generally prevents extraction of the full spectrum. Since the manuscript claims that the sequential deflation technique yields all eigenvalues, this is an independent correctness issue for Algorithm 1's stated output.","section":"§III, 'To obtain all the eigenvalues...'"},{"comment":"The logarithmic-memory scalability claim is not established. The number of qubits N=⌈log2 n⌉ is logarithmic, but the Hermitian components H_f and K_f must be represented as sums of Pauli strings. For a dense n×n matrix this decomposition already has O(n²)=O(4^N) terms, and for sparse Y_B,f the number of Pauli terms and the associated measurement overhead are not analyzed. A logarithmic qubit count does not imply logarithmic total resources: the classical preprocessing, the number of circuit evaluations, and the memory needed to store the Pauli decomposition may reintroduce the scaling bottleneck the paper claims to bypass.","section":"§II-B, Algorithm 1 step 7; §IV"}],"minor_comments":[{"comment":"Typo: 'Qauntum Simulation' should be 'Quantum Simulation'.","section":"Fig. 4"},{"comment":"The central RVVQE algorithm is delegated to an unpublished, same-author preprint (ref. [23]) with no independent proof or implementation details. Algorithm 1 is a useful summary, but important choices—ansatz design, optimizer, deflation-penalty strength, zero-eigenvalue removal threshold, diagonal-shift amount—are only described verbally, which hampers reproducibility.","section":"References [23]"},{"comment":"Padding the 5×5 Ybus to 8×8 introduces artificial zero eigenvalues. The manuscript says an 'automatic validation routine removes zero eigenvalues,' but it does not explain how these are distinguished from physically meaningful near-zero resonance modes without biasing the extracted critical modes.","section":"§III, zero-padding"},{"comment":"The identity R_f = L_f^{-1} = L_f^T is not automatically true for every complex symmetric matrix; it requires a particular scaling of the eigenvectors. Please state the normalization or justify the equality, since it underlies the participation-factor formula in Eq. (5).","section":"§II-A, Eq. (4)"},{"comment":"The validation shows only the magnitude of the critical modal impedance |Z_cm|. No comparison of the full eigenvalue spectra, eigenvectors, or participation factors is reported, and no convergence history or final C_var values are given. Such data would be necessary to assess whether the quantum simulation actually converged to the intended eigenstates.","section":"Fig. 4 and §III"}],"recommendation":"reject","confidential_remarks":"The paper's central numerical claim is contradicted by its own cost function. Even setting aside the dependence on an unpublished same-group preprint, Eq. (7) cannot select eigenvectors of a non-normal matrix, and the orthogonality-based deflation is invalid for non-Hermitian eigenproblems. These are fundamental issues that cannot be repaired by local edits. If the authors can demonstrate normality of their specific Y_B,f matrices or provide a different algorithm, a resubmission might be warranted, but the present manuscript does not meet the bar for publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"This paper applies the authors' own RVVQE algorithm to a 5-bus power-grid resonance analysis. The RMA background is fine, and the idea of encoding the admittance matrix in log number of qubits is standard but worth stating. The 5-bus test against classical SciPy diagonalization is a sensible sanity check, and the text is honest about the absence of error mitigation. The problem is real.\n\nBut the central math doesn't hold up. Eq. (7) minimizes the sum of variances of H and K. That sum is zero only for a state that is an eigenstate of both H and K. For a non-Hermitian (and generally non-normal) admittance matrix Y = H + iK, H and K do not commute, so there is no state that is an eigenstate of both. In fact, C_var is strictly positive on every exact eigenvector of Y. So a converged state with C_var ≈ 0 would be a simultaneous eigenstate of H and K, and λ = ⟨H⟩ + i⟨K⟩ is not generally an eigenvalue of Y. The sequential orthogonality penalty is likewise misguided: eigenvectors of a non-normal matrix are not mutually orthogonal, so forcing orthogonality keeps the optimizer away from the targets.\n\nThat leaves Fig. 4 unexplained. Either the 5-bus Ybus is accidentally normal at every frequency (which the paper never shows), or the red markers came from a different procedure than Algorithm 1. No code, data, or error bars are given, so there's no way to check. And the \"quantum simulation\" label suggests this is a classical simulation of a 3-qubit circuit, not a hardware run.\n\nThe scalability claim also overreaches. Logarithmic qubit count only addresses memory. The paper doesn't address the cost of preparing the trial state, estimating the variance, or extracting eigenvalues, and it explicitly leaves QRAM to future work.\n\nWho is this for? Someone curious about whether VQE-style methods can touch non-Hermitian eigenvalue problems might find the example useful as a caution. But as it stands, the paper doesn't support its claims. I wouldn't cite it, and I wouldn't send it to review — the flaw is fundamental and the evidence is too thin to justify referee time. A resubmission after a proper method (or a proof that Y is normal for this test case) might be worth another look.","headline":"The core variational cost function cannot select eigenvectors of a non-Hermitian matrix; the 5-bus validation is unexplained.","tokens_in":8662,"tokens_out":4126,"would_cite":false,"duration_ms":44327,"reading_group":"maybe","serious_thinker":"no","would_accept_peer_review":false},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proposes a quantum-classical hybrid solver that uses RVVQE to extract complex eigenvalues of a power grid's non-Hermitian admittance matrix, enabling resonance mode analysis with logarithmic memory.","keywords":["resonance mode analysis","non-Hermitian eigenvalue problem","variational quantum eigensolver","RVVQE","power system stability","harmonic resonance","quantum-classical hybrid","logarithmic memory scaling"],"falsifier":"Take a small non-normal admittance matrix (e.g., a 3×3 complex symmetric matrix) with known eigenvalues and non-commuting H and K. Run RVVQE with the same cost function: if the optimizer cannot drive C_var to zero and the recovered eigenvalues deviate from exact diagonalization beyond numerical tolerance, the central claim fails. A direct check is to compute the residual norm ||(Y − λ I)|ψ⟩|| for the obtained eigenpair on any test instance.","tokens_in":7719,"feed_emoji":"⚡","tokens_out":9354,"duration_ms":85444,"temperature":0.7,"pith_summary":"The paper aims to show that resonance mode analysis (RMA) of transmission grids — typically a memory-hungry diagonalization of large non-Hermitian matrices — can be performed on a quantum computer using only logarithmically many qubits. The authors propose mapping the admittance matrix Y_B,f = H_f + iK_f into Pauli operators and running the Real Variance-based Variational Quantum Eigensolver (RVVQE), which minimises the sum of variances of H_f and K_f to find a state that is a common eigenstate of both. They validate the routine on a padded 5-bus benchmark, reporting that the quantum-derived critical modal impedance |Z_cm| closely matches the classical eigen-solution over 100–3000 Hz. If the claim holds, engineers could analyse resonance instabilities in grids with tens of thousands of buses without storing the full matrix, since N = ⌈log2 n⌉ qubits suffice. The practical motivation is to make RMA scalable for continental-scale power networks.","feed_headline":"Quantum algorithm reproduces power-grid resonance modes","feed_subtitle":"RVVQE matches classical eigen-decomposition on a 5-bus test, enabling resonance analysis with logarithmic qubit count.","key_machinery":"The load-bearing mechanism is the Real Variance-based Variational Quantum Eigensolver (RVVQE), a VQE-style algorithm for non-Hermitian operators. It splits the non-Hermitian admittance matrix into its Hermitian parts H and K, maps each to Pauli strings on N = ⌈log2 n⌉ qubits, and minimises the real cost C_var = Var(H) + Var(K). Zero variance is interpreted as meaning that the trial state is a simultaneous eigenvector of both H and K, so the complex eigenvalue is read out as ⟨H⟩ + i⟨K⟩. This variance-minimization step, together with the logarithmic qubit encoding, is what carries the argument from a non-Hermitian eigenproblem to a near-term quantum measurement.","core_discovery":"The central claim is that the RVVQE algorithm reliably extracts the complex eigenvalues of the non-Hermitian admittance matrix that governs resonance mode analysis. For each frequency step, the algorithm writes Y_B,f = H_f + iK_f with Hermitian H_f and K_f, encodes these operators into N-qubit Pauli strings, and prepares a variational state |ψ(θ)⟩. The cost C_var(θ) = ⟨H_f²⟩ − ⟨H_f⟩² + ⟨K_f²⟩ − ⟨K_f⟩² is minimized; when it reaches zero the state is a simultaneous eigenstate of H_f and K_f, and the associated eigenvalue is λ = ⟨H_f⟩ + i⟨K_f⟩. Using a 3-qubit ansatz on a zero-padded 5×5 admittance matrix, the authors report that the critical resonance modal impedances obtained this way are in","pith_inferences":["The zero-variance objective implicitly requires H and K to have a common eigenvector; for a general non-normal admittance matrix this is not guaranteed, so the method's validity may be restricted to cases where H and K nearly commute or to matrices that are unitarily diagonalizable. The paper does not address this constraint.","The empirical validation uses a 5-bus system padded to 8×8, which is a very small non-normal matrix; scaling up may reveal that the variance landscape is rugged and that the classical optimizer struggles to reach global minima, a risk the paper does not quantify.","The same variance-based technique could be adapted to other non-Hermitian eigenvalue problems arising in open quantum systems (Lindbladians) or linear stability analysis, so the contribution may be a general template rather than a grid-specific solution."],"forward_implications":["Resonance mode analysis of a 70,000-bus grid would need about 17 qubits, eliminating the classical O(n²) memory bottleneck for admittance matrices.","The same RVVQE workflow applies directly to larger IEEE test systems (118-bus, 300-bus) and to synthetic continental-scale grids such as those proposed as future benchmarks.","The reported agreement was obtained without quantum error mitigation; the paper argues that adding ZNE or PEC should further improve accuracy on physical hardware.","The extracted modal impedances plug directly into stability criteria such as the positive-mode-damping condition, so the quantum output is usable for harmonic mitigation and contingency assessment."],"fun_headline_variants":["Quantum solver reproduces grid resonance modes","Hybrid quantum method scales grid resonance analysis","RVVQE eigensolver matches classical grid models","Few-qubit quantum circuit finds grid instabilities","Quantum memory bypasses RAM for grid eigenmodes"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The entire method rests on the existence of a quantum state that is simultaneously an eigenvector of both Hermitian components H and K of the non-Hermitian admittance matrix; for a general non-normal matrix this simultaneous eigenvector does not exist, so the variance cost may never reach zero and the extracted 'eigenvalues' could be meaningless.","fun_headline_variants_meta":{"raw":{"variants":["Quantum solver reproduces grid resonance modes","Hybrid quantum method scales grid resonance analysis","RVVQE eigensolver matches classical grid models","Few-qubit quantum circuit finds grid instabilities","Quantum memory bypasses RAM for grid eigenmodes"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000273,"raw_usage":{"total_tokens":1512,"prompt_tokens":822,"completion_tokens":690,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":566,"completion_tokens_details":{"reasoning_tokens":620}},"tokens_in":566,"tokens_out":690,"duration_ms":7926,"temperature":1.0,"reasoning_tokens":620,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-03T04:07:19.418156+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a small non-normal admittance matrix (e.g., a 3×3 complex symmetric matrix) with known eigenvalues and non-commuting H and K. Run RVVQE with the same cost function: if the optimizer cannot drive C_var to zero and the recovered eigenvalues deviate from exact diagonalization beyond numerical tolerance, the central claim fails. A direct check is to compute the residual norm ||(Y − λ I)|ψ⟩|| for the obtained eigenpair on any test instance.","supporting_citations":[],"review_version":1}