{"id":"982e8fa8-0083-4f75-b892-0d07a411c712","arxiv_id":"2607.13790","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"COMPASS combines adaptive recurrent neural networks with biorthogonal Monte Carlo to simulate non-Hermitian spin systems, revealing ansatz-induced PT breaking, frustration-gap shielding, and a diabolic ring of level crossings.","lead":"A neural-network framework called COMPASS computes ground states of non-Hermitian quantum magnets without first converting them into Hermitian problems. It shows that choosing real versus complex wave-functions matters more than raw expressive power, and that magnetic frustration can shield such systems from spectral instabilities.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"COMPASS's convergence to the ground-state biorthogonal pair is asserted, not established: the energy loss (Eq. B2) is a sum of pure left/right expectations, so large-N and frustrated-model claims rest on an untested optimization assumption.","rationale":"The reader's weakest_assumption correctly identifies the complementary loss as the load-bearing point. The paper's abstract and conclusion claim a 'direct study' of NH many-body systems without Hermitian embeddings; that claim is only as strong as the optimizer's ability to find the ground-state biorthogonal pair. The evidence for this is indirect: Appendix C shows the complementary protocol outperforms pure energy and pure variance on the PT-symmetric TFIM, Fig. 10 shows ED agreement for N=10 frustrated chains, and Appendix E reports high left-right overlap. These are good signs but none of them prove the loss landscape has the claimed property. In fact, the energy loss in Eq. (B2) is not the biorthogonal estimator; it is the sum of right and left pure expectations. For non-Hermitian H, such expectations are complex and can be minimized by states unrelated to the ground eigenpair (e.g., states with large negative real part in a wrong sector, or states that make ⟨ΨR|H|ΨR⟩ small while ⟨ΨL|H†|ΨL⟩ is small for a different eigenvalue). The variance term cannot disambiguate because pure variances vanish for any eigenstate pair sharing an eigenvalue. Thus the central scalability and topology results are conditional on a heuristic that has not been characterized. This is not a dispute with the physical community's consensus; it is an internal gap between the assertion 'ensures' and the provided mathematics. The test proposed—ED benchmarks with random seeds and a comparison against the mixed-estimator loss—would settle whether the heuristic is reliable in the small systems where exact answers exist; if it fails there, the large-N claims lose their foundation. The diabolic-ring topology and ansatz-selection findings are secondary; even if they are later refined, the convergence of COMPASS is the enabling assumption. Because the paper does provide N=10 ED agreement and a plausible warm-start mechanism, the appropriate verdict remains CONDITIONAL rather than rejection; no change from the reader's verdict is needed.","tokens_in":28287,"tokens_out":7233,"duration_ms":70094,"concrete_test":"Perform an ED-controlled benchmark on the two frustrated models (Eqs. 13 and 14) for N=10–14 over the same (J2/J1, γ/J1) grid used in Figs. 7–10: run COMPASS with the proposed L from multiple random seeds and record whether the converged pair matches the exact ground eigenpair (Re E minimal). In the same setup, run a variant whose energy loss is the real part of the biorthogonal mixed estimator E_bi = ⟨ΨL|H|ΨR⟩/⟨ΨL|ΨR⟩ while keeping the variance and warm-start protocol unchanged. If the proposed L picks the wrong eigenpair for any finite fraction of grid points/seeds, or if the mixed-estimator variant is systematically more reliable, the convergence guarantee in Sec. II.A.4 is unsupported and all large-N claims need to be re-benchmarked.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim—direct large-scale NH simulation without Hermitian embeddings/adiabatic continuation—rests on Sec. II.A.4's assertion that Eq. (8) 'ensures' convergence to the correct ground-state biorthogonal pair. The energy part of that loss, Eq. (B2), is L_e = ⟨ΨR|H|ΨR⟩/⟨ΨR|ΨR⟩ + ⟨ΨL|H†|ΨL⟩/⟨ΨL|ΨL⟩, a sum of two pure expectations, not the mixed biorthogonal estimator ⟨ΨL|H|ΨR⟩/⟨ΨL|ΨR⟩ that defines bVMC (Sec. II.A.1, Appendix A) and is used for the final energy. For a non-Hermitian H these pure expectations have no variational lower bound and are generally complex; no stationary-point argument or counterexample shows that minimizing L_e (with the variance term) selects the smallest-real-part biorthogonal eigenpair rather than an excited or spurious pair. The variance term only forces each state to be an eigenstate (right of H, left of H†) with a common eigenvalue; it does not select the ground state. The warm-start schedule is heuristic: Appendix C shows pure variance can converge to excited states and pure energy has large error, but the complementary loop is not proven to avoid these traps. ED benchmarks are limited to N≤10–14, so the N=200 (1D) and N=100 (2D) results and the large-N frustrated phase diagrams inherit this unverified convergence assumption. The N=10 ED agreement in Fig. 10 is real supporting evidence, but it does not cover the large-N regime where the strongest claims are made.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper introduces COMPASS, a non-Hermitian neural quantum state method built from two adaptive recurrent networks representing left and right eigenstates, trained with a complementary loss (Eq. 8) that switches between energy minimization and variance minimization, with exact autoregressive sampling rather than MCMC. The authors demonstrate the method on 1D and 2D PT-symmetric transverse-field Ising chains, report adaptive/static speed comparisons, and argue that real-valued ansätze are necessary in the PT-unbroken phase while complex ansätze are needed for generic complex spectra. They then study two non-Hermitian J1-J2 spin-chain models: one with a staggered imaginary field (Model 1) and one with a complex next-nearest-neighbor coupling (Model 2). For Model 1 they claim a 'frustration-gap shield' whereby the spectral gap sets a quantitative threshold for PT breaking; for Model 2 they claim a 'diabolic ring'—a closed curve of real-part level crossings with no Hermitian analog, associated with eigenvalue braiding. Large-system claims are made for N=200 (1D) and N=100 (2D) TFIM simulations, while the frustrated models are benchmarked only at N=10 against exact diagonalization.","tokens_in":28787,"tokens_out":7544,"duration_ms":82363,"significance":"If the convergence and topology claims hold, COMPASS would be a genuinely useful extension of neural quantum states to non-Hermitian, frustrated and non-stoquastic systems, and the ansatz-selection insight (real vs complex networks) is physically well motivated. The paper's strengths include exact autoregressive sampling without MCMC, a clean adaptive capacity schedule, explicit ED benchmarks at N=10, and an honest self-falsification of the naive |J2+iγ| effective-coupling hypothesis in Sec. V.B. However, the advertised guarantees are currently stronger than what is demonstrated: the complementary loss has no proven ground-state selection property, the diabolic-ring braiding is asserted rather than computed, and the 'quantitative' gap-shield relation is not backed by scaling data. These issues are load-bearing for the central claims, so a major revision is required.","major_comments":[{"comment":"The statement that the complementary loss 'ensures' convergence to the correct ground-state biorthogonal pair is not derived. With Eq. (B2), L_e = ⟨ΨR|H|ΨR⟩/⟨ΨR|ΨR⟩ + ⟨ΨL|H†|ΨL⟩/⟨ΨL|ΨL⟩, the energy loss is a sum of pure left/right expectations, not the mixed biorthogonal estimator of Eq. (2). For non-Hermitian H these expectations have no variational lower bound and are generally complex; the text does not specify how a complex loss is optimized. The variance terms force |ΨR⟩ and |ΨL⟩ to be eigenstates of H and H† with a common eigenvalue but do not select the ground state, and the λ_t warm-start schedule is heuristic. Appendix C benchmarks against ED only at N=10 for the PT-symmetric TFIM; the large-N comparisons use SE, not an independent NH solver. Since the N=200/100 claims and the frustrated-model results inherit this assumption, please provide a stationary-point argument for Eq. (8","section":"Sec. II.A.4 / App. B / App. C"},{"comment":"The central new-topology claim—the 'diabolic ring' supporting eigenvalue braiding—is not demonstrated. The ring is identified as the locus where Re(E1)=Re(E0) with distinct imaginary parts, i.e., a real-part degeneracy line. The paper asserts that encircling the ring exchanges E0 and E1 ('spectral flow topology'), but no closed loop in (J2,γ) is tracked, no eigenvector-following or monodromy calculation is shown, and no braid invariant is computed. The maximum eigenvector overlap of ≈0.89 reported in Appendix I does not establish the absence of EPs by itself. Without this evidence, the ring is an interesting level-crossing curve, but the 'topologically nontrivial' and 'no Hermitian analog' statements are unsupported. Please provide the braiding/monodromy data or substantially qualify the claim.","section":"Sec. V.C / Fig. 9"},{"comment":"The 'quantitative shield' relation γ_c ∝ Δ is not supported by the displayed data. At the Majumdar-Ghosh point (J2/J1=0.5), the zero-field gap within the Sz=0 sector is exactly zero (Fig. 7c), yet Fig. 8a shows |Im E0|=0 up to γ_c≈0.95. Thus the threshold is not set by the zero-field gap; at the MG point the protection must instead arise from the γ-induced splitting of the degenerate dimer states. A plot of γ_c versus the relevant gap, or a finite-size scaling analysis, is needed before calling the shield quantitative. As written, the mechanism is plausible but the quantitative claim is unjustified.","section":"Sec. V.B / Figs. 7-8"},{"comment":"The large-scale demonstrations (N=200 in 1D, N=100 in 2D) are only for the PT-symmetric TFIM, where the spectrum is real in the unbroken phase and the model is stoquastic-like. The frustrated J1-J2 models, which are the basis for the 'beyond stoquasticity' and 'direct study of NH many-body systems' claims, are benchmarked only at N=10 against ED. The abstract and conclusion should either distinguish 'method scales on the TFIM' from 'method works for frustrated NH systems' or include a large-N frustrated benchmark. This is not a fatal flaw, but it currently overstates the scope of the numerical evidence.","section":"Secs. III-V / Abstract"}],"minor_comments":[{"comment":"The caption contains 'δλ=?' with no value or definition; this should be specified.","section":"Fig. 4 caption"},{"comment":"The left/right variance losses are labeled L_R^e and L_L^e; these should be L_R^v and L_L^v to match Eq. (B3).","section":"Appendix B, Eq. (B4)"},{"comment":"The abbreviations cRNN, pRNN, and SE are used without explicit definition. 'pRNN' is later described as 'real (positive)', but the relation between 'real-valued' and the amplitude parametrization of Eq. (4) should be clarified.","section":"Table I / Sec. III"},{"comment":"The text says the snake scan is 'two orders of magnitude more accurate' but the inset shows ε_rel ≈ 10^-2, i.e., about two orders of magnitude smaller than the raster energy; the phrasing is confusing.","section":"Sec. IV.B, Fig. 6"},{"comment":"Minor typos: 'bugles outward' should be 'bulges outward'; the Note added contains 'antsatz'; Appendix F has malformed notation 'E ∼ |ΨR|2'.","section":"Sec. V.B and elsewhere"},{"comment":"The GitHub reference lists author names inconsistent with the manuscript author list; please verify the citation entry.","section":"Ref. [51]"}],"recommendation":"major_revision","confidential_remarks":"The stress-test concern about the complementary loss is valid and is the main reason for major revision. The paper's most novel physical claim—the diabolic ring—needs a concrete braiding computation before it can be accepted as topological. I would also ask the editor to check the novelty statement against Ref. [8] and Ref. [7], since the methodological overlap with those works is substantial and the present manuscript's distinct contribution must be crisply delineated."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague — quick take: the paper is worth reading for two things: the observation that complex-valued RNN ansätze spontaneously break PT symmetry during optimization in the unbroken phase, and the ED-backed finding that a frustration gap shields PT symmetry with γ_c ∝ Δ. The larger claim — COMPASS reaches N=200/100 without Hermitian embeddings or adiabatic continuation — is plausible but not actually established.\n\nWhat’s new and good: the package of adaptive RNN growth, independent left/right biorthogonal networks, exact autoregressive sampling, and alternating energy/variance minimization is a real methodological contribution. The ansatz-selection result is non-obvious and useful: a real ansatz constrains the optimization to the real-spectrum manifold for PT-symmetric systems, while complex ansätze are needed for generic complex spectra. The gap-frustration shield is a parameter-free ED observation with stated assumptions; the phase diagrams in Figs. 7–8 look consistent with that narrative. The diabolic ring is an interesting level-crossing locus, though I’d call it “observed” rather than “topologically nontrivial” without a formal braiding analysis — the authors themselves note no EPs are detected and leave eigenvalue tracking for larger γ.\n\nThe soft spot is load-bearing. Section II.A.4 and Appendix B define the energy loss as a sum of pure left/right expectations (Eq. B2), not the biorthogonal mixed estimator. For non-Hermitian H these pure expectations have no variational lower bound and are generally complex. The paper asserts that the complementary loss “ensures” convergence to the ground-state biorthogonal pair, but no stationary-point argument or counterexample shows that minimizing this sum selects the smallest-real-part pair. The variance term only forces each state to be an eigenstate of H (right) and H† (left) with a common eigenvalue; it doesn’t select the ground state. The warm-start protocol is heuristic, and Appendix C shows pure variance can land on excited states. Since ED benchmarks stop at N=10–14, the N=200 (1D) and N=100 (2D) claims and the large-N phase diagrams inherit this unverified assumption. The N=10 ED agreement in Fig. 10 is real supporting evidence, but it doesn’t cover the regime where the strongest scalability claims are made. Minor issues: Fig. 4 has a literal missing parameter (δλ=?), and the code is under embargo despite the data availability statement.\n\nOverall: this is a serious piece of work with credible small-system results and an honest tendency to report its own failed predictions — the effective-frustration hypothesis is explicitly falsified in the text. I’d send it to peer review with a request for substantial revision, not desk reject. The referee should ask for a proof or at least a counterexample-based argument for convergence of Eq. B2, a clear statement of what the large-N results actually validate, and a more careful topological characterization of the diabolic ring.","headline":"Useful ansatz-selection insight and credible ED-scaled observations, but the convergence guarantee underpinning the large-N claims is asserted, not proven.","tokens_in":29223,"tokens_out":3304,"would_cite":true,"duration_ms":52958,"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":"This paper introduces COMPASS, a biorthogonal adaptive neural-network method that simulates non-Hermitian many-body systems directly and reveals that frustration gaps shield spectra from non-Hermitian instability while complex couplings cre","keywords":["non-Hermitian quantum systems","neural quantum states","biorthogonal variational Monte Carlo","parity-time symmetry","frustrated quantum magnetism","diabolic ring","adaptive recurrent neural networks","exceptional points"],"falsifier":"Exact-diagonalize a small (N≤12) instance of the complex-NNN J1-J2 model and run COMPASS from many random initializations: if any run converges to a zero-variance pair whose real-part energy exceeds the exact ground-state energy, the complementary-loss convergence claim fails. A second, independent test: measure the maximum eigenvector overlap at a point on the diabolic ring—if the overlap approaches 1, the ring is exceptional, not diabolic.","tokens_in":28181,"feed_emoji":"🧲","tokens_out":6830,"duration_ms":60688,"temperature":0.7,"pith_summary":"COMPASS is a variational framework that lets neural quantum states tackle non-Hermitian many-body systems directly, without mapping them onto Hermitian problems or slowly switching on non-Hermiticity. The paper shows that the choice of ansatz is physically decisive: for parity-time-symmetric Hamiltonians, an unconstrained complex neural network can spontaneously break the symmetry and produce spurious imaginary energies, while a real-valued network stays on the physical manifold; for generic non-Hermitian Hamiltonians the reverse is true. With the framework the authors simulate frustrated spin chains and find that the energy gap from frustration protects the spectrum against non-Hermitian instability up to a critical strength, and that making the frustration coupling complex produces a closed ring of level crossings—the 'diabolic ring'—a spectral topology with no Hermitian counterpart. If correct, this opens non-Hermitian frustrated magnets to scalable numerical study and suggests that symmetry-aware architecture choice, not raw expressivity, is what makes variational simulations reliable.","feed_headline":"New method reaches 200-spin non-Hermitian systems directly","feed_subtitle":"COMPASS pairs left/right neural states with energy-variance optimization to probe frustrated quantum magnets.","key_machinery":"The central machinery is COMPASS: a pair of independent gated-recurrent-unit autoregressive neural networks, one for the right eigenstate and one for the left, whose log-amplitudes are complex and factorized as products of conditionals with probability and phase parts. It combines biorthogonal variational Monte Carlo (sampling from |Ψ_L Ψ_R|) with a complementary loss L = λ_t L_e + (1-λ_t)L_v that alternates energy minimization (to select the ground state) and variance minimization (to enforce the eigenstate condition), plus a warm-start schedule and an adaptive architecture that grows the hidden dimension while transferring parameters. Exact autoregressive sampling removes Markov-chain nois","core_discovery":"The paper's central claim is that a biorthogonal adaptive neural-network ansatz—two independent autoregressive recurrent networks representing the left and right eigenstates, trained with a complementary loss that alternates energy and variance minimization—converges to the correct ground-state eigenpair of a generic non-Hermitian Hamiltonian, enabling direct simulation of 1D (up to N=200) and 2D (up to N=100) systems without Hermitian embeddings or adiabatic continuation. Alongside the method, the paper establishes two physical results: (i) in PT-symmetric systems, real-valued ansätze are necessary to avoid spurious spontaneous PT breaking during optimization, while complex ansätze are esse","pith_inferences":["We infer that if the convergence guarantee holds, COMPASS should generalize to time-dependent non-Hermitian dynamics and to higher-dimensional frustrated lattices, where the diabolic ring might become a surface or higher-dimensional crossing manifold.","We infer that the ansatz-selection principle—constraining the network to the symmetry class of the target state beats unconstrained expressivity—likely extends to other variational families and could be formalized as a bias-variance trade-off specific to non-Hermitian optimization landscapes.","If the diabolic ring is robust to finite-size effects, it offers a testable signature of non-Hermitian frustration in cold-atom or trapped-ion simulators, where the complex next-nearest-neighbor coupling could be engineered via lossy intermediate sites.","The real part of the biorthogonal variance, which the paper notes changes sign at the PT transition, could serve as a cheap numerical order parameter for locating exceptional points in larger systems."],"forward_implications":["Ansatz selection is physically decisive: real-valued networks are required for PT-symmetric models to avoid spurious imaginary energies; complex networks are required for generic non-Hermitian spectra.","The frustration gap Δ sets the critical threshold for PT-symmetry breaking, γ_c ∝ Δ, so frustrated gapped phases are quantitatively shielded against non-Hermitian spectral instability.","Complexifying the frustration coupling yields a closed diabolic ring of real-energy level crossings with eigenvalue braiding, a spectral topology with no Hermitian analog and controlled by the phase of the complex coupling, not its modulus.","Direct simulation of non-Hermitian many-body systems is possible up to N=200 (1D) and N=100 (2D) without Hermitian embeddings or adiabatic continuation, with high biorthogonal fidelity.","The 'effective frustration' hypothesis (replacing the complex coupling by its modulus) is disproved by the absence of a phase transition along the predicted quarter-circle locus."],"fun_headline_variants":["Neural states go biorthogonal for non-Hermitian systems","COMPASS navigates frustrated magnets without Hermitian crutch","Direct 200-spin non-Hermitian simulations via neural ansatz","Biorthogonal neural nets map PT symmetry and frustration","Adaptive neural framework probes non-Hermitian many-body physics"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The load-bearing premise is that minimizing the complementary loss (sum of energy and variance) reliably lands on the smallest-real-part biorthogonal eigenpair; the paper asserts this convergence but offers no proof that the energy loss of the pure left and right expectations selects the biorthogonal ground state rather than some other eigenpair.","fun_headline_variants_meta":{"raw":{"variants":["Neural states go biorthogonal for non-Hermitian systems","COMPASS navigates frustrated magnets without Hermitian crutch","Direct 200-spin non-Hermitian simulations via neural ansatz","Biorthogonal neural nets map PT symmetry and frustration","Adaptive neural framework probes non-Hermitian many-body physics"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000184,"raw_usage":{"total_tokens":1205,"prompt_tokens":847,"completion_tokens":358,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":591,"completion_tokens_details":{"reasoning_tokens":270}},"tokens_in":591,"tokens_out":358,"duration_ms":4217,"temperature":1.0,"reasoning_tokens":270,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-02T03:42:03.258013+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Exact-diagonalize a small (N≤12) instance of the complex-NNN J1-J2 model and run COMPASS from many random initializations: if any run converges to a zero-variance pair whose real-part energy exceeds the exact ground-state energy, the complementary-loss convergence claim fails. A second, independent test: measure the maximum eigenvector overlap at a point on the diabolic ring—if the overlap approaches 1, the ring is exceptional, not diabolic.","supporting_citations":[],"review_version":1}