{"id":"bd4f3970-09a9-4616-a60a-3608683467ac","arxiv_id":"2411.08965","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"A driven-dissipative Bose-Hubbard chain with a phase-gradient drive can reach a steady state in which low- and high-density regions coexist and the low-density region carries a non-Hermitian winding number ν=1.","lead":"This paper simulates a chain of lossy, driven photon modes with a phase twist that breaks time-reversal symmetry, using Gaussian states to handle the nonlinearity. It reports a coexistence region where the steady state develops non-Hermitian topological character and directional amplification, and argues the setup is realizable with superconducting circuits.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The local winding number νj is applied at the coexistence interface where the quasi-homogeneous limit fails; no independent check links νj to the full inhomogeneous Green's function, so the claimed topological amplification is not yet established.","rationale":"The reader's primary concern is the quantitative reliability of the Gaussian ansatz for multi-site chains, and that is a serious point because the exact benchmark covers only N = 1. However, I regard the local winding-number construction as even more load-bearing for the central claim, because the topological characterization would remain unproven even if the Gaussian ansatz were exact. The paper acknowledges the quasi-homogeneous restriction on νj, yet applies it across the coexistence interface, where density gradients and fluctuations peak in Figs. 3(d-f). The independent Green's-function evidence in S VI is derived for homogeneous systems and is not directly tied to νj for the inhomogeneous chain; no finite-size scaling of the minimal singular value, and no singular-vector localization analysis, is presented. Near a critical line, large response functions are expected even without point-gap topology, so the 'topological amplification' label requires the exponential e^{-N/ξ} scaling to be demonstrated. My proposed test uses the same Gaussian steady states, so it isolates the topological question from the ansatz-reliability question. If the test passes, the central claim is substantially supported; if it fails, the authors would need a non-Bloch or local topological marker, or a weakened claim. I therefore keep the conditional verdict: the concern is a missing validation step rather than a demonstrated contradiction, and the manuscript should not be accepted without addressing it.","tokens_in":13018,"tokens_out":7808,"duration_ms":80626,"concrete_test":"Fix a coexistence point, e.g. N = 40, Δ/J = 0.5, ϕ = π/3, κ/J = 1, U/J = -2×10^{-4}, and ϵ/J = 40.5. Using the Gaussian steady-state parameters, construct the full fluctuation matrix H from Eq. (15). First, compute its singular values and the zero-frequency Green's function G(0) = -H^{-1} for N = 20, 40, 80, and 160, and verify whether the minimal singular value s0 scales as e^{-N/ξ} and whether ||G(0)||_F grows exponentially with N. Second, plot the spatial weight of the left and right singular vectors associated with s0 alongside the νj profile, and check whether this zero mode is localized at the boundary of the predicted νj = 1 region rather than at the open boundary or at the drive-profile defect. Third, compare the directional amplification profile G_ji, as in Fig. 4(b), with the region predicted by νj.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central topological claim rests on the local winding number νj = ν(gj, J, Δ̃j) introduced in the section 'Topology in the phase coexistence region'. The paper itself states that this local definition is 'correct in a quasi-homogeneous limit in which parameters vary slowly in space'. Yet in the coexistence region the parameters are not slowly varying: Figs. 3(d-f) show a sharp interface, with a large fluctuation maximum, separating the νj = 1 and νj = 0 regions, and S V defines the transition site by the gradient maximum of |α*_j|. The local formula is therefore applied exactly where its validity condition is most strained, namely at the moving interface that defines the coexistence phenomenon. Moreover, no independent validation connects νj to the actual response of the inhomogeneous chain: the Green's function analysis in Fig. 4 and S VI is performed for the full chain, but the paper never checks that the quasi-zero singular value of H in Eq. (15) is localized in, or bounded by, the νj = 1 region, nor that the minimal singular value s0 decays as e^{-N/ξ} with system size. Without such a check, a large Green's function could simply reflect proximity to a driven-dissipative transition rather than point-gap topology. This gap is load-bearing because it persists even if the Gaussian ansatz is assumed exact: the topological identification νj = 1 versus νj = 0 is an unvalidated local approximation at the interface that is central to the claimed phase coexistence.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript studies a driven-dissipative Bose-Hubbard chain with a uniform phase gradient, using a Gaussian variational ansatz to approximate the steady state of the master equation. It reports that an inhomogeneous spatial profile of the drive amplitude and detuning stabilizes steady states, and it presents a non-equilibrium phase diagram with low- and high-density regions separated by a coexistence region. The central claim is that the coexistence region exhibits non-Hermitian point-gap topology and topological amplification, identified through a spatially local winding number nu_j computed from the local parameters of the effective quadratic fluctuation Hamiltonian. Supporting results include Green's function data, correlation functions, a single-site benchmark of the Gaussian ansatz, and a finite-size scaling analysis of the apparent first-order transition.","tokens_in":13294,"tokens_out":7377,"duration_ms":71192,"significance":"If the central claim is correct, the paper would extend non-Hermitian point-gap topology and topological amplification from linear lattice models to an interacting driven-dissipative Bose-Hubbard model, with a concrete superconducting-circuit implementation. The manuscript has several strengths: the Gaussian equations of motion are derived explicitly, the single-site benchmark in SI II supports the method at weak nonlinearity, the inhomogeneous drive profile is a concrete stabilization mechanism, and the finite-size scaling in SI V provides a falsifiable prediction for the phase transition. However, the topological identification rests on a local winding number that is applied in a sharply inhomogeneous regime and is not independently connected to the full Green's function, so the main claim is not yet established at the level of rigor the paper intends.","major_comments":[{"comment":"The paper's central claim is based on the local winding number nu_j = nu(g_j, J, Delta_j) introduced in the section 'Topology in the phase coexistence region' and shown in Fig. 3(f). The authors state that this local definition is 'correct in a quasi-homogeneous limit in which parameters vary slowly in space.' In the coexistence region this condition is not met: Figs. 3(d,e) show a sharp interface between low- and high-density regions with a pronounced fluctuation maximum, and SI V defines the transition site by maximizing the gradient of |alpha*_j|. The local formula is therefore applied precisely in the regime where its validity is most strained, and no independent calculation of a topological invariant for the full inhomogeneous H is provided. This is load-bearing because the phase-coexistence claim of topological amplification rests on nu_j taking values 1 and 0 in the two regions.","section":"Topology in the phase coexistence region"},{"comment":"Fig. 4(a,b) and the discussion in SVI demonstrate enhanced and directional Green's function elements for the effective quadratic model, but they do not establish that the enhancement is due to non-Hermitian point-gap topology. The topological amplification theory in SVI predicts a quasi-zero singular value s_0 of H with s_0 ~ e^{-N/xi} in a homogeneous topological phase. The manuscript does not check this scaling for the inhomogeneous coexistence state, nor does it verify that the singular vector associated with the small singular value is localized in the nu_j = 1 region. Without such a check, the large values of ||G||_F in Fig. 4(a) and the spatial growth in Fig. 4(b) could in principle be attributed to proximity to the driven-dissipative transition (the fluctuation maximum in Fig. 3(e)) rather than to topology. A direct computation of the smallest singular values of the full N-site H and their system-size dependence would close this gap.","section":"Green's function and correlation functions"},{"comment":"The Gaussian ansatz is benchmarked only for a single site (Fig. 5(a)); for chains the paper substitutes the condition <b^dag_j b_j>/|alpha_j|^2 << 1. This ratio is necessary but not sufficient for the Wick factorization of four-operator correlators in Eq. (5) to be quantitatively accurate, and the interface region in Fig. 3(e) is exactly where fluctuations are largest. Given that the coexistence interface is the central object of the paper, the authors should provide a controlled check of the Gaussian truncation for small chains (for example N = 2 or 3 with a truncated local Hilbert space, or a higher-order truncation that quantifies the neglected fourth-order cumulants). Otherwise the steady-state profiles that feed into g_j and hence nu_j remain an assumption at the very point where the topological claim is made.","section":"Gaussian ansatz and SI II"}],"minor_comments":[{"comment":"The phrase 'numerically prove' overstates what numerical integration can show; recommend replacing it with 'demonstrate' or 'show'.","section":"Abstract and Introduction"},{"comment":"The value of N0 used in the simulations is never specified; since the bulk region is defined as N0 < j < N - N0, the choice of N0 affects the interpretation of the finite-size results in SI V.","section":"Eq. (6)"},{"comment":"The grey 'chaotic' region is not defined quantitatively; state the precise criterion used to decide that the Gaussian equations do not converge to a steady state.","section":"Fig. 2(a)"},{"comment":"The dynamical matrix H in Eq. (15) uses Delta_j without making clear whether this is the bare detuning or the dressed detuning Delta_tilde_j = Delta_j + 4U|alpha_j|^2 + U that appears in Eq. (3); this should be stated explicitly because the Green's function is computed from this matrix.","section":"SVI, Eq. (15)"},{"comment":"The in-text references to the supplementary sections (SII, SIII, SIV, SVI) would be easier to follow if they were given in a single consistent style, and if the supplementary section on the phi = 0 case were explicitly cross-referenced in the main text.","section":"Supplementary references"}],"recommendation":"major_revision","confidential_remarks":"The paper is a plausible extension of the authors' earlier work on topological amplification, and the numerical machinery is presented transparently. The load-bearing gap is the lack of a direct link between the local winding number and the Green's function of the full inhomogeneous chain; the requested checks are feasible within the manuscript's scope. If those checks are supplied, I would be willing to reconsider the paper favorably."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nThe genuinely new result is the Gaussian variational treatment of a chiral driven-dissipative Bose-Hubbard chain with a phase gradient, and the phase-coexistence region with a self-consistently moving interface between low- and high-density regions. It extends the group's earlier work on topological amplification in linear and parametrically driven lattices to an interacting model, which is the natural next step for circuit-QED implementations.\n\nThe paper does several things well. The derivation of the Gaussian equations of motion is careful, with a single-site exact diagonalization benchmark in the supplement that shows the method is accurate where it is tested. The comparison with the phi=0 case makes the role of time-reversal breaking concrete. The finite-size scaling of d|alpha*|/depsilon in SV is a genuine attempt to support the first-order transition claim, and the qualitative steady-state profiles are clear.\n\nThe soft spots are real, though fixable. The main one is the local winding number nu_j. The paper states it is only correct in a quasi-homogeneous limit, but then applies it at the coexistence interface, where the parameters vary sharply and fluctuations peak. The stress-test note is right: no check connects nu_j to the full inhomogeneous Green's function. The paper never shows that the quasi-zero singular value of H is localized in the nu_j=1 region or that s0 decays exponentially with system size. Without that, the large Green's function could simply reflect proximity to a driven-dissipative transition. The Gaussian ansatz itself is benchmarked only for N=1; for chains the validity is inferred from <b†b>/|alpha|^2 << 1, which is a necessary but not sufficient condition for Wick factorization to be accurate, especially at the interface. Also, N0 in Eq. (6) is never specified, and no code or data are provided.\n\nNone of this is fatal. The equations are internally consistent, the numerics appear reproducible, and the self-citation is not circular in the fitting sense—they are extending their prior results, not fitting to them. The central idea is plausible but not yet established.\n\nThis paper is for people working on driven-dissipative photonic lattices and topological amplification, and for experimental groups building directional amplifiers with superconducting circuits. It deserves a serious referee, but the referee should require a multi-site benchmark or a direct test of the local winding number against the full Green's function, specification of N0, and code/data.\n\nMy recommendation: send it to peer review with major revision. The load-bearing checks are missing, but the paper is honest and the direction is right.","headline":"A plausible extension of non-Hermitian topological amplification to an interacting driven-dissipative Bose-Hubbard chain, but the central topological claim rests on a local winding number applied exactly where its validity condition is strained.","tokens_in":13872,"tokens_out":2590,"would_cite":true,"duration_ms":22593,"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 driven-dissipative Bose-Hubbard chain with a linear phase gradient in its drive develops a steady-state coexistence region where a topological, amplifying domain (non-Hermitian winding number 1) borders a trivial domain (winding number…","keywords":["non-Hermitian topology","topological amplification","driven-dissipative Bose-Hubbard model","point-gap topology","Gaussian variational ansatz","winding number","phase coexistence","superconducting circuits"],"falsifier":"Compute the exact steady state of a short chain (N=4 to 8) with the same parameters (φ=π/3, κ/J=1, U/J=-2·$10^{-4}$, ε/J≈40) using a method that does not assume Gaussianity—e.g. matrix-product-operator simulation of the Lindblad master equation—and check whether the νj=1/νj=0 domain splitting, the interface fluctuation peak, and the exponential Green's-function growth survive.","tokens_in":12755,"feed_emoji":"📡","tokens_out":6786,"duration_ms":59815,"temperature":0.7,"pith_summary":"The paper introduces a chain of coupled lossy photonic modes with Kerr nonlinearity and a coherent drive whose phase advances linearly along the chain, breaking time-reversal symmetry. Using a Gaussian variational ansatz closed by Wick's theorem, it argues that a shaped drive profile (tanh ramps at the boundaries) stabilizes steady states across most of the phase diagram. The central discovery is a phase diagram with low- and high-density photon phases separated by a coexistence region in which the local non-Hermitian winding number is 1 in the low-density segment and 0 in the high-density segment. In that region the zero-frequency Green's function is exponentially enhanced, meaning the chain directionally amplifies inputs; fluctuations are largest at the moving interface between the topological and trivial domains. If correct, this is a concrete route to non-Hermitian topological phases in an interacting, dissipative bosonic lattice implementable with superconducting circuits.","feed_headline":"Chiral drive gives a photonic chain a topological amplifying phase","feed_subtitle":"In the coexistence region the chain splits into topological and trivial halves, amplifying signals one way.","key_machinery":"The argument runs through three objects. First, the Gaussian variational ansatz: writing b_j=a_j-α_j and closing the Heisenberg equations with Wick's theorem yields closed equations of motion for the coherent amplitudes α_j, normal correlators G_jk=⟨b†_j b_k⟩, and anomalous correlators F_jk=⟨b_j b_k⟩; this is what makes N=40 chains tractable and what produces the steady-state profiles. Second, the non-Hermitian dynamical matrix H=JV-i(κ/2)1 (in the Nambu basis), whose Fourier transform H(k) defines the winding number ν=(1/2π) Im ∮ dk Tr ∂_k log H(k); a local version νj=ν(g_j,J,Δ̃_j) is used to label spatial regions as topological or trivial. Third, topological amplification: the singular value decomposition of H, extended to a chiral Hermitian matrix, protects near-zero singular values when ν≠0, and because the Green's function G(0)=-$H^{{-1}}$ contains $s_n^{{-1}}$, these zero modes translate into exponentially large, directionally biased response. A boundary-shaped drive amplitude and detuning, ε_j=ε tanh(j/N0) (mirrored), damps boundary effects and is what lets a steady state exist at all.","core_discovery":"In the coexistence region between the low- and high-density steady states, the chain spontaneously divides into two spatial domains with different non-Hermitian topology. Concretely, with N=40 sites, hopping phase φ=π/3, κ/J=1 and U/J=-2·$10^{-4}$, the local winding number νj=ν(gj,J,Δ̃j) equals 1 across the low-density part of the chain and 0 across the high-density part, with a sharp interface whose position moves from right to left as the drive amplitude ε increases. The nonzero winding is tied to topological amplification: the SVD of the non-Hermitian dynamical matrix develops a topologically protected near-zero singular value, so the zero-frequency Green's function, and therefore the linear response, grows roughly exponentially across the topological segment. Fluctuations ⟨b†_j b_j⟩ peak at the interface, and normalized two-point correlations become long-ranged along the critical line, matching the effective quadratic model built from the steady-state order parameters. The paper presents this as evidence that point-gap non-Hermitian topology emerges from the interplay of interactions, dissipation, and the chiral drive.","pith_inferences":["If the Gaussian closure is accurate, the same mechanism should appear for smaller |U| and for other lattice geometries; a direct check would be to measure the position-resolved gain profile of a driven chain and compare it with the local winding number νj computed from the steady state.","The power-law growth of the critical derivative, d|α*_{N/2}|/dε ∝ N^{3.05}, suggests the coexistence boundary sharpens into a genuine first-order transition in the thermodynamic limit; an exact or tensor-network study of larger chains could test whether the exponent is universal.","Because gauges (i) and (ii) are equivalent, an experiment needs only a constant phase gradient in the drive rather than complex hopping; this turns the prediction into a comparatively simple superconducting-circuit layout.","The paper's local winding-number construction implicitly assumes parameters vary slowly; near the interface, where the density changes sharply, a better characterization may require a nonlocal or scattering-based invariant, which would either confirm or refine the νj=1/0 picture."],"forward_implications":["A chain tuned into the coexistence region should amplify weak input signals exponentially across the low-density topological segment, with gain that grows with system size roughly as e^{N/ξ} because the Green's function is controlled by a protected near-zero singular value.","Varying the drive amplitude ε moves the topological/trivial interface through the chain, so the spatial extent of the amplifying region becomes a tunable parameter and the same device can switch between amplification and transparent regimes.","The interface where the winding number changes is also the locus of maximal photon-number fluctuations and long-range correlations, making the phase boundary observable through noise measurements and correlation spectroscopy.","No such coexistence or amplification appears when the hopping phase is zero (ϕ=0); the chiral drive is therefore the ingredient that generates the topological response.","The necessary ingredients—Kerr nonlinearity, photon loss, a phase gradient, and shaped drive amplitude—are all available in superconducting-circuit arrays, so the predicted phase diagram is in reach of current experiments."],"supporting_citations":[{"why":"Supplies the definition of the winding number for Gaussian systems and the result that ν≠0 yields directional amplification through near-zero singular values of the dynamical matrix.","marker":"[37]"},{"why":"Establishes the topological-amplification mechanism (SVD of the dynamical matrix and chiral symmetry protecting zero modes) that the paper invokes for its Green's-function analysis.","marker":"[10]"},{"why":"Provides the non-Hermitian point-gap topology classification that the paper's winding number is said to belong to.","marker":"[39]"},{"why":"Justifies the Gaussian variational ansatz and the Wick-factorized equations of motion for bosonic systems, the method that generates the paper's steady-state results.","marker":"[41, 42]"},{"why":"Gives the superconducting-circuit implementation scheme (phase gradient plus inhomogeneous drive profile) that the paper relies on for experimental feasibility.","marker":"[43]"},{"why":"Earlier model of topological amplification in driven-dissipative linear photonic lattices that the paper extends to the interacting Bose-Hubbard setting.","marker":"[34]"}],"fun_headline_variants":["Chiral drive turns lossy photon chain into topological amplifier","Non-Hermitian winding number emerges in driven-dissipative chain","Photonic chain splits into topological halves, amplifying signals","Topological amplification in interacting driven-dissipative photons","Steady-state photon chain shows point-gap topology and amplification"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole chain-level result rests on the assumption that the two-point-correlation closure gives the true steady state; the authors verify this exactly only for one site, and for longer chains they substitute a smallness condition that does not guarantee accuracy exactly where the fluctuations are biggest (the interface).","fun_headline_variants_meta":{"raw":{"variants":["Chiral drive turns lossy photon chain into topological amplifier","Non-Hermitian winding number emerges in driven-dissipative chain","Photonic chain splits into topological halves, amplifying signals","Topological amplification in interacting driven-dissipative photons","Steady-state photon chain shows point-gap topology and amplification"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.0006,"raw_usage":{"total_tokens":2784,"prompt_tokens":908,"completion_tokens":1876,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":524,"completion_tokens_details":{"reasoning_tokens":1792}},"tokens_in":524,"tokens_out":1876,"duration_ms":13539,"temperature":1.0,"reasoning_tokens":1792,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T21:12:48.422264+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the exact steady state of a short chain (N=4 to 8) with the same parameters (φ=π/3, κ/J=1, U/J=-2·$10^{-4}$, ε/J≈40) using a method that does not assume Gaussianity—e.g. matrix-product-operator simulation of the Lindblad master equation—and check whether the νj=1/νj=0 domain splitting, the interface fluctuation peak, and the exponential Green's-function growth survive.","supporting_citations":[],"review_version":1}