{"id":"460b307d-8d29-46de-805c-fd01e93c949f","arxiv_id":"2505.09179","paper_version":3,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":7,"one_line_summary":"In a nonlinear Hatano-Nelson model and a microwave metamaterial, increasing the field intensity reverses the non-Hermitian skin effect, switching the edge localization from one end of the lattice to the other, and the third harmonic generated by the nonlinearity inherits the reversed localization.","lead":"This paper shows that raising the input power in a nonlinear microwave lattice can flip the direction in which waves bunch up at the edge, reversing the non-Hermitian skin effect. It also shows that the same nonlinearity generates a third harmonic whose spatial pattern follows the skin localization.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Comparison of driven multi-port measurements to fixed-intensity nonlinear eigenmodes lacks any driven-response model or P→I mapping; the claimed reversal could be a drive/loss effect rather than a skin-mode transition.","rationale":"I read the paper as making two coupled claims: an observable power-controlled localization reversal and its interpretation as a point-gap topological transition. The experimental reversal is supported by port and near-field measurements, but the bridge from those driven measurements to the fixed-intensity eigenmode theory is absent. The reader correctly flags this as the weakest step. My concern does not overturn the qualitative observation; it makes the theoretical interpretation conditional on a driven-response calculation and a P↔I calibration. The topological-phase label is also not fully justified without an invariant, but that is a framing issue, not a separate fatal flaw. I therefore keep the conditional verdict.","tokens_in":12078,"tokens_out":7434,"duration_ms":87965,"concrete_test":"Build a driven coupled-mode simulation of the 11-site lattice using the retrieved parameters (the same values used for Fig. 3e-h), with a coherent source term representing the power splitter at each port and Eq. (2) evaluated with the local steady-state intensities. Sweep input power from -25 to 11 dBm, compute the fundamental-frequency field distribution at the same frequencies as the experiment, and compare quantitatively with Fig. 3b-d. If the measured right-to-left reversal is reproduced without free parameters beyond the stated P↔I calibration, the eigenmode interpretation is validated; otherwise the central comparison fails.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central experimental demonstration (Fig. 3b-d, i-k) is read against eigensolutions of the fixed-intensity problem H(ψ)ψ=Eψ described in Methods, yet the measurements are steady states of an 11-port coherent drive at fixed input power P. The paper never states how P maps to the global intensity I used in the theory, nor does it solve the driven nonlinear problem. This matters because Eq. (2) makes the nonreciprocal coupling depend on local site intensity; under multi-port driving the local intensity profile is set by the drive amplitudes and losses, not by a global I parameter. The agreement in Fig. 3e-h could therefore be qualitative, and the high-power reversal might originate from drive/detuning or saturation effects rather than from a self-consistent skin-mode eigenphase. The 'point-gap topological phase transition' label is additionally unsupported by any computed nonlinear winding invariant, but the missing driven-response bridge is the more immediate threat to the central claim.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript reports a theoretical and experimental study of a nonlinear Hatano-Nelson model with saturable nonreciprocal coupling. The central claim is that increasing the field intensity drives a point-gap topological phase transition in which the non-Hermitian skin effect reverses direction, with modes localized at one end at low intensity and at the opposite end at high intensity. The authors implement the model in a microwave metamaterial where unidirectional coupling is provided by an LNA-based amplifier whose gain saturates with local intensity. They observe the localization reversal in three independent ways: port transmission measurements on an 11-resonator chain, near-field electric-field scans, and third-harmonic generation whose spatial profile follows the skin mode. The theory solves a fixed-intensity nonlinear eigenproblem H(ψ)ψ=Eψ, while the experiment uses driven steady states under continuous-wave excitation. The nonlinear coupling parameters are extracted from a two-resonator transmission measurement and then used in the 11-resonator eigenmode calculation.","tokens_in":12310,"tokens_out":7987,"duration_ms":83605,"significance":"If the claims hold, this would be a valuable advance in nonlinear non-Hermitian topology, showing that intensity alone can switch the direction of the skin effect and can be used for reconfigurable wave manipulation and harmonic generation. The paper is commendable for showing consistency across multiple measurement modalities and for using parameters extracted from a separate two-resonator characterization rather than fitting to the observed reversal. The main risk is that the experiment-to-theory comparison bypasses the driven nonlinear response, so the topological interpretation is not yet fully substantiated. With a driven-response simulation and an explicit input-power-to-intensity calibration, the work could become a convincing demonstration.","major_comments":[{"comment":"The measured steady-state field distributions in Fig. 3b-d and 3i-k are responses of an 11-resonator system to a coherent multi-port drive at fixed input power P, while the comparison in Fig. 3e-h is made to eigensolutions of H(ψ)ψ=Eψ at fixed total intensity I. The manuscript never specifies a mapping between P and I and does not solve the driven nonlinear problem. Without such a driven-response calculation, the agreement is qualitative, and the high-power reversal could in principle arise from drive, loss, or detuning effects rather than from the predicted self-consistent skin-mode eigenphase. Please add a driven coupled-mode simulation with the same retrieved parameters, or provide a clear argument that the uniform 11-port excitation selects the nonlinear eigenmode.","section":"Methods (Nonlinear eigenproblem) and Fig. 3"},{"comment":"The nonlinear coupling model is written as a function of local site intensity I_i, but the two-resonator characterization retrieves κ~2 as a function of input power P. The conversion between P and I_i (e.g., from the two-resonator coupled-mode equations) is not stated. Since t_c sets the saturation scale used in the N-site eigenproblem, this missing calibration weakens the quantitative interpretation of Fig. 3; please specify how I_i was obtained from the measured P or revise the comparison to use a directly calibrated intensity axis.","section":"Eq. (2) and Fig. 2e"},{"comment":"The transition is labeled a point-gap topological phase transition, but no point-gap winding number or equivalent invariant is computed for the nonlinear PBC spectrum. The x̄_c=0 and PBC-gap-closing criteria are suggestive but do not by themselves establish a change of the point-gap winding. Please compute the winding number for the self-consistent PBC solutions or clearly limit the claim to a gap-closing/localization-reversal transition.","section":"Fig. 1b,d-f and Eq. (3)"}],"minor_comments":[{"comment":"The caption statement 'with a fixed t0 = 2.05 (θ = −0.9π)' is ambiguous; please specify explicitly which panel uses which fixed parameter.","section":"Fig. 1 caption"},{"comment":"When connecting the chain into a closed loop for the PBC spectrum, the manuscript does not specify how the nonlinear coupling between sites N and 1 is defined, in particular which site intensity enters the saturable coupling; please clarify.","section":"Theoretical model, PBC treatment"},{"comment":"The term 'Simulated field distributions' is ambiguous; state clearly whether these are eigenmode intensities from the nonlinear eigenproblem or spectra from a driven-response model.","section":"Fig. 3e-g"},{"comment":"The harmonic-generation experiment drives site 1 only, whereas the skin-mode excitation in Fig. 3 uses uniform multi-port drive; please discuss whether the single-port drive affects the interpretation of the third-harmonic localization reversal.","section":"Fig. 4"},{"comment":"Please report parameter uncertainties for the fitted values t0, t∞, tc, and θ, and state how the fit was performed (e.g., least squares on magnitude and phase).","section":"Fig. 2e"}],"recommendation":"major_revision","confidential_remarks":"The missing driven-response bridge is the main obstacle to accepting the experimental demonstration as presented. I would require the authors to add a driven nonlinear simulation using the same fitted parameters and to state the power-to-intensity calibration. The topological-invariant point is also fixable and should be addressed."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe thing to know about this paper is that the experimental result is genuinely new: a saturable nonreciprocal coupling that lets input power alone flip the skin localization from one end of a microwave chain to the other, plus third-harmonic generation whose spatial profile follows the skin modes. The authors do honest parameter characterization on a two-resonator setup, then use those independently fitted parameters to predict the 11-resonator behavior. That is not curve fitting to the reversal, and it gives the central observation real weight.\n\nThe main soft spot is the theory-experiment bridge. The theory solves a fixed-intensity nonlinear eigenproblem, H(psi)psi = E psi, with total intensity I as the control parameter. The experiment drives all 11 sites uniformly at fixed input power P and measures the steady-state response. There is no driven-response calculation and no explicit mapping from P to I. The qualitative agreement across three power levels and three independent measurements (port transmission, near-field scanning, and eigenmode calculations) strongly suggests the effect is real, but without that bridge, one cannot rule out drive-detuning or LNA-saturation effects as contributors to the reversal. This is the issue a referee should push hardest on.\n\nMinor issues: the paper labels the reversal a 'point-gap topological phase transition' but never computes a nonlinear winding number or similar invariant. The point-gap closing is shown, but that is characterization, not a proof of a topological transition. Also, the experimental data lack error bars or repeated trials, a small but notable omission for an experimental paper.\n\nBottom line: this is a solid experimental paper with a clear mechanism and a worthwhile result. It deserves serious peer review, but the revision should include a driven nonlinear simulation or at least a credible argument for why the eigenmode picture applies to the driven measurement. I would likely cite it when discussing nonlinear NHSE experiments.","headline":"A well-executed experiment showing power-controlled reversal of the non-Hermitian skin effect, with a real theory gap between fixed-intensity eigenmodes and the driven measurements.","tokens_in":12832,"tokens_out":3888,"would_cite":true,"duration_ms":42376,"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":"This paper reports that raising the field intensity in a nonlinear non-Hermitian lattice reverses the direction of the non-Hermitian skin effect, switching microwave localization from one end of the sample to the other, with the…","keywords":["non-Hermitian skin effect","point-gap topology","nonlinear topological phase transition","microwave metamaterial","localization reversal","third harmonic generation","saturable nonreciprocal coupling"],"falsifier":"Simulate or measure the actual driven 11-resonator circuit, not its eigenmode model: sweep the continuous-wave input power and record the steady-state field profile; if the measured localization switches at a power that does not correspond to the intensity at which the point gap of the self-consistent spectrum closes, or if the switch is absent in a direct driven-response computation, the claim that the reversal is a point-gap topological transition would fail. A simpler check is to extract the winding of the complex transmission spectrum across the transition and see whether it changes sign at the same power as the field reversal.","tokens_in":11844,"feed_emoji":"🔁","tokens_out":8956,"duration_ms":82971,"temperature":0.7,"pith_summary":"This paper reports a nonlinearity-driven reversal of the non-Hermitian skin effect: in an array with a saturable nonreciprocal coupling, the eigenmodes accumulate at the right edge under low-intensity excitation and at the left edge once the intensity exceeds a threshold. The authors model the effect with a nonlinear Hatano-Nelson chain in which one hopping direction is a complex saturable function of local intensity, solve the nonlinear eigenproblem $\\hat H(\\psi)\\psi = E\\psi$ under a fixed total intensity, and identify the threshold as the closing of the point gap in the periodic-boundary spectrum: a topological phase transition between two nontrivial point-gap phases. They then observe the reversal in an 11-resonator microwave metamaterial at gigahertz frequencies, both through port transmission measurements and near-field scans. The same power-controlled switch appears in the spatial profile of third-harmonic signals generated by the skin modes. A sympathetic reader would therefore take the paper's contribution to be a demonstration that the boundary on which non-Hermitian wave energy accumulates can be chosen purely by pump power.","feed_headline":"Power alone flips which end of a metamaterial stores waves","feed_subtitle":"In a nonlinear microwave chain, intensity alone reverses where skin modes localize; the third harmonic follows.","key_machinery":"The load-bearing object is the nonlinear Hatano-Nelson model with a saturable, complex, nonreciprocal coupling: a one-dimensional tight-binding chain whose rightward hopping is the sum of a fixed reciprocal coupling and an intensity-dependent term $\\tilde\\kappa_{2,i}(I_i)$, while the leftward hopping is purely reciprocal. The saturable term makes the direction of the dominant hopping a function of field intensity: at small $I_i$ the rightward path exceeds the leftward one, and at large $I_i$ the saturated nonreciprocal part shrinks so the balance tilts leftward. The argument then identifies the point-gap winding of the periodic-boundary spectrum as the quantity that changes at the phase transition, because the reversal occurs exactly where this gap closes. The numerical instrument is a self-consistent nonlinear eigensolver with the fixed total intensity constraint, and the experimental instrument is an amplifier-plus-varactor coupling circuit that produces the required nonlinear nonreciprocal hopping.","core_discovery":"In the model, the lattice Hamiltonian is $\\hat H=\\sum_i f_0 \\hat c_i^\\dagger \\hat c_i + \\sum_i (t_{l,i}\\hat c_i^\\dagger \\hat c_{i+1}+t_{r,i}\\hat c_{i+1}^\\dagger \\hat c_i)$ with reciprocal hopping in both directions and an extra rightward hopping $t_{r,i}=\\kappa_{1,i}+\\tilde\\kappa_{2,i}$, where the nonlinear nonreciprocal coupling takes the saturable form $\\tilde\\kappa_{2,i}(I_i)=((t_0-t_\\infty)/(1+I_i/t_c)+t_\\infty)e^{i\\theta}$ and shrinks as the local intensity grows. Solving the self-consistent nonlinear eigenproblem at fixed total intensity $I=\\sum_i|\\psi_i|^2$ shows that the average mode position $\\bar{x}_c$ passes through zero at a threshold intensity, and that this threshold coincides with the closing of the point gap of the periodic-boundary spectrum. In other words, the winding of the complex eigenvalues around a point changes sign, so the point-gap topology and the associated skin localization both reverse even though no lattice parameter is changed. The experimental chain of 11 microwave resonators realizes the required unidirectional coupling with amplifiers whose gain saturates with power; measured port voltages and near-field maps show right-edge localization at low input power, a nearly uniform distribution at intermediate power, and left-edge localization at high input power, and the third harmonic at $3f_{\\rm in}$ follows the same spatial switch.","pith_inferences":["If the self-consistent eigenmode picture is correct, a direct driven-response simulation that maps input power $P_{\\rm in}$ to the total intensity $I$ should reproduce the measured transition power; that mapping is an implicit assumption of the paper and is the natural next test.","The same saturable-nonreciprocity mechanism could be embedded in two-dimensional lattices, where the point-gap winding is replaced by higher-dimensional invariants; one would then expect the localization edge or corner itself to switch with intensity, which is testable with the same amplifier-varactor building block.","Because the third harmonic inherits the skin profile, a power-switchable frequency converter could be built in which both the conversion efficiency and the direction of the emitted harmonic field are controlled by the pump level, even though no geometric reconfiguration occurs."],"forward_implications":["Raising input power alone moves the localization of all skin modes from one boundary of the array to the other, with an almost delocalized distribution near the transition.","The point gap in the periodic-boundary spectrum closes exactly where the average mode position $\\bar{x}_c$ crosses zero, so the transition is of point-gap topological type rather than a line-gap transition.","Third-harmonic fields generated by the nonlinearity inherit the spatial profile of the fundamental skin modes, so the spatial distribution of the harmonic signal can be switched by input power.","The amplifier-based saturable coupling is compact enough to realize nonlinear non-Hermitian lattices in one and two dimensions, so the scheme can serve as a platform for other nonlinear and non-Hermitian models."],"supporting_citations":[{"why":"Supplies the linear Hatano-Nelson model whose nonlinear version is the subject of the paper.","marker":"[64]"},{"why":"Establishes the edge-state and topological-invariant framework for non-Hermitian systems used to identify the point-gap transition.","marker":"[20]"},{"why":"Attributes the skin effect to point-gap topology, the classification the reported transition belongs to.","marker":"[24]"},{"why":"Connects point-gap winding numbers to skin-mode localization direction, the diagnostic used to detect the reversal.","marker":"[25]"},{"why":"Provides the H-shaped microwave resonator design from which the experimental chain is built.","marker":"[65]"},{"why":"Earlier demonstration that nonlinearity controls non-Hermitian topological phase transitions, which the paper extends from line-gap to point-gap phases.","marker":"[54]"}],"fun_headline_variants":["Intensity flips skin-effect localization in nonlinear metamaterial","Power alone flips wave edge in metamaterial","Nonlinearity reverses non-Hermitian skin effect","Saturable gain flips skin-mode direction","Nonlinear microwave chain switches wave edge"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The steady-state field distributions measured under continuous-wave drive at given input powers are assumed to be the same as the nonlinear eigenmode solutions at a fixed total intensity, but the paper does not calculate the driven response or connect input power to that intensity.","fun_headline_variants_meta":{"raw":{"variants":["Intensity flips skin-effect localization in nonlinear metamaterial","Power alone flips wave edge in metamaterial","Nonlinearity reverses non-Hermitian skin effect","Saturable gain flips skin-mode direction","Nonlinear microwave chain switches wave edge"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000486,"raw_usage":{"total_tokens":2448,"prompt_tokens":1050,"completion_tokens":1398,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":666,"completion_tokens_details":{"reasoning_tokens":1326}},"tokens_in":666,"tokens_out":1398,"duration_ms":10522,"temperature":1.0,"reasoning_tokens":1326,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T21:38:33.769969+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Simulate or measure the actual driven 11-resonator circuit, not its eigenmode model: sweep the continuous-wave input power and record the steady-state field profile; if the measured localization switches at a power that does not correspond to the intensity at which the point gap of the self-consistent spectrum closes, or if the switch is absent in a direct driven-response computation, the claim that the reversal is a point-gap topological transition would fail. A simpler check is to extract the winding of the complex transmission spectrum across the transition and see whether it changes sign at the same power as the field reversal.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Earlier demonstration that nonlinearity controls non-Hermitian topological phase transitions, which the paper extends from line-gap to point-gap phases."}],"review_version":1}