{"id":"af155ba9-284d-4eb0-acce-b0e791d52c84","arxiv_id":"2512.01645","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"Localised driving of a dissipative Lieb lattice creates strong antibunching on adjacent dark B sites through interference, extending unconventional photon blockade to extended lattices.","lead":"In a driven-dissipative Bose-Hubbard model on a Lieb lattice, the authors show that driving just one site can turn the usually dark corner sites into sources of antibunched light. The result suggests a practical route to producing single-photon-like correlations in polariton micropillar arrays with weak nonlinearities.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The flat-band control in Sec. III E changes Δ_C on the driven site, confounding flat-band destruction with an off-resonant drive; the claimed link between antibunching and flat-band interference is not established.","rationale":"After reading the paper in good faith, the central proposal — that localised driving of a C site in a DDBH Lieb lattice generates UPB-like antibunching on a neighbouring dark B site — is supported by the 3-site analytical calculation and the numerical scans. The positive-P method is established in the authors' prior work, and some error bars are given; while convergence tests are missing, this is not the most acute weakness. The most load-bearing issue is the flat-band control in Sec. III E. The observed increase of g^(2)(0) from 0.449 to 0.948 when Δ_C is changed from -0.2γ to -5.0γ is used to argue that destroying the flat band destroys the antibunching. However, this parameter change also detunes the driven C site from the laser, which reduces the effective drive and pushes the system toward the coherent (weakly driven) limit. A change in drive strength alone can produce a similar rise in g^(2)(0), so the control does not isolate flat-band physics. A definitive test is to break the flat band while keeping Δ_C fixed; this would discriminate between the resonance effect and the interference effect. The paper's own caveat that the single-unit-cell chain (no flat band) shows the effect further underscores that the flat-band link, as tested, is not causal. Therefore the central numerical result (e.g., g^(2)=0.468 at site 3B) stands, but the claimed connection to flat-band interference is unproven. This warrants maintaining the conditional verdict, pending the proposed control.","tokens_in":16297,"tokens_out":12425,"duration_ms":123968,"concrete_test":"Repeat the Sec. III E simulation with a flat-band-breaking perturbation that preserves Δ_C = -0.2γ on all C sites. For instance, add an on-site energy δ to the A sites (Δ_A = -0.2γ + δ) or introduce a weak next-nearest-neighbor hopping between C sites, and scan δ to, e.g., δ=±γ. If g^(2)(0) on site 11B remains ≈0.45, the original Δ_C control was compromised by resonance effects; if it rises to ≈0.95, the flat-band link is supported.","verdict_should_be":"UNCHANGED","load_bearing_attack":"To support the connection between the enhanced antibunching on the dark B site and the destructive interference that creates the Lieb flat band, Sec. III E compares a lattice with Δ_C = Δ = -0.2γ to one with Δ_C = -5.0γ (20 unit cells, driven on site 11C). This control simultaneously (i) destroys the flat band and (ii) shifts the driven C site 5γ away from the laser resonance (the laser is tuned so that A and B sites have Δ=-0.2γ). With the C site far off resonance, the effective drive that reaches the B site through the lattice is strongly reduced; the occupation of the dark site drops and the second-order correlation rises to 0.948, close to the coherent value. This is the expected behavior of any weakly driven nonlinear system at low occupancy, and does not by itself implicate the flat band. No alternative flat-band-breaking control that keeps the C sites on resonance (e.g., detuning the A sublattice or modifying hopping) is provided. Since the paper's interpretation (common destructive-interference origin of flat band and antibunching) rests on this single control, the evidence for that claim is confounded.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the driven-dissipative Bose-Hubbard model on Lieb lattices with a coherent drive applied locally to a C site. It first derives, in the weak-driving limit, analytic conditions for optimal antibunching on the middle B site of a three-site chain (Eqs. (6)–(8) and Appendix A). It then uses positive-P stochastic simulations to show that on quasi-1D Lieb lattices (and in one 2D example) the B site adjacent to the drive exhibits strong antibunching, e.g. g^(2)(0)=0.468 for U=0.1γ, Δ=-0.2γ, F=J=3.0γ in the 5-cell system. The authors optimize parameters for observability in polariton micropillar experiments, introduce a background drive to recover antibunching at lower J, and present a flat-band control in Sec. III E that they interpret as evidence for a common destructive-interference origin of the flat band and the enhanced antibunching.","tokens_in":16684,"tokens_out":6652,"duration_ms":72572,"significance":"If the numerical results are reliable, this is a useful extension of unconventional photon blockade from dimers to extended lattice geometries: it demonstrates that local driving can create non-classical correlations on selected dark sites of a Lieb lattice, and it identifies practical advantages (longer oscillation period in g^(2)(τ), reduced side peaks) that are relevant to polariton micropillar experiments. The analytic three-site derivation is transparent and internally consistent, and the positive-P method is, in principle, an exact stochastic technique for this model. The main qualifications are the absence of convergence/error-bar information for the numerics and the weak control used to connect the effect to flat-band interference.","major_comments":[{"comment":"The control for the flat-band claim is confounded. Changing Δ_C from -0.2γ to -5.0γ simultaneously destroys the flat band and moves the driven C site 5γ off resonance relative to the laser tuned to A/B. The reduced effective drive lowers the B-site occupation; at low occupancy any weakly driven nonlinear system tends to g^(2)(0)→1, so the observed 0.948 does not specifically implicate flat-band interference. Since Sec. IV uses this result to argue for a common destructive-interference origin, please supply a control that breaks the flat band while keeping the C sites on resonance (e.g., detuning the A sublattice, or a staggered hopping), or otherwise match the B-site occupation across the comparison.","section":"Sec. III E, Fig. 10"},{"comment":"No convergence tests are reported for the positive-P simulations. The paper states that 2000 samples are used, but most g^(2)(0) values in parameter scans are plotted without error bars, and some key numbers (e.g., g^(2)(0)=0.468 in Fig. 6) are quoted without uncertainty. Because the parameter choices in Secs. III C/D and Appendix B are optima of these curves, it is essential to know whether the reported minima are significant. Please add error bars or statistical bootstrap uncertainties, and provide a convergence check (e.g., trajectory number, or comparison with an exact master equation/DMRG calculation for a small system).","section":"Sec. II; Figs. 13–17"},{"comment":"The analytic optimal condition is derived in the weak-driving limit (amplitude hierarchy A2, linearized density matrix A3). The numerical demonstration of the 3-site optimum uses F=γ (Fig. 3), and the extended lattices use F=J=3γ, i.e., outside the assumed weak-driving regime. The agreement at F=γ is encouraging, but the paper does not quantify the range of validity of the analytic formulas or show that the same two-boson interference mechanism is operative at the drive strengths of the main results. Please test the ansatz against exact numerics as a function of F, or explicitly restrict the analytic-optimum claim to weak driving.","section":"Appendix A; Sec. III A"}],"minor_comments":[{"comment":"The abstract states that positive-P allows one to calculate correlations 'exactly' for extended lattices. With a finite number of trajectories (2000), the results are only exact in the limit of large samples and sufficient dissipation; please soften the wording or state the finite-sample nature explicitly.","section":"Abstract; Sec. II"},{"comment":"The text says that for sites A and C the g^(2)(τ) values are 'approximately equal' at all τ, but this is not shown or explained. Please either display these curves or give a brief reason for the equivalence.","section":"Fig. 3"},{"comment":"For the flat-band comparison, please report the B-site occupations n for both Δ_C cases. The interpretation of g^(2)(0)≈0.948 as 'weak antibunching' depends strongly on whether the occupation is also strongly reduced in the off-resonant case.","section":"Sec. III E"},{"comment":"Please specify the time step δt, the number of steady-state time samples N_t, and the total integration time used in the positive-P simulations, in addition to the total trajectory count.","section":"Sec. II; Appendix B"},{"comment":"The 'period' of the oscillations in g^(2)(τ) is described as the gap between the two maxima adjacent to τ=0, but this quantity is not formally defined. Please define it precisely, especially because it is used to claim a doubling of the period at lower J.","section":"Figs. 7 and 9"},{"comment":"The simulations use the xmds2 package, but no code or data availability statement is included. Providing scripts or data would strengthen reproducibility, particularly for the parameter scans in Appendix B.","section":"Reproducibility"}],"recommendation":"major_revision","confidential_remarks":"The core idea is interesting and the 3-site analytic calculation appears sound, but the manuscript currently needs two things before I can recommend acceptance: (i) an unambiguous control experiment for the flat-band claim, since the Δ_C=-5γ control is confounded by detuning the driven site, and (ii) error bars or convergence tests for the positive-P results that underpin all extended-lattice claims. The weak-driving ansatz is used at F=J=3γ in the main text, which is another point that should be addressed explicitly. The paper is not fatally flawed, but the numerical evidence needs to be made more robust."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Worth a look for the main numerical result: localised driving of a C site in a Lieb lattice produces strong antibunching on the adjacent dark B site (g2(0) ≈ 0.45 for U=0.1γ, J=F=3γ), and the effect is extended to quasi-1D and 2D lattices. The three-site analytic optimisation in Appendix A is a clean extension of Bamba et al., and the background-drive trick that doubles the g2(τ) period is a useful practical tweak for polariton experiments. These are real contributions.\n\nThe soft spots are real too. The flat-band control in Sec. III E changes Δ_C on the driven site to -5γ while the laser stays tuned to the other sublattices. That simultaneously destroys the flat band and pushes the driven site far off resonance. The resulting loss of antibunching (g2 rises to 0.948) is exactly what you'd expect from a weakly excited nonlinear system, so it does not implicate the flat band. The authors state they only claim a common origin, but the experiment they run cannot support that claim. A better control would detune the A sublattice or modify hopping while keeping the driven C site resonant.\n\nAlso, most g2 scans lack error bars, no code or data are deposited, and the positive-P convergence is asserted rather than shown—2000 trajectories may be fine, but the reader has to take it on trust. And the extended-lattice parameters are optimised against the same g2 target, so the reported values are demonstrations, not predictions; that's fine as long as it's framed that way.\n\nWho should read this: anyone working on UPB, flat-band correlations, or driven-dissipative polariton lattices. It deserves a serious referee, and the issues are fixable. I'd send it, with the flat-band section either revised or reframed as speculative.","headline":"A genuinely new UPB-in-lattices result with a solid three-site analytic core, but the flat-band connection is not proven by the confounded control in Sec. III E.","tokens_in":17086,"tokens_out":4759,"would_cite":true,"duration_ms":49451,"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":"Locally driving a single site in a dissipative Lieb lattice can turn the destructive interference that creates flat bands into a source of antibunched light on a neighbouring dark site.","keywords":["unconventional photon blockade","Lieb lattice","flat band","antibunching","driven-dissipative Bose-Hubbard","positive-P method","polariton micropillars","second-order correlations"],"falsifier":"Compute g^(2)(0) on site 3B of the five-unit-cell chain using an independent method (e.g., a matrix-product-state based master equation solver or exact diagonalization for a reduced lattice) at U=0.1γ, Δ=-0.2γ, F=J=3.0γ; if the result deviates significantly from the reported 0.468, the positive-P sampling is biased. Alternatively, in a Lieb micropillar lattice, measure g^(2)(0) on the dark site adjacent to a locally driven C site while detuning C sites to destroy the flat band; the predicted jump from ~0.45 to ~0.95 either occurs or it does not.","tokens_in":16234,"feed_emoji":"⚛️","tokens_out":4003,"duration_ms":42410,"temperature":0.7,"pith_summary":"This paper shows numerically that in a driven-dissipative Bose-Hubbard model on a Lieb lattice, applying a coherent drive to one C site produces strong antibunching (g^(2)(0)=0.468) on the adjacent dark B site. The effect shares its physical origin with unconventional photon blockade: destructive interference cancels the two-boson amplitude on the target site. By breaking the flat band through a large detuning on C sites, the antibunching essentially disappears (g^(2)(0) rises to 0.948), demonstrating that the same interference underlies both phenomena. The authors optimize parameters for polariton micropillar experiments and show that adding a background drive can double the oscillation period of the correlations, making the effect easier to observe with finite-time-resolution detectors. If correct, this provides a scalable route to generating non-classical photon statistics in extended open quantum lattices without requiring strong nonlinearities.","feed_headline":"Local drive makes Lieb-lattice dark sites emit antibunched light","feed_subtitle":"The same interference that creates flat bands can be steered to generate non-classical light on a chosen dark site.","key_machinery":"Unconventional photon blockade (UPB): the cancellation of the two-boson occupation amplitude on a target site by destructive interference between different excitation paths, here enabled by the same geometric frustration that makes B sites dark in a Lieb lattice. The analytical optimization for a single unit cell (three-site chain) is derived from a weak-driving two-boson ansatz, setting the C_020 amplitude to zero and yielding closed-form equations for the optimal detuning and hopping.","core_discovery":"The paper claims that in a locally driven Lieb lattice, the destructive interference which produces the flat band and dark B sites can be repurposed to suppress double occupation of a chosen dark site, yielding enhanced antibunching via a mechanism analogous to unconventional photon blockade. Concretely, for a five-unit-cell quasi-1D chain with drive F=J=3γ, U=0.1γ, and Δ=-0.2γ, the adjacent site 3B reaches g^(2)(0)=0.468, while A and C sites remain nearly coherent. Destroying the flat band by setting Δ_C=-5γ reduces the effect to g^(2)(0)=0.948, directly linking the antibunching to the same interference that creates the flat band. The work also demonstrates the effect in a 2D Lieb lattice a","pith_inferences":["The same interference mechanism could be exploited to generate entanglement or squeezed states between multiple dark sites in larger two-dimensional Lieb lattices, where multiple B sites are simultaneously addressed.","If the flat-band connection is robust, disorder or perturbations that break the flat band should degrade antibunching, making g^(2)(0) on dark sites a sensitive diagnostic for flat-band physics in experiments.","The positive-P results for extended lattices could be cross-checked against tensor-network or exact diagonalization methods on small systems; agreement would strengthen the reliability of the stochastic approach for multi-time correlations in open lattices.","Frustrated geometries other than Lieb lattices (e.g., kagome or pyrochlore) that also host dark sites may support analogous UPB-like antibunching under localized driving."],"forward_implications":["A single local drive can generate antibunching on a preselected site of an extended lattice, offering a scalable alternative to two-site UPB devices.","The effect provides experimentally accessible parameters for polariton micropillar lattices, with g^(2)(0) below 0.5 and occupations above 0.1 at U/γ=0.1.","Adding a background drive on other C sites allows the hopping J to be halved while maintaining antibunching, doubling the oscillation period of g^(2)(τ) and easing detector time-resolution requirements.","The link between flat-band interference and antibunching implies that measuring g^(2)(0) on dark sites can act as a probe of flat-band integrity in real lattices.","Local driving schemes can be extended to other lattice geometries and to other correlation functions, broadening the toolkit for quantum correlations in driven-dissipative systems."],"fun_headline_variants":["Local drive repurposes flat-band interference for antibunching","Dark-site antibunching via local drive in Lieb lattices","Lieb lattice: local drive yields antibunching on flat-band dark sites","Turning flat-band frustration into single-photon antibunching","Local drive creates antibunched photon source at Lieb dark site"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The central numerical claim rests on the positive-P method with 2000 trajectories yielding converged, unbiased steady-state and second-order correlation functions for the extended lattices, with no convergence tests or code provided; the analytical parameter optimization additionally assumes weak driving while the simulations operate at F≈γ.","fun_headline_variants_meta":{"raw":{"variants":["Local drive repurposes flat-band interference for antibunching","Dark-site antibunching via local drive in Lieb lattices","Lieb lattice: local drive yields antibunching on flat-band dark sites","Turning flat-band frustration into single-photon antibunching","Local drive creates antibunched photon source at Lieb dark site"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000985,"raw_usage":{"total_tokens":4001,"prompt_tokens":718,"completion_tokens":3283,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":462,"completion_tokens_details":{"reasoning_tokens":3194}},"tokens_in":462,"tokens_out":3283,"duration_ms":20581,"temperature":1.0,"reasoning_tokens":3194,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-03T19:07:53.948459+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute g^(2)(0) on site 3B of the five-unit-cell chain using an independent method (e.g., a matrix-product-state based master equation solver or exact diagonalization for a reduced lattice) at U=0.1γ, Δ=-0.2γ, F=J=3.0γ; if the result deviates significantly from the reported 0.468, the positive-P sampling is biased. Alternatively, in a Lieb micropillar lattice, measure g^(2)(0) on the dark site adjacent to a locally driven C site while detuning C sites to destroy the flat band; the predicted jump from ~0.45 to ~0.95 either occurs or it does not.","supporting_citations":[],"review_version":1}