{"id":"11b1f114-a4b7-4484-9e2f-7a73d5bd8841","arxiv_id":"2505.15343","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"A self-consistent variational calculation of the driven 2D Hubbard model finds multiphoton-tunneling cooperation and dephasing-induced linear absorption in subgap-pumped Mott insulators, along with real-time feedback from gap renormalization.","lead":"This paper models strong laser driving of antiferromagnetic Mott insulators with a time-dependent variational method and finds two new photoexcitation regimes in addition to the usual multiphoton and tunneling mechanisms. It also shows that the shrinking Mott gap feeds back into carrier generation in real time, and proposes acoustic phonons and momentum-resolved carrier maps as experimental probes.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The load-bearing concern is the variational ansatz itself: the unbenchmarked mean-field structure of Eq. (5) could generate the claimed cooperative-regime thresholds and gap-feedback effects, so the central new physics rests on an uncontrolled approximation.","rationale":"I read the paper as a legitimate time-dependent mean-field study: it derives a self-consistent SDW-type evolution, recovers the known multiphoton and dc-tunneling Keldysh limits, and makes a concrete experimental proposal via acoustic phonons. Those are real strengths, and the phonon probe argument is largely independent of the variational subtlety. However, the most novel claims—the cooperative multiphoton-tunneling region and the two gap-feedback effects (saturation and threshold lowering)—are established only within the ansatz of Eq. (5), which the paper itself does not validate against any exact method. The reader identified exactly this as the weakest assumption, and I agree. The analytic argument in Appendix D3b is explicitly incomplete, so the numerics are the sole evidence for the cooperative regime; those numerics inherit all limitations of the factorized two-state-per-k density matrix. A small-system exact benchmark is the natural, decisive check: it is feasible with current MPS/Lanczos methods, and it would settle whether the thresholds and feedback survive beyond mean-field. Since the concern is real but addressable, and since the paper does contain recoverable analytic limits and an independent phonon application, the correct verdict remains CONDITIONAL rather than REJECT; no change to the reader's assessment is needed.","tokens_in":48634,"tokens_out":5842,"duration_ms":58733,"concrete_test":"Benchmark the variational results against numerically exact time evolution on a small 2D Hubbard system. Concretely, run time-dependent MPS on a 4×8 Hubbard cylinder at half-filling (or time-dependent Lanczos on a tilted 4×4 cluster) with the same U = 1.93 eV, τ = 189 meV, and Gaussian pulse envelope of 100 fs, for ω_d/Δ0 ∈ [0.1, 0.5] and eE_p a/Δ0 ∈ [0.05, 0.4]. Compute total n_qp, the multiphoton index μ of Eq. (17), and the momentum-resolved distribution Δn_qp of Eq. (18), and overlay these on the phase diagrams of Figs. 2 and 5. Because Γ is small, first compare in the coherent limit Γ→0; then, if feasible, add local dephasing via quantum trajectories with Γ = 1.64 meV. A second part of the same check: compare the time-resolved order parameter S(t) at (ω_d/Δ0 = 0.34, eE_p a/Δ0 = 0.3) with Fig. 8.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim—two regimes beyond Keldysh plus dynamical gap feedback—rests on the time-dependent Gaussian ansatz in Eq. (5) and the reduced evolution equations (6). This ansatz factorizes the density matrix over momentum sectors, keeps only zero-total-momentum doublon-hole pairs, collapses all decoherence into a single Markovian rate Γ, and neglects intraband relaxation during the pump. In the cooperative region, eEa ≫ ℏω_d, the vector potential sweeps the full Brillouin zone many times per cycle and the final n_qp becomes large; this is precisely the regime where inter-sector correlations, doublon-doublon scattering, and non-Markovian effects should be largest. A single collective gap S(t) is the only channel coupling different k-sectors, so the sharp Δ0/n thresholds seen above the Keldysh line in Fig. 2(a), the momentum-space sharpening in Fig. 4(i,l), and the real-time saturation/threshold-lowering in Figs. 8-9 could all be mean-field artifacts. The analytic derivation of the cooperative tunneling rate in Appendix D3b is explicitly incomplete ('a more precise calculation needs to count all possible combinations'), and the numerical observation is not backed by any exact benchmark: no DMFT, MPS, or exact-diagonalization comparison is provided. The incoherent low-field regime is less vulnerable because it follows from a controlled Γ/Δ expansion, but it does not support the two new high-field regimes.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript studies photoexcitation of the half-filled two-dimensional Hubbard model under subgap ac driving, using a time-dependent Gaussian variational ansatz built on the SDW mean-field state. It derives coupled evolution equations for the doublon-hole coherences and the self-consistent Néel order parameter, then analyzes the rigid-band limit to obtain approximate analytic rates: multiphoton excitation (Eq. 12), tunneling (Eq. 13), and a new incoherent linear-absorption pathway (Eq. 16). On this basis it claims a four-regime landscape in the frequency-field plane, including a cooperative region in which the field dependence is tunneling-like while frequency thresholds remain at integer Δ/ω_d, and a low-field incoherent region. The numerical solution of the full variational equations for cuprate-like parameters is used to study gap-renormalization feedback, saturation and threshold-lowering dynamics, momentum-resolved carrier distributions, and coherent acoustic phonon generation as a probe of photocarrier density.","tokens_in":48904,"tokens_out":4449,"duration_ms":45503,"significance":"If correct, the paper would provide a unified variational framework that recovers the known Keldysh multiphoton/tunneling limits while predicting additional experimentally accessible regimes and real-time gap-feedback effects. The analytic derivations are careful in the rigid-band sector, the incoherent formula (D49) is explicit, and the parameters are physical inputs rather than back-fitted free parameters; the recovery of the dc tunneling and multiphoton functional forms of Ref. [23] is a genuine strength. The momentum-distribution metric of Fig. 5 and the acoustic-phonon probe of Sec. V are potentially useful and testable. However, the paper's central new claim, the cooperative regime, rests on an incomplete analytic derivation of γ(ω_d) and on an unbenchmarked variational ansatz in exactly the strong-drive regime where the ansatz is least controlled; the significance is therefore conditional on resolving those issues.","major_comments":[{"comment":"The cooperative regime, which is the paper's central new claim, is encoded in the frequency-dependent tunneling rate γ(ω_d) appearing in Eq. (13). However, Appendix D3b explicitly states that 'a more precise calculation of this enhancement needs to count all possible combinations of frequencies {n_i} satisfying the generalized resonance, rather than just the case n_1 ∼ n_2 ∼ ... ≡ n considered above'. No closed expression for γ(ω_d) is provided; the numerical fits in Fig. 3c extract the slope ex post, so the claimed coexistence of tunneling field dependence and multiphoton frequency thresholds is not an analytically established prediction. Please complete the calculation of γ(ω_d) from the generalized resonance condition (D36), or replace the claim with a direct numerical verification that is not based on fits to the asserted form.","section":"Section II B and Appendix D3b, Eq. (13)"},{"comment":"The variational density matrix factorizes over momentum sectors, keeps only zero-total-momentum doublon-hole pairs, collapses decoherence into a single Markovian rate Γ, and neglects intraband relaxation during the pump. No benchmark against numerically exact methods (DMFT, MPS, or exact diagonalization) is provided. The cooperative regime eEa ≫ ℏω_d is precisely where the vector potential sweeps the full Brillouin zone and the final carrier density becomes large, so inter-sector correlations, doublon-doublon scattering, and non-Markovian effects are expected to be strongest; the sharp thresholds in Fig. 2 and the saturation/threshold-lowering effects in Figs. 8 and 9 could therefore be artifacts of the ansatz. I ask for a concrete benchmark on a small cluster or 1D chain, or an explicit argument identifying a controlled limit in which the ansatz becomes exact, before the high-field predictions are presented as quantitative.","section":"Section II A, Eq. (5), and Appendix C"}],"minor_comments":[{"comment":"The footnote contains a duplicated phrase: 'At first, it may seem surprising that the regionsAt first, it may seem surprising that the regions in Fig. 2b...'.","section":"Footnote [58]"},{"comment":"There is a typo: 'intergral over δ' should be 'integral over δ'.","section":"Appendix D1, Eq. (D11)"},{"comment":"The caption reads 'at frequecy ω_d = Δ0/10'; this should be 'frequency'.","section":"Figure 3 caption"},{"comment":"The text switches between the continuous-wave amplitude E_0 used in the analytic rates and the peak pulse field E_p used in the numerics; the statement that E_p should replace E_0 is physically reasonable, but the approximation should be justified more explicitly, since the pulse envelope modifies the effective duration available for tunneling.","section":"Section III B"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is within the scope of the journal and the authors have made a serious attempt at a difficult problem; I do not see grounds for rejection. My recommendation of major revision is driven by two load-bearing gaps: the incomplete analytic derivation of the cooperative tunneling rate γ(ω_d), and the absence of any benchmark for the variational ansatz in the strong-drive regime where the new phenomena are claimed. Both issues are fixable in principle within the manuscript's scope. The citation pattern, including self-citations, appears reasonable."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe paper to know: a time-dependent Gaussian variational treatment of subgap pumping in the 2D Hubbard model at half filling. It recovers the familiar Keldysh multiphoton/tunneling crossover, then claims two regimes beyond it: a broad cooperative region where the field dependence is tunneling-like but frequency thresholds appear at Δ/ω_d integer, and a low-field incoherent regime controlled by dephasing. It also studies how the self-consistently evolving gap feeds back into the generation rate, producing real-time saturation or threshold lowering.\n\nThe genuinely new content is the self-consistent gap feedback, the finite-frequency tunneling enhancement, and the incoherent regime. The derivations are mostly careful: the evolution equations, the self-consistency condition, and the incoherent formula (D49) are well posed; the multiphoton and dc tunneling rates match Oka's results, which is a good sanity check. The momentum-space diagnostics and the acoustic-phonon proposal are concrete and experimentally useful.\n\nThe soft spot is real and localized. The cooperative regime rests on an analytic derivation that the authors themselves flag as incomplete — Appendix D3b says a more precise calculation must count all possible frequency combinations. The numerics supporting the sharp thresholds in that regime come entirely from the variational mean-field ansatz (5), which factorizes over momentum sectors and keeps only zero-total-momentum doublon-hole pairs with a single Markovian Γ. In the strong-drive regime, eEa≫ω_d, that is precisely where inter-sector correlations and non-Markovian effects should matter most. There is no DMFT, MPS, or ED benchmark. So the cooperative-regime boundaries and the gap-feedback features could shift or disappear under a more exact treatment. The incoherent regime is on firmer footing, because it follows from a controlled Γ/Δ expansion and matches the numerics there.\n\nThis is not a fatal flaw; the paper is well reasoned, and the Keldysh recovery shows the method is not vacuous. But the new physics is claimed at the edge of the approximation's validity. The fix is straightforward in principle: one exact benchmark in the strong-field regime plus a completed D3b.\n\nFor a reader working on driven Mott insulators or ultrafast cuprate experiments, this is worth a reading-group slot and a serious referee, but the referee should insist on those benchmarks before the cooperative regime is taken as established.\n\nMy recommendation: engage with it — peer review, conditional.","headline":"A careful variational treatment of subgap pumping in Mott insulators that plausibly recovers the known Keldysh limits and adds two new regimes, but the central cooperative-regime claim rests on an explicitly incomplete analytic derivation and an unbenchmarked mean-field ansatz.","tokens_in":49520,"tokens_out":2142,"would_cite":true,"duration_ms":20195,"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":"Subgap laser pumping of a Mott insulator yields four photoexcitation regimes, with the Mott gap feeding back on its own carrier generation in real time.","keywords":["Mott insulator","Hubbard model","Keldysh crossover","multiphoton excitation","Landau-Zener tunneling","gap renormalization","time-dependent variational method","doublon-hole pairs"],"falsifier":"Benchmark the same driven half-filled 2D Hubbard model ($U=1.93$ eV, $\\tau=189$ meV, 100 fs Gaussian pump) against a method that does not assume the ansatz, such as dynamical mean-field theory, tensor networks, or time-dependent exact diagonalization on small clusters, and compare total photocarrier density, momentum distributions, and real-time saturation curves. If the exact calculation shows no persistent multiphoton thresholds above the Keldysh line, no saturation plateau while the pump is still on, or no sharp onset of absorption when the transiently renormalized gap crosses $n\\omega_d$, the central claim fails.","tokens_in":48377,"feed_emoji":"⚡","tokens_out":12456,"duration_ms":102765,"temperature":0.7,"pith_summary":"The paper claims that strong subgap pumping of an antiferromagnetic Mott insulator is governed by four photoexcitation mechanisms, not just the familiar two. A time-dependent variational treatment of the half-filled 2D Hubbard model recovers the conventional multiphoton and Landau-Zener tunneling rates on either side of the Keldysh crossover, and adds a broad cooperative region where the generation rate has tunneling field dependence yet shows multiphoton frequency thresholds, plus a low-field regime of incoherent linear absorption controlled by the dephasing rate. The central new element is real-time feedback: the photocarriers themselves suppress the Mott gap, which in turn slows or accelerates further carrier production, producing saturation on resonance and effective threshold lowering below resonance. If correct, the framework unifies single-shot rate calculations with steady-state descriptions, and identifies the momentum distribution of excited quasiparticles and coherent acoustic phonons as experimentally accessible fingerprints of the active mechanism.","feed_headline":"Mott-gap feedback reshapes subgap photoexcitation into four regimes","feed_subtitle":"Multiphoton and tunneling coexist in a new cooperative regime, and the shrinking gap saturates or accelerates carrier generation.","key_machinery":"The central object is a time-dependent Gaussian variational density matrix for the antiferromagnetic (spin-density-wave) state of the Hubbard model, factorized over momentum sectors in the reduced Brillouin zone and restricted to doublon-hole pairs at zero total momentum. Its evolution is governed by a time-dependent Bogoliubov angle $\\varphi_k(t) = \\arctan[2\\epsilon_k(t)/(US(t))]$, a quasiparticle dispersion $E_k(t) = \\sqrt{\\epsilon_k^2(t) + (US(t)/2)^2}$, and the self-consistency condition on the N\\'eel order parameter $S(t)$, so the gap $\\Delta(t) = US(t)$ responds to the carriers it helps create. The analytic engine is a generalized resonance condition that matches the poles of a two-frequency propagator $H(\\Omega,\\omega)$ at the doublon-hole energy $2E_k$ against peaks of the spectral weight of $\\partial_t\\varphi_k$; this yields the multiphoton power law $(eE_0a/\\hbar\\omega_d)^{2\\Delta/\\omega_d}$, the tunneling exponential $\\exp(-\\gamma\\Delta/eE_0a)$, and, at finite $\\Gamma$, the linear-absorption rate proportional to $\\Gamma$.","core_discovery":"On its own terms, the paper's central discovery is that a subgap-driven antiferromagnetic Mott insulator has not two but four photoexcitation regimes. Besides the standard multiphoton and Landau-Zener tunneling mechanisms separated by the Keldysh crossover, the authors find a broad cooperative region in which the carrier density follows the exponential tunneling dependence on field, $n \\propto E^2 \\exp(-\\gamma\\Delta/eEa)$, while simultaneously displaying sharp multiphoton thresholds whenever the gap is an integer multiple of the pump frequency, $\\Delta/\\omega_d \\in \\mathbb{Z}$; below a multiphoton-onset line, an incoherent linear-absorption pathway set by the dephasing rate $\\Gamma$ dominates. Because the time-dependent variational treatment computes the N\\'eel order parameter, and therefore the Mott gap $\\Delta(t) = US(t)$, self-consistently during the pulse, gap suppression feeds back into the generation rate: on resonance the pump is driven out of resonance and carrier density saturates in real time, while below resonance transient micromotion can suppress the gap enough to bring the pump into resonance and effectively lower the multiphoton threshold. The momentum distribution of doublon-hole pairs (the charge excitations left when hopping transfers an electron across the Mott gap) immediately after the pump is proposed as a quantitative diagnostic of the active mechanism.","pith_inferences":[],"forward_implications":["A single set of evolution equations reproduces both the rigid-band analytic regimes and the strong-drive phenomenology, so pump shape and pump duration become control parameters that interpolate between rate-like generation, saturation, and quasi-steady states.","Experiments inferring the excitation mechanism from field dependence alone will misread the cooperative region: carrier density scales as tunneling in the field while frequency thresholds at $\\Delta/\\omega_d \\in \\mathbb{Z}$ persist, meaning multiphoton and tunneling signatures coexist above the Keldysh line.","Near a multiphoton resonance at strong drive, pump-probe measurements should show carrier density saturating before the pulse peak, because the renormalized gap has already pushed the system off resonance.","Pumping just below $\\Delta_0/n$ should show a drive-strength-dependent redshift of the absorption edge, with efficient excitation switching on once micromotion-driven gap suppression closes the detuning.","The momentum-space spread of photocarriers, and the amplitude of coherent acoustic phonons emitted at wavevector $\\sim \\lambda_{\\rm pump}^{-1}$, both track the total carrier density and can serve as regime diagnostics in momentum-resolved and transient-reflectivity experiments.","Inference beyond the paper: the threshold-lowering mechanism implies that pulse shape is an active control knob, so a chirped pulse that walks the instantaneous frequency across the dropping gap could sustain resonance longer and raise the saturated carrier density above what a transform-limited pulse produces.","Inference beyond the paper: the single-Markovian-rate ansatz makes the saturation effect a prediction to be stress-tested by numerically exact methods that include doublon-doublon interactions and non-Markovian relaxation, which could sharpen or wash out the saturation plateau and shift the boundary of the incoherent regime.","Inference beyond the paper: the momentum-distribution diagnostic suggests a direct cold-atom test, in which quantum gas microscopy maps the distribution of doublon-hole pairs after a lattice-shaking pulse, exposing the predicted crossover from flat tunneling profiles to contour-localized multiphoton profiles to striped incoherent profiles without phononic or disorder complications."],"supporting_citations":[{"why":"Supplies the analytic multiphoton and tunneling production rates for the 1D Hubbard model that the paper reproduces in the rigid-band limit and extends to 2D.","marker":"[23]"},{"why":"Provides the spin-density-wave variational treatment of the Hubbard model on which the time-dependent Gaussian ansatz is built.","marker":"[47]"},{"why":"Defines the tunneling-versus-multiphoton ionization crossover whose semiconductor picture the paper generalizes to four regimes.","marker":"[21]"},{"why":"Reports the pump-probe experiment on a strongly driven Mott insulator that observed the Keldysh crossover, giving the regime map its main experimental benchmark.","marker":"[34]"},{"why":"Computes nonequilibrium steady states of ac-driven Mott insulators, the comparison point the paper contrasts with its real-time treatment of frequency-resonant features.","marker":"[33]"},{"why":"Describes screening of the interaction in photodoped Mott insulators, the prior gap-suppression theory the paper goes beyond by feeding the gap back into the generation rate.","marker":"[35]"},{"why":"Gives the phenomenological deformation-potential mechanism for coherent acoustic phonon generation that the paper derives microscopically from the SSH coupling.","marker":"[50]"},{"why":"Reports the original observation of surface generation and detection of phonons by picosecond light pulses, anchoring the phonon-probe channel in experiment.","marker":"[48]"},{"why":"Observes long-lived coherent acoustic waves, supporting the proposal that phonon amplitude tracks photocarrier density.","marker":"[49]"}],"fun_headline_variants":["Four photoexcitation regimes revealed in subgap-driven Mott insulators","Cooperative multiphoton-tunneling regime in subgap-pumped Mott","Gap feedback drives four regimes in subgap photoexcitation","Subgap pumping: four regimes, gap feedback, diagnostic signatures"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The entire picture rests on the variational ansatz: the driven state is assumed to be a product of independent momentum sectors containing only zero-total-momentum doublon-hole pairs, with a single Markovian dephasing rate and no carrier redistribution among momenta during the pump, so that all correlations beyond this simple structure are neglected on the pump timescale.","fun_headline_variants_meta":{"raw":{"variants":["Four photoexcitation regimes revealed in subgap-driven Mott insulators","Cooperative multiphoton-tunneling regime in subgap-pumped Mott","Gap feedback drives four regimes in subgap photoexcitation","Subgap pumping: four regimes, gap feedback, diagnostic signatures"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000694,"raw_usage":{"total_tokens":3146,"prompt_tokens":955,"completion_tokens":2191,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":571,"completion_tokens_details":{"reasoning_tokens":2116}},"tokens_in":571,"tokens_out":2191,"duration_ms":14761,"temperature":1.0,"reasoning_tokens":2116,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T15:19:41.832919+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Benchmark the same driven half-filled 2D Hubbard model ($U=1.93$ eV, $\\tau=189$ meV, 100 fs Gaussian pump) against a method that does not assume the ansatz, such as dynamical mean-field theory, tensor networks, or time-dependent exact diagonalization on small clusters, and compare total photocarrier density, momentum distributions, and real-time saturation curves. If the exact calculation shows no persistent multiphoton thresholds above the Keldysh line, no saturation plateau while the pump is still on, or no sharp onset of absorption when the transiently renormalized gap crosses $n\\omega_d$, the central claim fails.","supporting_citations":[{"cited_title":"Oka, Nonlinear doublon production in a mott insu- lator: Landau-dykhne method applied to an integrable model, Phys","cited_arxiv_id":null,"evidence_quote":"Supplies the analytic multiphoton and tunneling production rates for the 1D Hubbard model that the paper reproduces in the rigid-band limit and extends to 2D."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Defines the tunneling-versus-multiphoton ionization crossover whose semiconductor picture the paper generalizes to four regimes."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Reports the pump-probe experiment on a strongly driven Mott insulator that observed the Keldysh crossover, giving the regime map its main experimental benchmark."},{"cited_title":"Murakami and P","cited_arxiv_id":null,"evidence_quote":"Computes nonequilibrium steady states of ac-driven Mott insulators, the comparison point the paper contrasts with its real-time treatment of frequency-resonant features."},{"cited_title":"Goleˇ z, M","cited_arxiv_id":null,"evidence_quote":"Describes screening of the interaction in photodoped Mott insulators, the prior gap-suppression theory the paper goes beyond by feeding the gap back into the generation rate."},{"cited_title":"Bozovic, M","cited_arxiv_id":null,"evidence_quote":"Observes long-lived coherent acoustic waves, supporting the proposal that phonon amplitude tracks photocarrier density."}],"review_version":1}