{"id":"994b7a90-85bb-46fc-82f6-d3489365f4ad","arxiv_id":"2607.14239","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"A new amplitude-amplification technique gives O(n^{1/4}) gate count for Fock-state preparation and 1/√p scaling for single-photon generation.","lead":"This paper proposes new quantum control protocols for generating single photons and Fock states using amplitude amplification. It claims a quadratic improvement in gate count for Fock-state preparation and a reduction in cavity loss for single-photon generation.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Very oblivious AA requires [Ū,N]=0, but N in Eq. (25) is not conserved by H in Eq. (20); single-photon scaling is approximate, though the Fock-state O(n^{1/4}) claim is unaffected.","rationale":"The paper presents two central claims: improved single-photon generation via very oblivious amplitude amplification, and Fock-state preparation via fixed-point amplitude amplification with O(n^{1/4}) SNAP gates. The Fock-state protocol is mathematically sound: the sequence in Eq. (41) implements the standard fixed-point reflections exactly in the ideal model, and the gate-count scaling follows from Eq. (10) and the coherent-state overlap. The single-photon protocol, however, depends on a simplification that is only valid under an operator commutator assumption. The N defined in Eq. (25) is not a conserved charge of the Hamiltonian in Eq. (20): the laser and cavity couplings both change N when |e,0⟩ is populated. Thus [Ū,N]=0 cannot hold exactly for any finite-duration pulse that achieves the desired transfer. This means the very oblivious AA identity is approximate, with corrections that vanish only in the adiabatic limit. The paper's error analysis in Appendix B and numerical simulations show the approximation is well-controlled for the studied parameters, so the central efficiency improvement likely survives. But the analytic claim of 1/√p scaling is not a rigorous theorem; it is an asymptotic result in the double limit of weak pulse and adiabaticity. The reader's CONDITIONAL verdict appropriately captures this: the protocol is plausible and supported by numerics, but the exactness of the simplification is an assumption that needs further verification or explicit error accounting. We therefore agree with the reader's weakest assumption and recommend no change to the verdict.","tokens_in":23403,"tokens_out":22831,"duration_ms":216617,"concrete_test":"Simulate the exact single-photon protocol — the full sequence S_{2k}...S_0 from Eq. (29) with T operations included, using the control pulses from Appendix A — and compare its output mode fidelity to the simplified protocol that omits T operations (as in Fig. 4). For the parameters of Fig. 4 (κ_exT=100, g=5κ_ex, κ_i=0.25κ_ex, γ=0.003κ_ex), if the difference in 1−η_2 exceeds the bound in Eq. (B42) or scales worse with k, the very oblivious AA simplification is the limiting factor for the claimed 1/√p improvement.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The most load-bearing concern is the validity of the very oblivious amplitude amplification identity used for the single-photon protocol. In Sec. II A, G3 = Ū R̄1 Ū† R̄1 (Eq. 19) requires [Ū,N]=0, with N defined in Eq. (16) and physically in Eq. (25). However, the Hamiltonian in Eq. (20) does not conserve this N: the laser term Ω(t)|e⟩⟨s| couples |s,0_p⟩ (N=1) to |e,0_p⟩ (N=0), and the cavity term g a|e⟩⟨g| couples |g,1_p⟩ (N=1) to |e,0_p⟩ (N=0). Thus [H,N]≠0, and a finite-duration pulse implementing Ū will generically have [Ū,N]≠0. The paper's assertion in Sec. III A that Ū† 'conserves the number of excitations N' is therefore only an approximation that holds in the adiabatic limit κ_ex/(g^2T)→0, where |e⟩ is never populated. Consequently, the simplification G3 is not exact; the analytic N~1/√p scaling derived from ideal Grover amplification is approximate. The non-adiabatic error bound in Eq. (B42) grows as k κ_ex/(g^2T), meaning that for the large k needed when p is small, the approximation error can become the limiting factor unless κ_ex/(g^2T) is made correspondingly small. This is a genuine limitation, but it is acknowledged and numerically analyzed; it does not invalidate the Fock-state protocol of Sec. III B, which uses standard fixed-point AA and is exact under ideal SNAP/displacement gates.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper introduces a modification of oblivious amplitude amplification—called very oblivious amplitude amplification—in which the reflection about the initial state is replaced by ŪR̄1Ū† under a conservation condition [Ū,N]=0, and applies it to two cavity-QED settings. In the first setting, a Λ-type emitter coupled to a cavity is used to generate traveling single photons with weak control pulses; the ideal protocol length is claimed to scale as N∼1/√p rather than N∼1/p. Loss formulas and numerical simulations are given, including a comparison with a single-strong-pulse protocol. In the second setting, fixed-point amplitude amplification is implemented using SNAP gates and displacements in dispersive circuit QED, yielding an explicit Fock-state preparation sequence with gate count 2l=(2π)^{1/4}log(2/δ)n^{1/4}−1 (Eq. 42), i.e., O(n^{1/4}) scaling, and an extension to NOON states is described.","tokens_in":23838,"tokens_out":11032,"duration_ms":114659,"significance":"The Fock-state result is the strongest part of the paper: it is an explicit, analytically derived gate sequence with no fitted parameters, based on published fixed-point amplitude amplification, and it gives a clear asymptotic improvement in SNAP-gate count. If correct, this is a practically useful contribution for circuit-QED state preparation and is likely to be of interest to the quantum-optics and quantum-information communities. The proposed very oblivious amplitude amplification is also a conceptually clean tool that may find applications beyond the two protocols studied here. The single-photon protocol, if it can be made rigorous in a finite-parameter regime, would be a valuable alternative to strong-pulse generation. The numerical work in the paper supports the main practical claims, but the formal part has a gap, described below, that should be fixed before the single-photon claims are accepted at face value.","major_comments":[{"comment":"The very-oblivious identity G3 = Ū R̄1 Ū† R̄1 relies on [Ū,N]=0. The physical Hamiltonian in Eq. (20) does not conserve the N defined in Eq. (25): the term Ω(t)|e⟩⟨s| maps |s,0_p⟩ (N=1) to |e,0_p⟩ (N=0), and the term g a|e⟩⟨g| maps |g,1_p⟩ (N=1) to |e,0_p⟩ (N=0). Thus [H,N]≠0 and, for a finite-duration pulse, [Ū,N]=0 is not an exact identity; it is an adiabatic-limit statement that holds only as κ_ex/(g^2T)→0. Consequently, the simplification G3 and the derived N∼1/√p scaling are approximate, not exact. The loose non-adiabatic bound in Eq. (B42) does not close this gap because it bounds the final output error rather than the error in the very-oblivious reflection step itself, and it grows linearly with k. The authors should state this approximation explicitly in the main text and either provide a direct bound on ∥[Ū,N]∥ or a quantitative numerical convergence study of the Grover-step err","section":"Sec. III A, Eqs. (20), (25), (26)"},{"comment":"The claim of a quadratic improvement over Ref. [35] in SNAP-gate count is supported in Fig. 6 by a numerical fit to the gate count of Ref. [35], not by an exact gate count or a published scaling bound. As written, the figure convincingly demonstrates the O(n^{1/4}) scaling of the new protocol, but it does not quantitatively establish the claimed improvement over Ref. [35] unless the fit is known to reproduce the exact gate count of that protocol. The authors should provide the exact gate-count expression for Ref. [35] or cite a rigorous derivation of its O(n^{1/2}) scaling.","section":"Sec. III B, Fig. 6"},{"comment":"The non-adiabatic error bound is acknowledged in the text to be 'very loose' and contains the unspecified constant c_h through ∥ḣ∥2 = c_h/T. The bound grows as (10.2 + 5.1k) κ_ex/(g^2T). For the large k values relevant when the single-photon success probability p is small, this bound is not useful unless κ_ex/(g^2T) is made extremely small, which is in tension with the finite-loss regime where the protocol is claimed to be advantageous. A tighter, interpretable bound—or a clear statement of the achievable (k, κ_ex/(g^2T)) trade-off—is needed to support the analytic claims for the single-photon protocol.","section":"Appendix B, Eq. (B42)"}],"minor_comments":[{"comment":"The expression 2l = (2π)^{1/4}log(2/δ)n^{1/4} − 1 contains a −1 that makes 2l non-integer for generic δ. Please state that the physical gate count is ⌈2l⌉ and check the offset convention against Eq. (10), which already contains a −1/2.","section":"Eq. (42)"},{"comment":"The phrase 'as [κ_ex/g, 1/(gT)] decrease' is ambiguous. Please specify that both dimensionless ratios are decreased, and in what order or jointly.","section":"Fig. 2 caption"},{"comment":"The statement that the single-photon output is 'invariant under T to high accuracy' should explicitly identify the order of the approximation as the same non-adiabatic order as Eq. (B42), so that the reader does not mistake it for an exact symmetry.","section":"Sec. III A, after Eq. (30)"},{"comment":"The time-reversal operation T is defined for real mode functions f(t)→f(T−t). For complex mode functions, a complex conjugation is generally required in the definition of T; please clarify the intended generalization.","section":"Appendix A"},{"comment":"The two-mode protocol assumes that the SNAP/displacement operations on mode i can be performed without cross-talk with mode j. This is stated implicitly via U_n^{(i)} but would benefit from an explicit sentence about the required addressing or frequency separation between the two cavities.","section":"NOON-state protocol"}],"recommendation":"major_revision","confidential_remarks":"The Fock-state protocol appears sound and is the most publishable contribution; the single-photon section needs a careful treatment of the [Ū,N]=0 approximation before the claims about 1/√p scaling and error reduction can be accepted. The comparison to Ref. [35] via a numerical fit should also be strengthened. I see no concern about novelty or scope."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know before you open it. The Fock-state result is the real one: an explicit unitary sequence with 2l = (2π)^{1/4} log(2/δ) n^{1/4} − 1 SNAP gates, a clean quadratic improvement over the O(n^{1/2}) of Krastanov et al., and it follows directly from Yoder–Low–Chuang fixed-point amplification once you notice the coherent-state overlap scales as (2πn)^{-1/4}. The single-photon part is genuinely clever but more approximate than the text sometimes implies.\n\nWhat the paper does well. The 'very oblivious' amplitude amplification is a new simplification — it replaces the reflection about the full initial state, which would be hard to implement, with Ū R̄1 Ū†, needing only signal-qubit reflections and Ū/Ū†. That is a real structural improvement for scattering-type settings. The loss-reduction effect is also convincing: the intrinsic-cavity-loss term accumulates as κ_i sin²θ Σ_j cos²(jθ) ≈ κ_i/k, so total loss has a minimum at k ∝ √(κ_i C_ex/κ_ex), and the numerics in Fig. 4 show a roughly 10× better inefficiency than the single-strong-pulse protocol in a concrete parameter set. The NOON-state extension is a reasonable by-product with a derived fidelity bound.\n\nSoft spots, proportional. (1) The paper states in Sec. III A that Ū† conserves N and therefore [Ū†,N]=0, and this is the load-bearing assumption for G3. It is not exact: H in Eq. (20) takes |s,0p⟩ and |g,1p⟩, both N=1, into |e,0p⟩, which has N=0. The simplification holds only in the adiabatic limit κ_ex/(g²T)→0. To the authors' credit, Appendix B and the figures analyze the non-adiabatic errors, and the numerics simulate the physical Hamiltonian rather than the idealized identity, so the protocol stands — but the assertion should be qualified where it appears, and the N∼1/√p scaling should be read as valid up to O(k κ_ex/g²T) corrections. The same adiabatic assumption justifies omitting the experimentally costly T operations. (2) The comparison to Ref. [35] in Fig. 6 uses a numerical fit, not exact gate counts from that paper. The asymptotic scaling is clear; the crossover for practical n is less certain. (3) The error bound in Eq. (B42) is loose by the authors' own admission — minor.\n\nWho it's for: circuit-QED people preparing Fock or NOON states get the gate-count result; atom/quantum-dot cavity people get the single-photon loss analysis. It deserves a serious referee. The issues are text-level qualifications and a fairer comparison to prior work, not fatal flaws.","headline":"The Fock-state O(n^{1/4}) gate-count result is the cleanest part of this paper; the single-photon protocol is clever but rests on a conservation-law identity that is only approximate in the physical system, and the text should qualify it.","tokens_in":24261,"tokens_out":11095,"would_cite":true,"duration_ms":104764,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["81P68","81V80"],"pacs":["42.50.Pq","03.67.-a"],"model":"deepseek-v4-flash","headline":"Amplitude amplification cuts cavity Fock-state preparation from n^(1/2) to n^(1/4) gates.","keywords":["amplitude amplification","cavity QED","Fock state preparation","SNAP gates","single-photon generation","fixed-point amplitude amplification","NOON state","oblivious amplitude amplification"],"falsifier":"For the single-photon protocol, measure 1−η₃ as a function of k at fixed κ_ex, g, T, κ_i, γ and check the predicted minimum at k≈(π/8)√(κ_i C_ex/κ_ex) and the loss scaling of Eq. (31); a disagreement in how loss grows with k would falsify the adiabatic treatment. For the Fock-state protocol, implement U_n on the coherent state D(√n)|0⟩ for n=1,4,16,64 and verify that the SNAP-gate count needed for fidelity 0.99 grows as n^{1/4}, not n^{1/2}.","tokens_in":23360,"feed_emoji":"⚛️","tokens_out":7973,"duration_ms":63908,"temperature":0.7,"pith_summary":"The paper develops two cavity-QED state-preparation protocols built on amplitude amplification. For a three-level emitter coupled to a cavity, it introduces 'very oblivious amplitude amplification,' a simplification that uses only signal-qubit reflections plus U and U†, and shows that repeated weak control pulses can generate a traveling single photon with protocol length scaling as 1/√p instead of 1/p, while also reducing loss from intrinsic cavity decay in many regimes. For a superconducting qubit dispersively coupled to a microwave cavity, it applies fixed-point amplitude amplification to a sequence of SNAP gates and displacements, proving that the Fock state |n⟩ can be prepared with O(n^{1/4}) gates, a quadratic improvement over the previous O(n^{1/2}). The same techniques yield a NOON-state preparation protocol with explicit fidelity bounds.","feed_headline":"Amplitude amplification cuts Fock-state gate count to n^1/4","feed_subtitle":"New cavity-QED protocols build single photons with 1/√p overhead and NOON states without resonant couplings.","key_machinery":"The key object is the very-oblivious-amplitude-amplification unitary G3 = Ū R̄1 Ū† R̄1, in which R̄1 = I−2|0⟩⟨0| acts only on the signal qubit; it replaces the two-register reflection of standard oblivious amplitude amplification and is exact when [Ū,N]=0. For Fock states, the load-bearing sequence is the fixed-point amplification U_n = ∏_j D(λ)S_0(α_j)D(λ)†S_n(β_j), built from SNAP gates (conditional phases on Fock states) and cavity displacements, whose length is set by analytic phase schedules. The conserved-observable condition is what makes the simplification work; in the cavity implementation it corresponds to never populating the excited state at the end of a step.","core_discovery":"The central claim is that amplitude amplification, normally requiring reflections about the good state and about the initial state, can be reorganized into a 'very oblivious' form—G3 = Ū R̄1 Ū† R̄1—where R̄1 is a reflection on the signal qubit alone, valid whenever the unitary Ū conserves the observable N = |s⟩⟨s|⊗I + I⊗(1−|0p⟩⟨0p|). In the Λ-system cavity setting this makes each amplification step a sequence of weak control pulses, yielding single-photon generation whose length scales as 1/√p and whose intrinsic-cavity-loss error can be much smaller than with a single strong pulse. In circuit QED, using fixed-point amplitude amplification with SNAP gates and displacements, the number of","pith_inferences":["The n^{1/4} scaling is likely close to optimal for fixed-point protocols using only SNAP and displacement gates, because the initial coherent state's overlap with |n⟩ already falls as (2π n)^{-1/4}; any local protocol may need at least that many controlled-phase steps.","The same fixed-point construction could be adapted to prepare other target states, such as bosonic code states, by replacing the reflection about |n⟩ with a reflection about the target subspace; the cost would scale roughly with the inverse square-root of the initial-state overlap.","The single-photon loss analysis suggests an adaptive version where k is tuned on the fly to the minimum of Eq. (31); a quick calibration of κ_i/C_ex would select the optimal k per device, and repeated amplification rounds could yield multiphoton states with shaped modes.","If the time-reversal operations can be implemented without excessive overhead, the protocol becomes deterministic up to losses, since the T-steps preserve the output mode exactly; the bottleneck then shifts entirely to the adiabaticity parameter κ_ex/(g²T)."],"forward_implications":["Traveling single photons can be produced with much weaker control pulses—Rabi frequencies that fall as 1/k—while preserving a short photon duration, easing experimental constraints such as off-resonant transitions and charge noise.","The protocol's intrinsic-cavity-loss error decreases as κ_i/(κ_ex k) over k steps, so there is a regime where amplitude amplification outperforms a single strong control pulse, suggesting its use for error reduction.","Fock-state preparation with O(n^{1/4}) SNAP gates makes larger-n Fock states practical with currently available circuit-QED hardware at fixed target fidelity.","NOON states can be prepared without resonant transmon-cavity interactions or three-level transmons, simplifying the experimental setup.","The general very-oblivious construction applies to any amplitude-amplification problem whose unitary conserves the relevant excitation number, so other quantum algorithms may inherit the simplified gate structure."],"fun_headline_variants":["Amplitude amplification makes single photons in 1/√p steps","Fock states prepared in O(n^1/4) steps via amplitude amplification","Amplitude amplification yields 1/√p photon generation scaling","Amplitude amplification enables NOON state preparation"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"Everything rests on the conservation condition [Ū,N]=0, which in the physical cavity implementation holds only in the adiabatic limit κ_ex/(g²T)→0; if the excited state is populated at the end of a step, the simplified amplification step G3 is inexact and extra errors appear.","fun_headline_variants_meta":{"raw":{"variants":["Amplitude amplification makes single photons in 1/√p steps","Fock states prepared in O(n^1/4) steps via amplitude amplification","Amplitude amplification yields 1/√p photon generation scaling","Amplitude amplification enables NOON state preparation"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001071,"raw_usage":{"total_tokens":4365,"prompt_tokens":832,"completion_tokens":3533,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":576,"completion_tokens_details":{"reasoning_tokens":3458}},"tokens_in":576,"tokens_out":3533,"duration_ms":23339,"temperature":1.0,"reasoning_tokens":3458,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-02T02:41:24.656220+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"For the single-photon protocol, measure 1−η₃ as a function of k at fixed κ_ex, g, T, κ_i, γ and check the predicted minimum at k≈(π/8)√(κ_i C_ex/κ_ex) and the loss scaling of Eq. (31); a disagreement in how loss grows with k would falsify the adiabatic treatment. For the Fock-state protocol, implement U_n on the coherent state D(√n)|0⟩ for n=1,4,16,64 and verify that the SNAP-gate count needed for fidelity 0.99 grows as n^{1/4}, not n^{1/2}.","supporting_citations":[],"review_version":1}