{"id":"8f150784-348b-4dca-b92a-e4fdb132f6f3","arxiv_id":"2411.19412","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For stochastic AC fields, the quantum Fisher information for estimating the frequency separation of two close signals is approximately 2/ωr², and the bound is approachable with Dicke-state superpositions.","lead":"Quantum sensors usually struggle to tell apart two close AC frequencies because of measurement bandwidth limits. This paper shows that for noisy, fluctuating AC fields, entangled states of many qubits can beat that limit, and derives the ultimate precision bound, which improves as the inverse square of the frequency separation.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The claim that 2/omega_r^2 is the ultimate separation bound depends on an unproven optimality assumption for the pi-pulse control; the paper proves a conditional bound for that control, not a global quantum-optimal limit.","rationale":"The paper's central derivation is internally sound as a conditional bound: the Gaussian phase model and the environment-state data-processing inequality lead to Eq. (25), and the single-qubit and GHZ limits reproduce the expected results from [12]. The weak point is the interpretive step from 'a bound for the pi-pulse-controlled channel' to 'the ultimate quantum limit'. For the central claim to hold, one would need to know that the pi-pulse sequence is the optimal control for the effective phase distribution, including for N-qubit probes; the manuscript does not supply that proof. The reader's weakest assumption identifies exactly this point, so I agree with the CONDITIONAL verdict. The numerical optimal-control check proposed above would directly test whether the assumed control is actually optimal; in the absence of such a test, the conditional verdict should stand unchanged.","tokens_in":25404,"tokens_out":28194,"duration_ms":266792,"concrete_test":"For the stochastic two-frequency Hamiltonian (3) with a control field and omega_r t = 0.1, numerically optimize a piecewise-constant pulse sequence (e.g., 100 time steps) for a single-qubit probe, maximizing the QFI of the environment state (6); compare with t^2/(2 tan^2(omega_r t/2)) ~ 2/omega_r^2. Repeat the search for N=2,3 collective probes by direct fidelity-based QFI evaluation. If any optimized sequence exceeds Eq. (25), the pi-pulse protocol is not optimal and the 'ultimate' bound fails; if none does in a sufficiently broad search, the conditional interpretation is supported.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Eq. (25)'s 2/omega_r^2 is presented as the ultimate QFI bound, but it is computed from the effective Hamiltonian (21) produced by the pi-pulse sequence. The environment-state argument (Eqs. 5-7) upper-bounds any probe state only for a fixed channel of the form integral dphi q(phi) U_phi^{otimes N}(.)U_phi^{dagger otimes N} dphi. Since the control determines q(phi), a different control can produce a different environment state with a different QFI. The paper takes the pi-pulse control from the single-qubit literature [12] and does not prove that it optimizes J(sigma_omega) over all controls, nor that it remains optimal for N-qubit collective probes. Therefore 'quantum-optimal' and 'ultimate' in the abstract are not established by the paper's own argument; the proven statement is conditional on this control choice. The accompanying achievability claim is also not demonstrated for the separation result: Fig. 3 shows QFI values of order 10^-3 against 2/omega_r^2 ~ 4.08 for N up to 60, with no data or proof in the asymptotic regime N >> 2pi/sigma_phi where saturation would be expected.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper treats frequency estimation of stochastic AC fields as estimation of a parameter in a collective dephasing channel. The central technical step is an environment-state representation of the channel, combined with the data-processing inequality, to upper-bound the quantum Fisher information (QFI) achievable by any probe state and adaptive protocol. The authors derive QFI bounds for a single frequency, for the centroid of a bi-frequency signal, and for the frequency separation, with the headline result that the separation QFI is approximately 2/ωr², i.e. inversely proportional to the separation. They also analyze GHZ states, which give Heisenberg scaling in the low-bandwidth limit, and superpositions of Dicke states, which are claimed to approach the bounds in certain regimes. The supplementary material contains derivations of the effective Hamiltonian under π-pulse control, the environment-state QFIs, the GHZ-state calculations, and a noise-robustness analysis.","tokens_in":25627,"tokens_out":7828,"duration_ms":69608,"significance":"If the central claims hold, the paper provides a useful framework: it reduces a non-unitary AC sensing problem to a channel-estimation problem, gives parameter-free QFI bounds in which the amplitude variance σ cancels, and reproduces known single-qubit and coherent-signal limits (Refs. [12,22]) in the appropriate limits. The data-processing argument is conceptually clean and the extension from single-qubit to collective probes is potentially important for quantum-enhanced frequency resolution. However, the main claim of a global 'quantum-optimal' bound of 2/ωr² is conditional on a specific control choice whose optimality is not proven, and the claimed saturation by Dicke-superposition states is not demonstrated in the plotted regime. These issues are load-bearing for the paper's advertised conclusions.","major_comments":[{"comment":"The central claim that the QFI for frequency separation is ultimately bounded by approximately 2/ωr² is conditional, not unconditional. The environment-state argument of Eqs. (5)–(7) bounds the QFI only for the fixed channel whose phase distribution q(ϕ) is generated by the chosen control. Here q(ϕ) is generated by the π-pulse sequence with δs t = 2π taken from Ref. [12]; different controls produce different q(ϕ) and hence different environment-state QFIs. Since the paper does not optimize J(σω) over controls, nor prove that this sequence is optimal for N-qubit collective probes, the words 'quantum-optimal' and 'ultimate' in the abstract and around Eq. (25) overstate what is proven. The proven statement is a bound for this control family; a control-optimality analysis, or a carefully qualified claim, is needed.","section":"Estimating the frequency separation (Eqs. 21–25)"},{"comment":"The claim that the 2/ωr² bound is achievable by superpositions of Dicke states is not demonstrated for the separation problem. For the plotted parameters (t = 0.7, ωr = 0.7), the numerically computed QFI of |˜+_N> is several orders of magnitude below 2/ωr² ≈ 4.08 for all N ≤ 60, and the paper provides no asymptotic evaluation in the regime N ≫ 2π/σ_φ or ωr t ≪ 1 where saturation is claimed. The bosonic saturation result of Ref. [39] is invoked by analogy but not transferred with a proof. Please add either an explicit N→∞ analysis or numerical data in the asymptotic regime.","section":"Fig. 3 and SM 'Spin number-position states'"},{"comment":"The label 'exact' is not applied consistently. The single-frequency environment-state QFI in Eq. (11) is exact for the Gaussian model, but Eq. (25) and the simplified bound 2/ωr² are derived under the approximations δs t = 2π, δr ≪ δs, and ωr t ≪ 1 (see the SM 'Frequency separation' section). The abstract's 'exact quantum Fisher information bounds' should be qualified so that readers do not take the small-separation asymptotic expression as a global exact result.","section":"Abstract and Eqs. (22)–(25)"}],"minor_comments":[{"comment":"The displayed relation 'ω2 = ωs − ωs' contains a typo and should read 'ω2 = ωs − ωr'.","section":"SM 'Two frequencies' after Eq. (54)"},{"comment":"The phrase 'Assuming δs ≪ 1' mixes a frequency with a dimensionless number; the needed small parameter is ωr t ≪ 1, equivalently δr ≪ δs.","section":"Main text near Eq. (22)"},{"comment":"The quantity on the vertical axis is not defined; if the plotted quantity is J_ω σ_ω or J_ωr σ_ω, this should be stated explicitly in the caption.","section":"Fig. 2 and Fig. 3 captions"},{"comment":"The expression 'Jωs = N²σ²t⁶ωs²/(18σ²)' has a spurious σ² in the denominator; it should read 'Jωs = N²σ²t⁶ωs²/18'.","section":"SM after Eq. (166)"},{"comment":"The state |˜+_N> is called both a 'number-superposition state' and a 'number-position state'; please use one term consistently.","section":"Main text near Eq. (12) and Fig. 2"}],"recommendation":"major_revision","confidential_remarks":"The paper's advertised novelty is stronger than what the current proofs establish. The environment-state data-processing bound is sound, but the 'ultimate' separation bound depends on an unverified optimality assumption about the π-pulse control, and the Dicke-state achievability claim lacks asymptotic evidence. These are fixable with additional analysis or a reframed claim, so I do not recommend rejection; however, the manuscript should not be accepted in its present form."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Zixin's group has a real result here. The mapping of stochastic AC frequency estimation to a global dephasing channel, followed by the environment-state data-processing inequality, is clean and does genuine work: it reproduces the known single-qubit results [12] and coherent limits [22] in the right limits, and it yields new bounds for the frequency centroid and, strikingly, an inverse-square dependence on the frequency separation omega_r. Table I is useful. The Dicke-state construction is an interesting addition, and the numerics suggest saturation in some regimes. This is not a fake paper.\n\nThe soft spots are real but localized. The biggest one is the word \"quantum-optimal\" in the title. The bound in Eq. (25) is derived for the effective Hamiltonian produced by the pi-pulse sequence with delta_s t = 2pi. That control is taken from the single-qubit literature. The environment-state argument bounds any probe state for that fixed channel, but a different control produces a different q(phi) and therefore a different environment state, so nothing here rules out a better control. The 2/omega_r^2 is a conditional bound, not the ultimate quantum limit. The paper should say that explicitly.\n\nSecond, \"exact\" is doing too much work. The separation variance in Eq. (23) is derived under delta_r << delta_s and then expanded, and Eq. (25) itself is the small-omega_r t limit. The abstract says \"approximately\", but the body calls these exact bounds. Third, there is a concrete internal inconsistency: main-text Eq. (20) gives J_omega_s ~ (1/18) sigma^2 t^6 omega_s^2 while SM Eq. (132) gives (1/72) omega_s^2 t^4. That needs fixing. Finally, the achievability claim for the separation bound is not demonstrated in the asymptotic regime: Fig. 3 shows QFI values around 10^-3 for N up to 60, versus 2/omega_r^2 ~ 4.08, so saturation is a hope, not a shown fact.\n\nNone of this kills the paper. The core environment-state technique is sound, the new bounds are worth having, and the single-qubit and coherent checks give confidence. With a revised title, a clear statement of what is and is not optimized, and a corrected SM, this is a solid contribution. I'd send it to a serious referee.","headline":"A genuinely new environment-state bound for stochastic AC frequency estimation, but the headline 2/omega_r^2 separation limit is proven only for a fixed pi-pulse control, not as a global quantum-optimal bound.","tokens_in":26175,"tokens_out":2760,"would_cite":true,"duration_ms":24398,"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":"Estimating the frequency separation of two stochastic AC fields is ultimately bounded by a quantum Fisher information of about $2/\\omega_r^2$, so closer signals become easier, not harder, to resolve.","keywords":["quantum Fisher information","frequency estimation","stochastic AC fields","dephasing channel","Dicke states","GHZ states","Heisenberg scaling","quantum control"],"falsifier":"On a bi-frequency stochastic field with known $\\omega_r$, run the prescribed $\\pi$-pulse sequence with $\\delta_s t = 2\\pi$ on a GHZ or spin number-superposition probe, measure in the Fourier basis, and compare the estimator variance to the predicted Cramér–Rao bound with $J_{\\omega_r} = 2/\\omega_r^2$; if any control or state yields precision beyond that bound, or if this measurement falls short for reasons not attributable to technical noise, the claimed ultimate limit is settled.","tokens_in":25162,"feed_emoji":"⚛️","tokens_out":7607,"duration_ms":64295,"temperature":0.7,"pith_summary":"This paper derives the ultimate quantum limits for estimating the frequency of a stochastic AC field and for separating two close frequencies in such a field. Its central new result is that the quantum Fisher information for the frequency separation $\\omega_r$ is approximately $2/\\omega_r^2$, so the smaller the separation, the more precisely it can in principle be estimated—the opposite of the classical bandwidth intuition that resolving close frequencies demands long observation times. The argument works by mapping frequency estimation onto estimation of a global dephasing channel, then bounding any probe's precision by the quantum Fisher information of an auxiliary environment state. Superpositions of Dicke states approach the bound in certain regimes, while GHZ states give an $N^2$ Heisenberg scaling over unentangled states in the low-bandwidth limit.","feed_headline":"Classical 1/t limit beaten: AC frequency split bound is 2/ωr²","feed_subtitle":"When two stochastic signals nearly coincide, quantum sensing makes the split easier to measure, not harder—reversing classical spectroscopy.","key_machinery":"The load-bearing object is the environment-state decomposition of a collective dephasing channel: the probe sees a random phase $\\phi$ drawn from $q(\\phi)$, and one defines $\\sigma_\\omega = \\int d\\phi\\, q(\\phi)|\\phi\\rangle\\langle\\phi|$ on an auxiliary orthogonal basis. By the data-processing inequality, no adaptive protocol using any probe state can have quantum Fisher information larger than that of $\\sigma_\\omega$, turning the search over states into a single calculation of the quantum Fisher information of a Gaussian phase distribution. The second ingredient is the $\\pi$-pulse control sequence with detuning $\\delta_s$ chosen so $\\delta_s t = 2\\pi$, which produces the effective Hamiltonian $H_{\\rm eff}(t) = (2/\\pi)\\sum_i [A_i\\cos((\\delta_s\\pm\\delta_r)t)+B_i\\sin((\\delta_s\\pm\\delta_r)t)]\\sigma_z$ and makes the separation phase $\\phi_{\\omega_r}\\approx (A_2-A_1)t\\sin(\\omega_r t)/\\pi^2$; that phase's Gaussian variance gives the separation bound. Spin number-superposition states $\\frac{1}{\\sqrt{N+1}}\\sum_{n=0}^N |D_N^n\\rangle$, where $|D_N^n\\rangle$ is a Dicke state with $n$ excitations, are the states shown to approach the bound.","core_discovery":"The discovery is a set of exact quantum Fisher information formulas for stochastic AC frequency sensing, capped by the separation bound $J_{\\omega_r} = t^2/(2\\tan^2(\\omega_r t/2)) \\approx 2/\\omega_r^2$ for two close signals. The paper shows that for stochastic amplitudes, the accumulated phase is Gaussian, so the whole channel is a collective dephasing channel with an environment state $\\sigma_\\omega$ that carries all estimable information. The quantum Fisher information of that environment state is an upper bound for any probe state, and the paper evaluates it exactly for single-frequency, centroid, and separation estimation. It then shows that spin number-superposition states $|\\tilde{+}_N\\rangle$, the spin analogue of a bosonic number-superposition state, approach the bound at large $N$ in some regimes, while GHZ states fall short of the optimum but still deliver $N^2$ Heisenberg scaling in the low-bandwidth limit.","pith_inferences":["A direct extension the authors do not develop: because the accumulated phase remains Gaussian for any number of frequency components, the same environment-state method should yield exact quantum Fisher information bounds for multi-frequency stochastic fields beyond two tones.","The inverse-square scaling in $\\omega_r$ is reminiscent of superresolution imaging of two incoherent point sources, where the quantum Fisher information for separation also improves as sources approach; we read both as manifestations of measurement noise vanishing near an eigenstate, and expect a unified treatment to be possible.","Testable design consequence: the optimal qubit number $N$ for a GHZ probe depends on the noise strength $\\sigma$, since larger noise favors smaller $N$, so a practical sensor would tune $N$ against the noise level rather than always maximizing it."],"forward_implications":["For two close stochastic AC signals, the achievable variance in estimating $\\omega_r$ scales as $\\omega_r^2/2$, so the resolution limit improves as the frequencies move closer together, within the regime where the approximations hold.","GHZ states provide an $N^2$ enhancement over single-qubit probes for single-frequency, centroid, and separation estimation in the low-bandwidth limit, despite the channel being non-unitary.","The spin number-superposition state saturates the environment-state bound for certain noise strengths and large qubit number, identifying a concrete state family and Fourier-basis measurement that achieves the limit.","To leading order the optimal separation bound $2/\\omega_r^2$ is independent of the noise amplitude $\\sigma$ and the interrogation time $t$, so the limiting factor is the separation itself."],"supporting_citations":[{"why":"It supplies the single-qubit $\\pi$-pulse control and effective Hamiltonian used for separation estimation, and it gives the single-qubit quantum Fisher information baseline that the multi-qubit results extend.","marker":"[12]"},{"why":"It provides the optimal adaptive-control method for coherent time-dependent Hamiltonians that the paper extends to the coherent rows of its comparison table.","marker":"[22]"},{"why":"It reports the experimental demonstration of single-qubit quantum-enhanced frequency resolution, supporting the claim that projection-noise suppression underlies the advantage.","marker":"[27]"},{"why":"It gives the channel-estimation distinguishability setting in which quantum Fisher information is bounded via the environment state and data processing.","marker":"[38]"},{"why":"It establishes that number-superposition states saturate exact bounds for bosonic dephasing channels, the technique adapted here to spin Dicke superpositions.","marker":"[39]"}],"fun_headline_variants":["Quantum sensing flips the rule: close AC signals easier to split","Exact quantum bound: frequency split precision ~ 2/ωr²","Stochastic AC sensing: Dicke states reach quantum Fisher limit","GHZ states still give Heisenberg scaling for AC frequency","Quantum metrology breaks 1/t barrier for stochastic AC fields"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the $\\pi$-pulse control sequence, with detuning chosen so $\\delta_s t = 2\\pi$, realizes the effective Hamiltonian used in the derivation and is the optimal control for $N$-qubit probes; the paper takes this control from the single-qubit case and does not prove its optimality here.","fun_headline_variants_meta":{"raw":{"variants":["Quantum sensing flips the rule: close AC signals easier to split","Exact quantum bound: frequency split precision ~ 2/ωr²","Stochastic AC sensing: Dicke states reach quantum Fisher limit","GHZ states still give Heisenberg scaling for AC frequency","Quantum metrology breaks 1/t barrier for stochastic AC fields"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000242,"raw_usage":{"total_tokens":1527,"prompt_tokens":948,"completion_tokens":579,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":564,"completion_tokens_details":{"reasoning_tokens":490}},"tokens_in":564,"tokens_out":579,"duration_ms":5056,"temperature":1.0,"reasoning_tokens":490,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T10:14:08.428608+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"On a bi-frequency stochastic field with known $\\omega_r$, run the prescribed $\\pi$-pulse sequence with $\\delta_s t = 2\\pi$ on a GHZ or spin number-superposition probe, measure in the Fourier basis, and compare the estimator variance to the predicted Cramér–Rao bound with $J_{\\omega_r} = 2/\\omega_r^2$; if any control or state yields precision beyond that bound, or if this measurement falls short for reasons not attributable to technical noise, the claimed ultimate limit is settled.","supporting_citations":[{"cited_title":"Gefen, A","cited_arxiv_id":null,"evidence_quote":"It supplies the single-qubit $\\pi$-pulse control and effective Hamiltonian used for separation estimation, and it gives the single-qubit quantum Fisher information baseline that the multi-qubit results extend."},{"cited_title":"Robust asymptotic entanglement under mul- tipartite collective dephasing","cited_arxiv_id":null,"evidence_quote":"It provides the optimal adaptive-control method for coherent time-dependent Hamiltonians that the paper extends to the coherent rows of its comparison table."},{"cited_title":"Optical magne- tometry","cited_arxiv_id":null,"evidence_quote":"It reports the experimental demonstration of single-qubit quantum-enhanced frequency resolution, supporting the claim that projection-noise suppression underlies the advantage."},{"cited_title":"Semiconductor spin qubits","cited_arxiv_id":null,"evidence_quote":"It gives the channel-estimation distinguishability setting in which quantum Fisher information is bounded via the environment state and data processing."},{"cited_title":"Braunstein and Carlton M","cited_arxiv_id":null,"evidence_quote":"It establishes that number-superposition states saturate exact bounds for bosonic dephasing channels, the technique adapted here to spin Dicke superpositions."}],"review_version":1}