{"id":"98e8ef59-c22e-45b0-bb9c-e4f8823e2422","arxiv_id":"2507.11395","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"Inter-well beam splitting of independent BECs in an array of double wells yields a quantum Fisher information of about M n^2 / 2, exceeding the standard quantum limit without prior spin squeezing.","lead":"This paper shows that mixing atoms between neighboring double wells before an interferometer sequence can create useful entanglement and beat the shot-noise limit, even when each well starts as an ordinary uncorrelated Bose-Einstein condensate. The effect could improve the precision of array-based atomic sensors, such as lattice gravimeters, without requiring spin-squeezed input states.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The n^2 enhancement in Eq. (34) depends critically on the fixed-number Fock input |0,n>; with Poissonian coherent-state inputs a passive beam splitter gives product coherent states and only SQL scaling. Section IIIC addresses only imbalance fluctuations, not total-number statistics.","rationale":"The reader's weakest_assumption correctly identifies the fixed-number input as load-bearing, but the reader's rationale underweights it: the three cited weaknesses (optimal-measurement proof, HOM labeling, beam-splitter implementation) are secondary to the fact that Eq. (34) is derived only for the fixed-N Fock/CSS sector. The paper's Section IIIC, which is meant to address atom-number fluctuations, only varies the distribution between the left and right sites at fixed total n per well. It does not test Glauber coherent-state inputs, where a passive beam splitter leaves product coherent states and the QFI is at most SQL. This is not an internal contradiction of the pure-state calculation, so the paper could still be correct as an idealized result; however, the physical protocol's central advantage rests on a nonclassical number state, and the manuscript does not provide evidence that such an input is achievable or that the scaling survives Poissonian number statistics. A concrete numerical check with coherent-state inputs would settle whether the claimed n^2 scaling is robust. Since the paper is already CONDITIONAL and this concern does not change the verdict category, the reader's verdict remains appropriate, but the stated rationale should emphasize the input-state dependence rather than treating it as an aside.","tokens_in":14251,"tokens_out":33505,"duration_ms":395057,"concrete_test":"Compute I_q = 4 Var(S_y) for the same protocol (e^{-iπ/2 S_x} followed by e^{-iθ J_y}) with input per well |α>_r |0>_l, mean n=|α|^2, for M=2 and M=10 and several n (e.g., n=5, 20). Use exact diagonalization for small n or Gaussian-state covariance-matrix formulas. If I_q scales as M n (SQL) instead of (1/2) M n^2, the enhancement is an artifact of the Fock assumption. Alternatively, compute the QFI for the Poissonian mixture over total number sectors with mean M n; if it still exhibits M n^2 scaling, the coherent-state concern would be falsified.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Eq. (33) fixes every double well to exactly n atoms in the right mode, and Eq. (34) is derived inside that fixed-N sector. If instead each well is loaded with a Glauber coherent state |α>_r |0>_l (mean n=|α|^2), the beam splitter e^{-iπ/2 S_x} is a passive linear transformation: coherent states remain product coherent states in the rotated modes. For such Gaussian states, the variance of the rotated generator S_y is O(n) per mode (Wick's theorem), so I_q ~ M n, the SQL, and the n^2 term in Eq. (34) disappears. Section IIIC does not cover this case: the mixture in Eq. (41) keeps the total number per well fixed at n and only broadens the left/right distribution p(m_i); it does not model Poissonian number statistics of the modes. The paper's caveat that Eq. (33) 'neglects shot-to-shot atom-number fluctuations' refers to the imbalance distribution, not to coherent-state number fluctuations. Thus the central metrological advantage is tied to the nonclassical fixed-number input; if real BECs are better described by coherent states, the proposed scheme does not beat the SQL. This is a load-bearing assumption, not a minor parameter issue, because it directly controls the headline scaling.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript analyzes a metrological protocol based on a one-dimensional array of M double wells. Each well is initially prepared with exactly n atoms in the right mode, i.e., the product Fock state ∏_i |0,n>_i (Eq. 33). A nearest-neighbor beam-splitting network e^{-iπ/2 S_x} (Eq. 25) couples adjacent wells, and the parameter θ is imprinted by the collective rotation e^{-iθ J_y}. The authors derive the quantum Fisher information for this protocol analytically (Eq. 34: I_q = M(n^2+2n)/2), which exceeds the standard quantum limit N=Mn for n>2. They repeat the calculation for a symmetric CSS input (Eq. 37) and for one-axis-twisted states (Eq. 40), finding that the OAT states give no advantage over the simple CSS input. They then analyze the effect of shot-to-shot fluctuations described by a Gaussian mixture over the left-right distribution (Eqs. 41-43) and claim that the measurement of atom numbers in each site saturates the quantum Cramér-Rao bound.","tokens_in":14491,"tokens_out":32283,"duration_ms":340564,"significance":"The strength of the paper is the analytic, parameter-free derivation of the QFI for the beam-splitting network (Appendices A and B), which I have spot-checked for small M and n; the result is internally consistent. The protocol is appealing because it uses passive linear optics on product Fock inputs and achieves a Heisenberg-like scaling I_q ~ M n^2/2, without resorting to nonlinear interactions. The comparison with OAT is a useful benchmark. However, the significance is conditioned on the fixed-number Fock input: if the BEC loading is described by Glauber coherent states with Poissonian number statistics, the passive network gives only SQL scaling. The paper's fluctuation analysis (Section IIIC) does not cover this case. The optimal-measurement claim is also incomplete for the post-mixing states. With these caveats addressed, the result would be a solid contribution to quantum-enhanced atom interferometry.","major_comments":[{"comment":"The central n^2 scaling in Eq. (34) is derived for the fixed-number Fock input |0,n>_i (Eq. 33). The robustness analysis in Section IIIC replaces the perfect left-right occupation by a mixture over p(m_i)|m_i,n-m_i> (Eq. 41), but the total number of atoms per well, n, is fixed in every component. If the actual BEC loading has Poissonian number statistics, i.e., each well is in a Glauber coherent state |α>_r|0>_l, then the beam splitter e^{-iπ/2 S_x} is a passive linear transformation: a product of coherent states remains a product of coherent states in the rotated modes. For such a state, Wick's theorem gives Var(J_y) = O(n) per well, so I_q ~ M n (the SQL), and the n^2 term in Eq. (34) disappears. The sentence before Eq. (34) says the state 'neglects the shot-to-shot atom-number fluctuations,' but Eq. (41) does not model total-number fluctuations; it models only the left-right distribution. The paper should either analyze the coherent-state/Poissonian case explicitly or clearly restrict the claim to number-squeezed or Fock inputs.","section":"Sec. III.A, Eq. (33) and Sec. IIIC, Eq. (41)"},{"comment":"The proof that population-imbalance measurement saturates the QCRB assumes the input state |ψ_in> has real coefficients in the Fock basis (sentence before Eq. (47)). The states that enter the interferometer after the mixing step are generated by e^{-iπ/2 S_x}, whose beam-splitter transformation contains factors of -i (see Eq. (28)); consequently, the post-mixing Fock-basis coefficients are complex already for M=2, n=1. The claim 'This applies to both the pure CSSs considered in the previous section' is therefore valid only for the CSSs before the mixing, not for the states for which Eq. (34) is derived. The optimality of the number measurement for the actual protocol is unproven. Please either generalize the proof to complex coefficients or qualify the statement.","section":"Sec. IIID, Eqs. (47)-(53)"}],"minor_comments":[{"comment":"The condition 'pm ⩾ 9' should read 'pm ≥ 0'.","section":"Eq. (6)"},{"comment":"There are typos 'buisnessman' and 'modificiation'; please correct them.","section":"Sec. I and Sec. II.D"},{"comment":"The state |0,n>_i is called a coherent spin state, but it is a fixed-number Fock state, not a Glauber coherent state; clarifying this distinction would help readers assess the total-number-fluctuation discussion.","section":"Sec. II.A, Eq. (13)"},{"comment":"The notation 'Fq' is used for the QFI, while the main text uses 'Iq'; please unify the notation.","section":"Appendix A, Eq. (A6)"},{"comment":"The states from Eqs. (33) and (36) are both coherent spin states, not optical coherent states; consider rephrasing 'coherent states' to 'coherent spin states'.","section":"Fig. 3 caption"},{"comment":"The process is repeatedly described as a 'many-body equivalent of the Hong-Ou-Mandel effect.' In the standard HOM effect, one boson is incident on each input port of the beam splitter. In Eq. (33) the input to each inter-well beam splitter is n atoms in one port and vacuum in the other (right mode of well i, left mode of well i+1). The demonstrated enhancement is a consequence of the fixed-number Fock input and the resulting path entanglement, not of two-particle interference between independently prepared sources; I recommend rephrasing the HOM analogy.","section":"Sec. III.A and abstract"}],"recommendation":"major_revision","confidential_remarks":"To the editor: The central analytical result appears correct and the paper is likely to be of interest to the quantum-metrology community. The two issues that require major revision are the incomplete treatment of total-number fluctuations (which could invalidate the claimed robustness for realistic BEC loading) and the optimal-measurement proof (which does not cover the post-mixing complex states). If the authors can fix these by adding an analysis of coherent-state inputs and extending or qualifying the measurement proof, the paper should be suitable for publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThis is a solid, analytically clean paper on using an array of double wells with nearest-neighbor beam-splitter mixing for quantum-enhanced metrology. The main result—I_q = (1/2)M(n^2+2n) for N=Mn atoms prepared as |0,n> Fock states in each well—is a genuine extension of the earlier two-BEC interference work (Ref. [55]), and the multi-well scaling is new. The paper also does useful work comparing against one-axis twisting and showing robustness to a class of shot-to-shot imbalance fluctuations.\n\nThe QFI derivations are careful and the appendix is detailed; I checked the central algebra and it holds up. The authors are upfront that the input state is an idealization, which is good.\n\nThe soft spots are real but not disqualifying. The fixed-number Fock input is load-bearing: swap in coherent states (Poissonian number statistics) and the passive beam splitter just rotates a product state to another product state—the QFI falls to the SQL. Section IIIC only broadens the left-right split while keeping the per-well total n fixed; it does not model total-number fluctuations. So the advantage depends on number-squeezed or Fock-like preparation, and the manuscript should say that explicitly.\n\nSecond, the optimal-measurement proof (Sec. IIID) only works for states with real coefficients in the Fock basis. The states that actually enter the Mach-Zehnder after the mixing have complex coefficients because of the i in the beam-splitter, so the proof doesn't cover the full protocol. It's a secondary claim but a real gap.\n\nThird, the 'many-body HOM' label is a stretch—each beam splitter sees n atoms in one input and vacuum in the other, not the standard one-photon-per-port HOM setting. Naming aside, the physics is fine.\n\nFinally, the experimental realization of the inter-well beam splitter is assumed without discussion. Minor.\n\nBottom line: the core result is correct and the paper deserves a serious referee. It should be revised to address the number-statistics caveat and refine the optimal-measurement claim. I'd send it out.","headline":"A clean analytic result for multi-well interferometry with Fock-state inputs; the n^2 scaling is real but tied to number squeezing, and the optimal-measurement proof has a gap.","tokens_in":15056,"tokens_out":7102,"would_cite":true,"duration_ms":87133,"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":"The paper argues that a passive beam-splitting operation on an array of double wells holding independent Bose-Einstein condensates produces a state whose metrological sensitivity beats the standard quantum limit, scaling as $n^2$ per well…","keywords":["quantum metrology","double-well potentials","Bose-Einstein condensates","Hong-Ou-Mandel effect","quantum Fisher information","beam splitter","Heisenberg scaling","Mach-Zehnder interferometry"],"falsifier":"Prepare the double-well array with coherent-state loading (random atom numbers) rather than number-squeezed Fock states, apply the nearest-neighbor beam-splitting operation, and measure the variance of the collective phase generator or the sensitivity from per-site counting. If the variance grows only linearly with $n$ rather than as $n^2$, so that the quantum Fisher information stays at $I_q \\approx Mn$, the central scaling claim is refuted; the same measurement on number-squeezed inputs should reproduce $I_q = \\frac{1}{2}M(n^2+2n)$.","tokens_in":14001,"feed_emoji":"⚛️","tokens_out":9985,"duration_ms":107630,"temperature":0.7,"pith_summary":"This paper argues that a passive, nearest-neighbor beam-splitting step applied to a one-dimensional array of double wells can create metrologically useful entanglement from initially uncorrelated Bose-Einstein condensates. For $M$ wells each loaded with $n$ atoms in a definite Fock state, the quantum Fisher information after the beam splitter is $I_q = \\frac{1}{2}M(n^2+2n)$, which beats the standard quantum limit $I_q = N = Mn$ for $n>2$ and grows as $n^2$ per well. The effect is identified as a many-body Hong-Ou-Mandel mechanism: bosonic interference at the beam splitter builds nonclassical correlations between neighboring wells. The paper further shows that the $n^2$ enhancement survives shot-to-shot atom-number fluctuations, that one-axis twisting gives no advantage over the simple Fock input, and that counting atoms per site saturates the quantum Cramér-Rao bound. If correct, this provides a scalable route to quantum-enhanced atom interferometry that does not rely on interaction-based squeezing.","feed_headline":"Beam splitting alone beats the standard quantum limit","feed_subtitle":"Just splitting the wells' modes gives n²-per-well sensitivity, beating the shot-noise bound without squeezing","key_machinery":"The load-bearing object is the collective beam-splitting operator $\\hat{S}_x = \\frac{1}{2}\\sum_{i=1}^{M-1}(\\hat{a}_r^{(i)\\dagger}\\hat{a}_l^{(i+1)} + \\mathrm{h.c.})$, which mixes the right mode of well $i$ with the left mode of well $i+1$, and the Heisenberg-picture generator $\\hat{S}_y = e^{i\\pi \\hat{S}_x/2} \\hat{J}_y e^{-i\\pi \\hat{S}_x/2}$ that results after the mixing step. The quantum Fisher information is computed as $I_q = 4\\sum_i \\mathrm{Var}(\\hat{S}_y^{(i)}) + 8\\sum_{i\\neq j}\\mathrm{Cov}(\\hat{S}_y^{(i)},\\hat{S}_y^{(j)})$; the inter-well covariance terms are nonzero only because the beam splitter creates correlations between neighboring wells, which is the signature of the many-body Hong-Ou-Mandel effect. The argument then reduces to evaluating variances and covariances of these bilinear bosonic operators, which the paper does analytically for the two coherent-spin-state inputs, for the one-axis-twisted input, and for the Gaussian fluctuation model.","core_discovery":"The central claim is that the unitary $e^{-i\\pi \\hat{S}_x/2}$, where $\\hat{S}_x$ couples the right mode of each well to the left mode of its neighbor, converts a product of $n$-particle Fock states $|0,n\\rangle$ into a state with quantum Fisher information $I_q = \\frac{1}{2}M(n^2+2n)$ for collective phase estimation. This is above the separable-state limit $I_q = N = Mn$ whenever $n>2$, so the scheme achieves Heisenberg-like scaling in the per-well atom number while retaining shot-noise scaling in the number of wells. The same beam-splitting step applied to symmetrically loaded coherent spin states gives $I_q \\approx \\frac{1}{8}Mn^2$, still sub-SQL, while input states prepared by one-axis twisting reach at most roughly $\\frac{1}{2}Mn^2$, matching rather than improving on the Fock input. For incoherent Gaussian fluctuations of the atom distribution between the two sites, the limiting QFI is $\\frac{3M-2}{8}(n^2+2n)$, so the enhancement is robust. Finally, for any input state with real coefficients in the Fock basis, the classical Fisher information of per-site atom-number measurements at $\\theta=0$ equals $4\\langle \\hat{J}_y^2 \\rangle$, saturating the quantum Cramér-Rao bound.","pith_inferences":["A testable consequence not pursued in the paper: sweeping the beam-splitter strength on a tunable superlattice should show the predicted $n^2$ variance growth together with sub-Poissonian number correlations between neighboring wells.","If the mechanism is as described, the same passive mixing idea could be adapted to fermionic atoms, where antisymmetrization would replace bunching with antibunching and likely reverse the sign of the covariance terms, suppressing the gain.","The analysis assumes the beam-splitting step is instantaneous and lossless; a natural extension would include finite-depth lattice shaking or particle loss and check whether the $n^2$ scaling survives realistic decoherence times.","The authors note that long-range coherence would be needed to beat the SQL in $M$ as well; quantifying the QFI for all-to-all beam-splitting couplings is a natural, still-open extension."],"forward_implications":["With the Fock input, the predicted phase sensitivity is $\\Delta\\theta = \\sqrt{2/[M(n^2+2n)]}$, beating the SQL $1/\\sqrt{Mn}$ for $n>2$.","Symmetrically loaded coherent spin states still beat the SQL with $I_q \\approx \\frac{1}{8}Mn^2$, so the protocol does not require perfect single-site loading.","Shot-to-shot atom-number fluctuations only reduce the enhancement to $I_q \\approx \\frac{3}{8}Mn^2$ in the large-fluctuation limit, preserving the $n^2$ scaling.","Number counting at each site is an optimal measurement for real-coefficient inputs, so the predicted QCRB sensitivity is saturable with a simple detection scheme.","One-axis twisting offers no advantage over the product Fock state in this setup, meaning the interaction-based entanglement step can be omitted."],"supporting_citations":[{"why":"Supplies the quantum Fisher information and the Cramér-Rao bound that set the sensitivity limits used throughout.","marker":"[1]"},{"why":"The recent Mach-Zehnder atom-interferometry experiment on an array of double wells whose scheme this paper analyzes and extends.","marker":"[47]"},{"why":"Defines the Hong-Ou-Mandel interference whose many-body analog is identified as the mechanism creating the entangled state.","marker":"[48]"},{"why":"Demonstrates long-lived coherence in the double-well array, supporting the assumed scalability of the setup.","marker":"[46]"},{"why":"Provides the one-axis-twisting protocol used as a reference entangled input for comparison.","marker":"[49]"}],"fun_headline_variants":["Beam splitter alone beats shot-noise limit","Double-well arrays reach Heisenberg scaling","No squeezing: Beam splitting boosts sensitivity","Atom arrays: simple splitting surpasses SQL","HOM effect in double wells enhances metrology"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The central calculation assumes each well starts in a pure number state with exactly $n$ atoms in one site and none in the other; if the wells instead start with random atom numbers, as in an ordinary coherent state, a passive beam splitter creates no number entanglement and the $n^2$ enhancement is lost.","fun_headline_variants_meta":{"raw":{"variants":["Beam splitter alone beats shot-noise limit","Double-well arrays reach Heisenberg scaling","No squeezing: Beam splitting boosts sensitivity","Atom arrays: simple splitting surpasses SQL","HOM effect in double wells enhances metrology"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000205,"raw_usage":{"total_tokens":1418,"prompt_tokens":998,"completion_tokens":420,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":614,"completion_tokens_details":{"reasoning_tokens":354}},"tokens_in":614,"tokens_out":420,"duration_ms":5728,"temperature":1.0,"reasoning_tokens":354,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T17:12:05.793371+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Prepare the double-well array with coherent-state loading (random atom numbers) rather than number-squeezed Fock states, apply the nearest-neighbor beam-splitting operation, and measure the variance of the collective phase generator or the sensitivity from per-site counting. If the variance grows only linearly with $n$ rather than as $n^2$, so that the quantum Fisher information stays at $I_q \\approx Mn$, the central scaling claim is refuted; the same measurement on number-squeezed inputs should reproduce $I_q = \\frac{1}{2}M(n^2+2n)$.","supporting_citations":[{"cited_title":"Mach-Zehnder atom interferometry with non-interacting trapped Bose Einstein condensates","cited_arxiv_id":"2504.17391","evidence_quote":"The recent Mach-Zehnder atom-interferometry experiment on an array of double wells whose scheme this paper analyzes and extends."},{"cited_title":"beat-note","cited_arxiv_id":null,"evidence_quote":"Demonstrates long-lived coherence in the double-well array, supporting the assumed scalability of the setup."},{"cited_title":"Płodzień, M","cited_arxiv_id":null,"evidence_quote":"Provides the one-axis-twisting protocol used as a reference entangled input for comparison."}],"review_version":1}