{"id":"504b772b-44a9-4e65-8df7-37b986df5204","arxiv_id":"1908.08904","paper_version":3,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"A variational algorithm finds non-symmetric quantum probe states that significantly outperform conventional symmetric states for noisy quantum metrology on up to 9 qubits.","lead":"Quantum metrology researchers used variational quantum circuits to search for probe states that stay precise under realistic noise. For systems up to 9 qubits, they found asymmetric states that outperform conventional symmetric states like GHZ and squeezed states by up to a factor of 2 under amplitude damping and Pauli noise.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 'beats every symmetric state' conclusion depends on the symmetric-subspace optimizer having found the global optimum, which the paper explicitly does not guarantee.","rationale":"Reader's verdict CONDITIONAL is appropriate. My independent read agrees that the paper is a solid numerical study with a genuine symmetry-breaking finding, and the main risk is not internal inconsistency but the strength of the quantifier 'every symmetric state' in the presence of unverified global optimality. I focus the concern on the symmetric subspace because the ansatz state itself is an explicit witness; the only way the qualitative conclusion fails is if a better symmetric state exists. This is exactly the assumption the paper disclaims. I also considered the Appendix C analytical model: its per-basis ratio estimates (C.4-C.5) plus the factor N from multiple bases is heuristic and fitted to the numerics, but it is presented as explanatory rather than as proof, and the numerical QFI computations are exact for the simulated circuits. So the analytical gap does not threaten the central numerical claim. The concrete test is inexpensive because the symmetric subspace is small; if it passes, the paper's conditional verdict can be upgraded; if it fails, the central claim would need to be weakened to 'outperform previously known symmetric states'. Thus I recommend keeping the CONDITIONAL verdict.","tokens_in":25164,"tokens_out":5168,"duration_ms":56226,"concrete_test":"Run a dedicated global optimization on the symmetric subspace for the amplitude-damping N=8 and N=9 cases (and one inhomogeneous Pauli case, e.g., N=8), using the released QuEST/Mathematica code. For example, use multi-start differential evolution or a dense quasi-Monte Carlo grid over the normalized complex coefficients c_m (dimension 2(N+1)-2 = 18 for N=8), with the same objective gamma/T (Delta omega)^{-2}_max. If the best symmetric precision found exceeds the reported green ansatz precision, the claimed advantage fails; if it reproduces the brown curve, the load-bearing assumption is supported.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central numerical claim for amplitude damping and inhomogeneous Pauli errors is that ansatz states (green in Fig. 4) outperform 'any symmetric state' (brown). Both curves come from the same randomized adaptive coordinate descent routine (Sec. 5.2), and the paper states it 'cannot guarantee global optimality in general' (Sec. 5.2) and that 'verifying global optimality of the symmetry-breaking states is beyond the scope of the current work' (Sec. 7). The non-symmetric side of the comparison is less fragile: the optimized ansatz circuit parameters provide an explicit witness state, so even if the non-symmetric search is incomplete, the found state still exists. The fragile side is the symmetric baseline: the witness outperforms the symmetric family only if the symmetric optimization did not stall in a local optimum. The symmetric subspace is only N+1 dimensional (coefficients c_m in Eq. 12), so for N<=9 it is a small continuous search, but coordinate descent with random restarts is still not a global optimizer and the cost landscape is nonconvex. The paper validates the optimizer on dephasing and Ornstein-Uhlenbeck cases where analytic optima are known, but no such validation exists for amplitude damping or inhomogeneous Pauli errors. Hence the key existence claim remains conditional on the completeness of the symmetric search.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a variational quantum algorithm for finding near-optimal probe states for quantum metrology in the presence of noise. The probe state is prepared by a parameterized shallow circuit, allowed to evolve under a field Hamiltonian together with a continuous noisy channel, and the precision is quantified by the quantum Fisher information computed via a fidelity-based finite-difference approximation. The authors simulate systems of up to 9 qubits exactly with QuEST for dephasing, amplitude damping, inhomogeneous Pauli errors, and Ornstein-Uhlenbeck noise, optimizing both the circuit parameters and the sensing time. Their central numerical finding is that, for amplitude damping and inhomogeneous Pauli errors, permutation-non-symmetric ansatz states outperform the permutation-symmetric states they optimized (GHZ, squeezed, and general symmetric Dicke mixtures) by up to a constant factor of about 2. For amplitude damping, they provide an analytical model in which a broken-symmetry component permits first-order T1 decay events to be individually resolved, yielding additional Fisher information. They also outline an experimental implementation with encoder and decoder circuits and analyze the effect of imperfect preparation gates in an appendix.","tokens_in":25334,"tokens_out":5082,"duration_ms":54218,"significance":"If the central numerical claim holds, the paper makes a useful and somewhat counterintuitive contribution to noisy quantum metrology: it shows, for small numbers of qubits, that relaxing permutation symmetry can yield a constant-factor advantage over the symmetric states that have dominated the literature, and it gives an intuitive analytical explanation for the amplitude-damping case. The work is strengthened by exact numerical simulations using QuEST, by validation against known analytic optima for dephasing and Ornstein-Uhlenbeck noise, and by the public release of the simulation code. The analytical model in Appendix C, while heuristic, is a concrete falsifiable description of the mechanism. The main weakness is that the 'outperforms any symmetric state' claim depends on the completeness of a randomized optimization over the symmetric subspace, which the authors explicitly do not guarantee; this tempers the strength of the headline claim but does not undermine the existence of the explicit ansatz states themselves.","major_comments":[{"comment":"The central claim that the ansatz states 'outperform any symmetric state' is not established by the reported numerics, because the symmetric baseline is produced by the same randomized adaptive coordinate descent routine whose global optimality the authors explicitly disclaim ('we cannot guarantee global optimality in general', Sec. 5.2; 'verifying global optimality of the symmetry-breaking states is beyond the scope of the current work', Sec. 7). Since the symmetric subspace has dimension N+1, for N up to 9 a substantially more exhaustive search (dense grid plus local refinement, or a certified global optimizer) is feasible and would make the comparison conclusive; alternatively, the claim should be weakened to 'outperform all symmetric states found by our optimizer and all previously known explicit state families'.","section":"Sec. 5.2, Sec. 7, Fig. 4(a)(mid/right)"},{"comment":"The optimizer is validated only for dephasing and Ornstein-Uhlenbeck noise, where analytic optima are known and the ansatz, symmetric, and squeezed curves essentially coincide. No independent validation is supplied for amplitude damping or inhomogeneous Pauli errors, which are precisely the two channels where the symmetry-breaking advantage is claimed. A concrete test would be to compare the optimized symmetric curve against known asymptotic upper bounds for these channels, or to repeat the symmetric-subspace optimization for N up to 9 with a qualitatively different global search method, to confirm that the brown curve in Fig. 4 is not a local optimum.","section":"Sec. 5.3 and Fig. 4"},{"comment":"The analytical argument that the non-symmetric scheme gains a factor proportional to N relies on the ratios in Eqs. (C.4) and (C.5) being O(N^0) for every j, but the derivation only states this asymptotic order without showing the constants or the range of N for which the individual contributions remain comparable to the symmetric single-basis contribution. Since the claimed advantage is a constant factor at finite N (not an asymptotic scaling), the argument would be more convincing if the authors provided the explicit expressions for prob(Aj) and its derivative, or a plot of the ratio F_a/F_s versus N for the optimized coefficients.","section":"Appendix C, Eqs. (C.1)–(C.5)"}],"minor_comments":[{"comment":"There are several typos and OCR artifacts, including 'Winger' in the Fig. 1 caption, 'follwing' in the introduction, 'its its' in the Fig. 7 caption, and garbled author names in Ref. [40]; these should be corrected.","section":"Throughout"},{"comment":"The squeezed-state definition contains an extraneous time variable t in the exponents e^{-iθ3 t Jz}, e^{-iθ2 t Jx}, and e^{-iθ1 t Jz^2}; since t is the sensing time and is optimized separately, the formula should use the rotation angles θ_i alone or define the effective angles explicitly to avoid confusion.","section":"Sec. 5.1"},{"comment":"The text refers both to 'all 2N bases' and to 'N distinguishable measurement bases'; because each j contributes two bases (the ± signs), the counting in Eqs. (C.1)–(C.3) should be clarified, and the sum index should be explicit.","section":"Appendix C"},{"comment":"The phrase 'passively correct first-order decay events' may overstate the mechanism; the analysis in Appendix C describes extracting additional Fisher information from first-order decay outcomes, not correcting the state, so consider rewording to 'extract additional Fisher information from first-order decay events'.","section":"Sec. 5.4"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is a solid numerical study with reproducible code and a clear analytical hypothesis. The revision should focus on either strengthening the symmetric-subspace baseline search for N up to 9 or softening the 'any symmetric state' claim; with that change, the paper would be publishable. I do not see concerns about novelty or citation behavior."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Here's the take: this paper has a real result. For amplitude damping and inhomogeneous Pauli noise, a shallow variational circuit finds N≤9 qubit states that are not permutation symmetric and that beat the best permutation-symmetric states found by the same optimization routine by up to a factor of about 2. The states are explicit witnesses, the simulations are exact (QuEST), and the pipeline validates against known analytic optima for dephasing and Ornstein-Uhlenbeck noise. I believe the core numerical claim: symmetry-breaking helps under these noise models.\n\nThe paper does several things right. It reproduces analytic baselines where they exist, which is exactly the right check for a heuristic optimizer. The ansatz is shallow and plausible for near-term hardware. The code is public. And the mechanism for the improvement—resolving individual T1 events via a basis that is not symmetric—is concrete and, as far as I know, new.\n\nThe soft spots are proportional. The headline 'outperforms any symmetric state' is stronger than what the numerics establish. The symmetric baseline is produced by the same randomized coordinate descent, and the paper explicitly says it cannot guarantee global optimality for the symmetry-breaking cases. On the non-symmetric side this is fine: the ansatz state is an honest witness. On the symmetric side, the claim depends on the optimizer having found the symmetric optimum. The symmetric subspace is small (N+1 dimensions) and the optimizer is validated on dephasing and OU, so the risk is moderate, not severe. But it is real. A more exhaustive symmetric search could in principle close part of the gap. 'Any symmetric state' should be softened to 'the symmetric states we optimized' or supplemented with a more thorough search.\n\nThe analytical appendix is the second soft spot. The argument—per-basis sensitivities are comparable and there are O(N) such bases—is a plausible heuristic, but the step to a total Fisher-information advantage is sketched, not derived, and the coefficients are fitted to the optima. The paper mostly labels it as an analysis of the optimal states, but Fig. C1 is a post-hoc account, not a proof. Also, the appendix uses '2N bases' and 'N distinguishable bases' seemingly interchangeably; that needs cleaning up.\n\nMinor reproducibility point: the GitHub repo has no commit hash and the code is Mathematica-dependent. For a numerical paper, that is a small but avoidable hurdle.\n\nWho is this for? Anyone working on variational quantum algorithms, quantum metrology with noise, or near-term sensing. It deserves a serious referee. The main claim should be qualified, and the analytical model should either be tightened or explicitly marked as a conjecture, but the result is solid enough to justify the referee time.\n\nRecommendation: send it to review, with the expectation of revision.","headline":"Solid variational metrology paper with a real finding—symmetry-broken states beat symmetric ones under amplitude damping—but the headline claim rests on a non-global optimizer for the symmetric baseline.","tokens_in":25922,"tokens_out":11528,"would_cite":true,"duration_ms":107802,"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":"Variational metrology finds asymmetric entangled probe states that outperform every symmetric state under noise by up to a factor of 2.","keywords":["variational quantum metrology","quantum Fisher information","noise-robust probe states","permutation symmetry breaking","amplitude damping","non-symmetric entangled states","near-term quantum hardware","Ramsey interferometry"],"falsifier":"Run a certified global search over the permutation-symmetric subspace, i.e., Dicke-state superpositions, for $N = 6$ and $N = 8$ under amplitude damping with $p(t) = 1 - e^{-\\gamma t}$, optimizing $\\gamma/T\\,(\\Delta\\omega)^{-2}$ over $t$ and the coefficients $c_m$, and compare with the best symmetry-broken ansatz state reported. If any symmetric state reaches or exceeds the ansatz value, the claimed symmetry-breaking advantage is refuted for that $N$; equivalently, the explicit state $|\\psi_a\\rangle$ in Eq. (19) can be tested against the best symmetric state through the classical-Fisher decomposition in Appendix C.","tokens_in":24861,"feed_emoji":"⚛️","tokens_out":7482,"duration_ms":74609,"temperature":0.7,"pith_summary":"The paper proposes a variational algorithm that prepares a parameterized probe state on a quantum circuit and optimizes its metrological precision directly, without assuming a particular noise model. Simulating systems of up to 9 qubits under dephasing, amplitude damping, inhomogeneous Pauli errors, and non-Markovian Ornstein-Uhlenbeck noise, it finds that the best states are not always permutation-symmetric. For amplitude damping and inhomogeneous Pauli errors, non-symmetric highly entangled states outperform every symmetric state the authors optimized, improving the dimensionless precision by up to a factor of 2. This matters because previously known metrology states such as GHZ, squeezed, and symmetric Dicke states are permutation-symmetric, and the paper gives an analytical model for why breaking that symmetry helps: it lets the measurement resolve individual first-order decay events. The result is a practical, device-tailored route to noise-robust quantum sensing on near-term hardware.","feed_headline":"Asymmetric states double metrology precision under noise","feed_subtitle":"A variational search over up to 9 qubits finds entangled probes that beat every symmetric rival under amplitude damping.","key_machinery":"The central object is a shallow variational encoder circuit $U_E(\\theta)$ with a linear number of rotation parameters, together with a cost function built from the quantum Fisher information via the fidelity formula $F_Q = 8\\lim_{\\delta\\omega\\to 0}[1 - \\mathrm{Fid}(\\rho_0, \\rho_1)]/(\\delta\\omega)^2$. The optimizer maximizes the dimensionless precision $\\gamma/T\\,(\\Delta\\omega)^{-2}$ over both circuit parameters and sensing time $t$. The load-bearing mechanism is the symmetry-breaking component $|D\\rangle$ in Eq. (19): unlike the symmetric Dicke state $|J,J-2\\rangle$, $|D\\rangle$ contains only paired excitations on specific qubit pairs, so after amplitude damping the optimal measurement includes $N$ distinguishable bases $|A_j\\rangle = T_+^{(j)}(b_1|11\\cdots 1\\rangle \\pm b_2\\sqrt{N/2}|D\\rangle)$. Each such basis carries a first-order relaxation event, giving a sum of $N$ comparable Fisher-information contributions instead of one, which is the analytical reason for the superior performance.","core_discovery":"The paper claims that under amplitude damping and inhomogeneous Pauli errors, optimal metrological probe states for 3 to 9 qubits are not symmetric under permutations, and that these symmetry-broken states achieve up to twice the precision of the best permutation-symmetric states, including GHZ states, one-axis twisted squeezed states, and general symmetric Dicke superpositions. The advantage comes from an additional component $|D\\rangle = \\sqrt{2/N}(|1100\\cdots 000\\rangle + |0011\\cdots 000\\rangle + \\cdots + |0000\\cdots 011\\rangle)$ added to a superposition of $|00\\cdots 0\\rangle$ and $|11\\cdots 1\\rangle$, which allows first-order $T_1$ decay events on individual qubits to be resolved in $2N$ separate measurement bases rather than one collective basis. The paper also confirms in the dephasing and Ornstein-Uhlenbeck cases that its variational search reproduces known optimal states, so the non-symmetric finding is presented as a genuine exception to the usual symmetric ansatz rather than a failure of the optimizer.","pith_inferences":["If the Appendix C mechanism, $N$ distinguishable first-order error-resolving measurement bases, persists for larger $N$, the constant-factor advantage may grow with qubit number even though the asymptotic scaling stays linear; the paper only claims the advantage for $N \\leq 9$.","A testable heuristic follows: any noise model with a non-rotationally-symmetric error axis and distinguishable single-qubit decay sectors should show a similar symmetry-breaking benefit, while rotationally symmetric dephasing should not.","The same cost function could be extended to include active error correction or pulse-control parameters, potentially combining passive first-order correction with active protection; this extends the paper's variational scheme beyond passive probe states."],"forward_implications":["A near-term quantum device can run the variational loop directly with encoder and decoder circuits, and the resulting probe state is tailored to the device's own dominant noise without separate process tomography.","For amplitude damping the symmetry-breaking advantage appears for $N \\geq 5$ and persists when preparation circuits are subject to depolarizing gate noise at realistic rates, so it is compatible with imperfect hardware.","Under dephasing and Ornstein-Uhlenbeck noise the same ansatz reproduces known optimal behavior, squeezed-like states and GHZ states respectively, showing the method recovers established results where symmetry is not broken.","The gain over symmetric states is a constant factor, not a new scaling: the precision remains linear in $N$ under Markovian noise, consistent with asymptotic no-go bounds."],"supporting_citations":[{"why":"Supplies the dephasing benchmark formulas and establishes that GHZ states achieve only standard quantum limit scaling under dephasing, the key comparison baseline.","marker":"[23]"},{"why":"Defines the quantum Fisher information, Cramér-Rao bounds, and symmetric Dicke-state formalism that the whole optimization target relies on.","marker":"[12]"},{"why":"Provides the spin-squeezing decoherence limit and the squeezed-state family that the paper optimizes and compares against.","marker":"[38]"},{"why":"Gives the general framework for ultimate precision limits in noisy quantum-enhanced metrology, supporting the claim that only constant-factor gains are possible.","marker":"[39]"},{"why":"Establishes linear asymptotic quantum Fisher information bounds for uncorrelated Markovian noise, used for the amplitude-damping scaling discussion.","marker":"[24]"},{"why":"Provides the dimensionless precision formulation and optimal sensing-time analysis that structure the numerical optimization.","marker":"[37]"},{"why":"Shows improved scaling with GHZ states under non-Markovian Ornstein-Uhlenbeck noise, the comparison case for that noise model.","marker":"[26]"},{"why":"Supplies the explicit Kraus operators and time-dependent dephasing probability used to simulate Ornstein-Uhlenbeck noise.","marker":"[67]"},{"why":"The exact quantum-circuit simulator used to model noisy evolution and compute the fidelity-based precision.","marker":"[51]"}],"fun_headline_variants":["Symmetry-breaking probes double metrology precision","Variational search finds asymmetric states that beat symmetric rivals","Quantum metrology: broken symmetry boosts precision twofold","Asymmetric entangled states outperform symmetric ones in noisy metrology","Variational metrology finds asymmetric states that double precision"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the randomized coordinate-descent search reached the true global optimum within both the shallow ansatz family and the symmetric subspace for $N$ up to 9 under amplitude damping and Pauli errors, and that the shallow ansatz is expressive enough to contain the optimal state; the paper explicitly states it cannot guarantee global optimality.","fun_headline_variants_meta":{"raw":{"variants":["Symmetry-breaking probes double metrology precision","Variational search finds asymmetric states that beat symmetric rivals","Quantum metrology: broken symmetry boosts precision twofold","Asymmetric entangled states outperform symmetric ones in noisy metrology","Variational metrology finds asymmetric states that double precision"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000547,"raw_usage":{"total_tokens":2601,"prompt_tokens":917,"completion_tokens":1684,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":533,"completion_tokens_details":{"reasoning_tokens":1610}},"tokens_in":533,"tokens_out":1684,"duration_ms":12518,"temperature":1.0,"reasoning_tokens":1610,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:27:28.653971+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run a certified global search over the permutation-symmetric subspace, i.e., Dicke-state superpositions, for $N = 6$ and $N = 8$ under amplitude damping with $p(t) = 1 - e^{-\\gamma t}$, optimizing $\\gamma/T\\,(\\Delta\\omega)^{-2}$ over $t$ and the coefficients $c_m$, and compare with the best symmetry-broken ansatz state reported. If any symmetric state reaches or exceeds the ansatz value, the claimed symmetry-breaking advantage is refuted for that $N$; equivalently, the explicit state $|\\psi_a\\rangle$ in Eq. (19) can be tested against the best symmetric state through the classical-Fisher decomposition in Appendix C.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the dephasing benchmark formulas and establishes that GHZ states achieve only standard quantum limit scaling under dephasing, the key comparison baseline."},{"cited_title":"Oberthaler, Roman Schmied, and Philipp Treutlein","cited_arxiv_id":null,"evidence_quote":"Defines the quantum Fisher information, Cramér-Rao bounds, and symmetric Dicke-state formalism that the whole optimization target relies on."},{"cited_title":"Spin squeezing and decoherence limit in Ramsey spectroscopy","cited_arxiv_id":null,"evidence_quote":"Provides the spin-squeezing decoherence limit and the squeezed-state family that the paper optimizes and compares against."},{"cited_title":"General framework for estimating the ultimate precision limit in noisy quantum-enhanced metrology","cited_arxiv_id":null,"evidence_quote":"Gives the general framework for ultimate precision limits in noisy quantum-enhanced metrology, supporting the claim that only constant-factor gains are possible."},{"cited_title":"Efficient tools for quantum metrology with uncorrelated noise","cited_arxiv_id":null,"evidence_quote":"Establishes linear asymptotic quantum Fisher information bounds for uncorrelated Markovian noise, used for the amplitude-damping scaling discussion."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the dimensionless precision formulation and optimal sensing-time analysis that structure the numerical optimization."},{"cited_title":"Benjamin, and Joseph Fitzsimons","cited_arxiv_id":null,"evidence_quote":"Shows improved scaling with GHZ states under non-Markovian Ornstein-Uhlenbeck noise, the comparison case for that noise model."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the explicit Kraus operators and time-dependent dephasing probability used to simulate Ornstein-Uhlenbeck noise."},{"cited_title":"QuEST and high performance simulation of quantum computers","cited_arxiv_id":null,"evidence_quote":"The exact quantum-circuit simulator used to model noisy evolution and compute the fidelity-based precision."}],"review_version":1}