{"id":"cc2d2ba5-d7a8-4f3c-80bc-e510e3694278","arxiv_id":"2608.08130","paper_version":1,"verdict":"REJECT","confidence":"HIGH","novelty_score":5.0,"correctness_risk":"high","formal_verification":"none","parameter_count":0,"one_line_summary":"A correlated-noise correction protocol preserves stored quantum Fisher information for dephasing and bit-flip noise, but its claimed protection against amplitude-damping noise is incorrect.","lead":"The paper proposes a method to protect a quantum sensor's stored probe from errors that happen while the probe waits between measurement rounds. Special extra qubits detect the errors, which could let quantum sensors keep their precision in larger quantum computing tasks.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Amplitude-damping protection fails as written: the control-X gate of Eq. (13) does not satisfy the correlation condition Eq. (10) for σ_-, so a single damping jump erases the encoded phase while leaving an ambiguous syndrome.","rationale":"The paper's core new contribution is protecting already-encoded metrological information during storage. For pure dephasing and bit-flip, the control gates and syndrome logic are standard: T†(I⊗σ_z)T=σ_z^a⊗σ_z^p and T†(I⊗σ_x)T=σ_z^a⊗σ_x^p, so the QFI preservation claim for those channels is credible. The decisive question is whether the amplitude-damping channel can be handled by the same framework. It cannot as written: the gate in Eq. (13) conjugates σ_- to a conditional |0><0|σ_- + |1><1|σ_+ operator, violating Eq. (10). Working through a single jump shows the ancilla syndrome is ambiguous and the probe conditional state has no φ dependence. This directly falsifies the central abstract/conclusion claim that CNC protects the quantum Fisher information against amplitude-damping noise. The reader's verdict correctly identifies this as load-bearing. I see no reason to soften the rejection: the error is in a central advertised capability, not in a peripheral numerical detail. I would keep the REJECT verdict. A concrete simulation with the Kraus operators of amplitude damping would settle the issue definitively.","tokens_in":10854,"tokens_out":16806,"duration_ms":176501,"concrete_test":"Compute the final QFI for the Sec. IV C protocol using the exact Kraus operators of amplitude damping (E0=diag(1,√(1-p)), E1=σ_-), with Tz as in Eq. (13), ancilla |+>, and the conditional corrections claimed in Fig. 4. For a single jump the post-correction state is |+>_p or |−>_p with equal probability, independent of φ, so the phase QFI vanishes in that branch and the total F_Q is strictly below 4t_s^2 for any p>0. If a controlled-Z gate is substituted for Eq. (13), the jump branch is flagged, but postselecting it leaves F_Q=4t_s^2·4(1-p)/(2-p)^2, again less than ideal; either way Fig. 4's ideal restoration is contradicted.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The amplitude-damping section (Sec. IV C) reuses the dephasing gate Tz=|0><0|⊗I+|1><1|⊗σ_x (Eq. 13). For L=σ_-, the required condition Eq. (10) would be T_z†(I⊗σ_-)T_z = P^(a)⊗σ_-, but the actual identity is T_z†(I⊗σ_-)T_z = |0><0|⊗σ_- + |1><1|⊗σ_+. Acting on |+>_a(α|0>+β|1>) and applying T† after one jump gives (β|0>_a|0>_p + α|1>_a|1>_p)/√2 (unnormalized). An X-basis measurement of the ancilla then yields + or − with equal probability, and the conditional probe state is |+> or |−>, in which the encoded phase φ has been completely erased. The no-jump branch is not a clean syndrome either: it produces both ancilla outcomes with coefficients α±β, so a single amplitude-damping event cannot be distinguished or corrected by this protocol. Thus the claim that Fig. 4 restores the ideal QFI is unsupported. Since the abstract and Sec. VI explicitly include amplitude damping among the protected noise channels, this is a load-bearing failure, not a peripheral gap. The dephasing and bit-flip constructions are standard and appear sound.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a correlated-noise correction (CNC) protocol for protecting quantum Fisher information during the storage stage after signal encoding. The protocol entangles the probe with auxiliary qubits before storage, allows the memory channel to act, then applies the inverse entangling gates and reads out all syndromes at the end. The authors claim that this preserves Heisenberg-limited QFI for single- and multi-qubit probes under dephasing, bit-flip, and amplitude-damping noise, and they distinguish between preserving QFI for immediate readout and fully restoring the probe state for later quantum processing. The dephasing and bit-flip constructions are standard and appear sound; the amplitude-damping analysis, however, is not justified by the general framework as written, although the final QFI claim can be recovered by an explicit calculation that the manuscript does not provide.","tokens_in":11128,"tokens_out":55192,"duration_ms":522058,"significance":"The storage-stage viewpoint is a legitimate and relatively unexplored gap in QEC-enhanced metrology: once a parameter is encoded, the probe may need to survive a memory interval before further processing. The paper's key observation that preserving QFI does not require full state recovery when the probe is read out immediately is useful and illustrated clearly in the bit-flip/dephasing example. If the amplitude-damping section is corrected with a proper derivation, the central claim would be solid; the proposed protocol is simple, uses only fixed two-body gates, and naturally extends to multi-qubit probes. The numerical demonstrations support the main idea, though the figures lack sufficient detail for full reproducibility.","major_comments":[{"comment":"The gate T_z does not satisfy the required correlation condition Eq. (10) for L=σ_-. A direct calculation gives T_z^†(I⊗σ_-)T_z = |0><0|⊗σ_- + |1><1|⊗σ_+, which is not of the form P^(a)⊗σ_-. Consequently, the statement that amplitude damping can be handled by 'the same one-auxiliary CNC construction used for dephasing noise' is not supported by the general framework. This is load-bearing because amplitude damping is explicitly listed in the abstract and conclusions. However, the conclusion of Fig. 4 can be recovered: modelling the channel by Kraus operators E_0=|0><0|+k|1><1| and E_1=c|0><1| with c=√(1-k²), and applying σ_z when the ancilla outcome is |−>, gives ρ_out = p|ψ><ψ|+(1-p)σ_x|ψ><ψ|σ_x with p=(1+k²)/2, and this state has exactly the same QFI (1 in the φ parametrization, i.e., 4t_s² for the σ_z encoding) for all k. The paper must either include such a derivation or weaken the amplitude-damping claim to what the current text actually proves.","section":"Sec. IV C, Eq. (13)"},{"comment":"The framework is presented as necessary: 'the entangling gate T_i is required to satisfy Eq. (10).' The amplitude-damping example violates this requirement, making the framework internally inconsistent. The correct statement is that Eq. (10) is sufficient for converting Pauli-type errors into clean ancilla syndromes, but it is not necessary for preserving QFI, as the amplitude-damping twirl effect shows. The authors should reformulate the general condition as a sufficient design rule, or generalize it to cover non-self-adjoint and non-Pauli noise, and they should explicitly flag where the amplitude-damping protocol departs from the clean-syndrome picture.","section":"Sec. III, Eq. (10)"}],"minor_comments":[{"comment":"The correction rule for the amplitude-damping case is not specified. The reader must infer from the dephasing case that σ_z is applied when the ancilla is found in |−⟩. Please state the rule explicitly, together with the amplitude-damping Kraus operators and the relation between k, c, and the damping rate γ.","section":"Sec. IV C"},{"comment":"The numerical simulations are not reproducible from the text: no master equations, damping rates, auxiliary-noise rates, or initialization/measurement details are given. Please add the parameter values and the precise protocol used for each figure, especially the amplitude-damping simulation in Fig. 4.","section":"Figs. 2–6"},{"comment":"The QFI formula uses F(ρ_ω,ρ_{ω+dω}) without specifying whether F is the Uhlmann fidelity or its square. The standard relation is F_Q = 8(1−√F)/dω² when F is the Uhlmann fidelity; please clarify to avoid a factor-of-two ambiguity.","section":"Eq. (4)"},{"comment":"'using a Shur code' should presumably read 'using a Shor code.' Also, several typographical issues appear in the extracted text (e.g., 'withωencoded', 'F C andF Q'); please correct these in the final version.","section":"Sec. IV A"}],"recommendation":"major_revision","confidential_remarks":"The stress-test concern about amplitude damping is not fatal as stated. A Kraus-operator calculation shows that the proposed T_z gate plus σ_z correction produces a twirled state p|ψ><ψ|+(1−p)σ_x|ψ><ψ|σ_x whose QFI for the Z-encoding is exactly the ideal value for all p. The real problem is that the manuscript does not provide this calculation and instead implies that Eq. (10) is satisfied. I recommend requiring the authors to supply the missing derivation and to correct the general framework's sufficiency claim. If they do so, the paper could be acceptable; the current presentation is too incomplete for the amplitude-damping claim to stand as written."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The reader and the stress test both flag the amplitude-damping section as incorrect, and I think that objection does not hold up on the math. For a single jump, after T† and an X-basis ancilla readout, the conditional probe states are (β|0> ± α|1>)/√2 — that is, X|ψ> and ZX|ψ>, not |+> and |−> as the stress test claims. Both are unitarily equivalent to the ideal state with φ-independent unitaries, so the QFI is exactly preserved. The no-jump branch is also clean: it gives |ψ> for one ancilla outcome and Z|ψ> for the other, and the same σ_z correction used for dephasing maps both to |ψ>. The final state is a mixture of |ψ> and X|ψ>, and that mixture has QFI 4t_s² for any mixture weight. So the amplitude-damping claim is actually correct, at least for the single-qubit case with the natural correction rule.\n\nWhat the paper does well: the storage-stage framing is a legitimate extension, the dephasing and bit-flip constructions are standard and correctly executed, and the partial-correction observation — QFI can survive even when the stored state is not fully recovered — is real and clearly demonstrated in the bit-flip/dephasing section. The multi-qubit and multi-round numerics are consistent with the claims.\n\nThe real soft spot is not the physics but the paper's own framework. The general condition Eq. (10) is not satisfied by the amplitude-damping gate: T†(I⊗σ_-)T = |0><0|⊗σ_- + |1><1|⊗σ_+, which is not of the form P⊗σ_-. The paper never flags this and does not supply the missing calculation; the amplitude-damping section is asserted rather than derived. That is a genuine rigor gap, and a referee should ask for the explicit calculation. The novelty is modest — this is essentially textbook syndrome extraction applied to the storage stage — but the claims are basically sound.\n\nWho this is for: people working on sensing-enabled quantum information processing or quantum memories. It is not a breakthrough, but it is a correct and useful building block. I would send it to peer review, not desk-reject it, with a request to fix the Eq. (10) condition and add the amplitude-damping derivation.","headline":"The stress test's amplitude-damping objection is wrong: the protocol actually preserves QFI under amplitude damping, though the paper's general framework condition Eq. (10) is not satisfied and the section lacks a real proof.","tokens_in":11642,"tokens_out":37819,"would_cite":false,"duration_ms":319167,"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":"A correlated-noise correction protocol protects Heisenberg-limited metrological information during storage by entangling the probe with auxiliary qubits and extracting all syndromes once, after the memory stage.","keywords":["quantum metrology","Heisenberg limit","quantum Fisher information","quantum error correction","quantum memory","correlated-noise correction","amplitude damping","multi-qubit probes"],"falsifier":"Prepare a $Z$-encoded single-qubit probe in $|+\\rangle$, let one amplitude-damping event $\\sigma_-$ occur during storage, run the one-auxiliary CNC circuit of Sec. IV.C, and measure the ancilla in the $X$ basis many times. If the ancilla outcome is uncorrelated with the damping event, the recovered QFI will fall below $4t_s^2$, falsifying the claimed preservation of Heisenberg-limited information.","tokens_in":10663,"feed_emoji":"⚛️","tokens_out":9906,"duration_ms":101689,"temperature":0.7,"pith_summary":"This paper tries to establish that the storage stage of a quantum sensing task—the waiting time after the unknown parameter has been encoded but before the probe is measured or reused—can be protected against noise without continuous error correction. It proposes a correlated-noise correction (CNC) protocol in which each memory error channel is coupled to an auxiliary qubit by a fixed two-body entangling gate; the auxiliary qubits coherently record whether errors occurred, and all syndromes are read out once at the end of storage. The paper argues that this preserves the quantum Fisher information of a probe that has already reached Heisenberg-limited encoding under dephasing, bit-flip, and amplitude-damping noise, and that the construction extends from one qubit to multi-qubit probes. It also argues that protecting the encoded Fisher information during storage is weaker than restoring the full probe state: partial correction suffices if the probe is measured immediately, while full state recovery is required if the probe will be reused. This matters because sensing is increasingly an intermediate step in larger quantum workflows, where probes must wait inside quantum memory before the next operation.","feed_headline":"Storage noise no longer costs Heisenberg-limited precision","feed_subtitle":"Ancilla qubits record memory errors so one final fix preserves the encoded probe.","key_machinery":"The machinery is the set of entangling gates $T_i$ conjugating each memory error $L_i$ into a Pauli on the auxiliary qubit times the probe error, via Eq. (10). Each gate is fixed and two-body, applied before storage and inverted after storage; because the conjugation condition depends only on the memory error operators and not on the sensing Hamiltonian, the correction strategy decouples from the encoding stage. The same construction is tensored for multiple noise channels and for multi-qubit probes, so the resource count is one auxiliary qubit per independent error channel. The protocol deliberately reads syndromes only once at the end, avoiding repeated measurements that would disturb the stored phase.","core_discovery":"The central claim is that storage-stage noise can be corrected with a protocol that never measures the probe and never touches it during memory. Before storage, the probe is entangled with auxiliary qubits through gates $T_i$ satisfying $T_i L_j = L_j T_i$ for $j \\neq i$ and $T_i^\\dagger L_i T_i = P^{(a_i)} \\otimes L_i$, so each independent memory error becomes a correlated probe–auxiliary error. After storage, inverse gates map the accumulated errors onto the auxiliary qubits, a single $X$-basis measurement reads all syndromes, and conditional Pauli corrections restore either the quantum Fisher information or the full probe state. For a $Z$-encoded single qubit, one auxiliary qubit with a control-$X$ gate detects dephasing; a second auxiliary qubit detects bit-flip; and the paper argues the same one-auxiliary construction protects against amplitude damping for immediate readout because only the $Z$-degrading component matters. For $N$-qubit GHZ probes, local dephasing is corrected with one auxiliary qubit per probe qubit, preserving the Heisenberg-scaled QFI $4N^2 t_s^2$, and repeated sensing-storage rounds maintain the ideal QFI when bit-flip errors are fully corrected between rounds.","pith_inferences":["A direct check implied by Eq. (10): for the control-$X$ gate in Eq. (13), the conjugation of a single amplitude-damping event $\\sigma_-$ does not obviously produce a Pauli operator on the ancilla, so whether the one-ancilla construction actually detects a decay event is a concrete experiment that would settle the amplitude-damping claim.","Since the conjugation condition is independent of the sensing Hamiltonian, the same gate family should protect phase encodings beyond $\\sigma_z$, such as collective or multiparameter encodings; the paper demonstrates only $Z$-encoding.","The partial-versus-full correction distinction suggests a resource-adaptive schedule: use the cheaper partial correction while the probe is only waiting, and apply full state recovery only immediately before the probe is reused in a later stage; the paper does not optimize this transition.","The numerics show QFI degrading with auxiliary dephasing rate $\\kappa_a$; an extension would identify the maximum ancilla noise rate below which CNC still outperforms uncorrected storage."],"forward_implications":["A probe at the Heisenberg limit can be stored for the full memory interval without loss: for a single-qubit $Z$-encoded probe the QFI remains $4t_s^2$, and for an $N$-qubit GHZ probe it remains $4N^2t_s^2$, provided the auxiliary qubits are effectively noiseless.","One syndrome extraction per storage interval suffices; continuous QEC during memory is unnecessary for preserving the Fisher information.","When the probe is measured immediately after storage, correcting only the parameter-degrading error component (for $Z$-encoding, the dephasing component) is enough to recover ideal precision even if bit-flip or amplitude-damping distortions remain in the state.","When the stored probe must be reused for further sensing or processing, full state recovery is required; the protocol supplies this by correcting all monitored error channels, enabling multi-round Heisenberg scaling.","The protocol extends to multi-qubit probes with one ancilla per independent local error channel, so overhead grows linearly with probe size for local noise."],"supporting_citations":[{"why":"Supplies the fidelity-based quantum Fisher information metric that the paper uses to evaluate all corrections.","marker":"[50]"},{"why":"Provides the quantum Fisher information and multiparameter estimation background behind the precision bound in Eq. (3).","marker":"[51]"},{"why":"States the no-go theorem that unitary control alone cannot restore the Heisenberg limit under noise, motivating the QEC approach.","marker":"[18]"},{"why":"Gives the HNLS condition for sensing-stage QEC that the storage-stage construction is designed to bypass.","marker":"[32]"},{"why":"Introduces quantum error correction for metrology, the framework the storage-stage protocol extends.","marker":"[28]"},{"why":"Shows improved quantum metrology through quantum error correction, establishing the Heisenberg-restoring baseline.","marker":"[29]"},{"why":"Relaxes the noise-signal orthogonality constraint in sensing-stage QEC, the most recent protocol that CNC complements by moving correction to storage.","marker":"[34]"}],"fun_headline_variants":["Ancilla qubits shield stored probes from memory noise","Heisenberg limit survives storage with correlated-noise correction","Storage noise corrected without touching the probe","One-shot syndrome fix preserves quantum Fisher information","Correlated-noise correction keeps metrological precision during memory"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The amplitude-damping result assumes that a single downward decay of the probe flips the auxiliary qubit in the same way a dephasing error does; if the decay instead leaves the auxiliary qubit untouched on the branch where the phase is erased, the one-ancilla protocol loses the parameter information silently.","fun_headline_variants_meta":{"raw":{"variants":["Ancilla qubits shield stored probes from memory noise","Heisenberg limit survives storage with correlated-noise correction","Storage noise corrected without touching the probe","One-shot syndrome fix preserves quantum Fisher information","Correlated-noise correction keeps metrological precision during memory"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000241,"raw_usage":{"total_tokens":1563,"prompt_tokens":1028,"completion_tokens":535,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":644,"completion_tokens_details":{"reasoning_tokens":464}},"tokens_in":644,"tokens_out":535,"duration_ms":6970,"temperature":1.0,"reasoning_tokens":464,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T00:23:38.696416+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Prepare a $Z$-encoded single-qubit probe in $|+\\rangle$, let one amplitude-damping event $\\sigma_-$ occur during storage, run the one-auxiliary CNC circuit of Sec. IV.C, and measure the ancilla in the $X$ basis many times. If the ancilla outcome is uncorrelated with the damping event, the recovered QFI will fall below $4t_s^2$, falsifying the claimed preservation of Heisenberg-limited information.","supporting_citations":[{"cited_title":"Bai and J.-H","cited_arxiv_id":null,"evidence_quote":"States the no-go theorem that unitary control alone cannot restore the Heisenberg limit under noise, motivating the QEC approach."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Introduces quantum error correction for metrology, the framework the storage-stage protocol extends."},{"cited_title":"D¨ ur, M","cited_arxiv_id":null,"evidence_quote":"Shows improved quantum metrology through quantum error correction, establishing the Heisenberg-restoring baseline."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Relaxes the noise-signal orthogonality constraint in sensing-stage QEC, the most recent protocol that CNC complements by moving correction to storage."}],"review_version":1}