{"id":"e36c13be-823b-4041-81dc-4b0838afd0c7","arxiv_id":"2608.10490","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Indefinite evolution maps two noncommuting signal parameters onto separate degrees of freedom, enabling simultaneous Heisenberg-limited multiparameter estimation.","lead":"A new protocol uses coherently superposed evolution paths, controlled by an auxiliary qubit, to separate two incompatible quantum signals into different subsystems. This allows both parameters to be estimated at Heisenberg-limited precision simultaneously, which is an open goal in multiparameter quantum metrology.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Unspecified compensation operations C± are the load-bearing gap: the natural parameter-independent choice fails from the second slice on, and no fixed controlled-unitary correction can satisfy the first two slice conditions; Eq. (7) is therefore not established.","rationale":"The reader's conditional verdict is based partly on the unspecified C±. My stress-test agrees and sharpens it: it is not merely that C± are unspecified; the natural parameter-independent implementation fails on the second slice, and a general first-order analysis shows no fixed controlled-unitary compensation can satisfy the first two slices. The only apparent ways out are parameter-dependent operations (circular) or a supplementary construction that is absent from the main text. This does not prove the entire IE idea false—some other protocol or measurement strategy might still achieve SHL—but it does mean the presented derivation of Eq. (7) is not valid as written. Since the reader already assigned CONDITIONAL, my verdict remains unchanged: the paper should be accepted only if the compensation operations are explicitly constructed and verified to be parameter-independent, or if the full protocol is otherwise shown to attain Eq. (7) without parameter knowledge.","tokens_in":10983,"tokens_out":39290,"duration_ms":340011,"concrete_test":"Set G1=X, G2=Z and assume C± are arbitrary β-independent controlled unitaries, possibly different at each slice. Model one slice as E=e^{-iβZ_aG2Δt}e^{-iαG1Δt} followed by C=|+><+|⊗U_+^{(k)}+|-><-|⊗U_-^{(k)}. Expand the condition that after one and two slices the output equals T=e^{-iβZ_aΔt}e^{-iαG1Δt} applied once and twice, to first order in Δt. The first-slice equations give U_+^{(1)}=I and U_-^{(1)}=G2; the second-slice equations then require U_-^{(2)}(G2+I)=2I on the probe, which has no unitary solution because I+G2 is singular. A numerical companion check: run N=10^4 slices with C+=I and C−=Z_p and verify that the |−> branch contains a residual −iβt(G2−I)|φ> at order βt, confirming the accumulation.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Equation (7), the product-state evolution that yields the diagonal Heisenberg-limited CFIM, is reached by 'history-dependent compensation operations C±' that the main text never defines. This is not just an exposition gap. For the orthogonal qubit case (G1=X, G2=Z), the first-slice condition (6) forces C+ = I and C− = G2 to first order in Δt. That choice makes the first slice correct, but on the second slice the input auxiliary state is e^{-iβZ_aΔt}|+> ≈ |+> - iβΔt |−>. The same compensation leaves a residual −iβΔt (G2−I)|φ> on the |−> branch, whereas the ideal T^2|+>|φ> requires −2iβΔt |φ>. Eliminating this residual with a parameter-independent controlled unitary would require U_−(I+G2)=2I on all probe states; since I+G2 is singular for G2^2=I, no unitary U_− exists. The error is O(βΔt) per slice and accumulates to O(βt), so it does not vanish as Δt→0. Thus either C± depend on α and β, making the protocol circular, or the Supplementary Material must contain a non-obvious parameter-independent construction. The full text provides neither, so the central claim is unsupported at its core.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a multiparameter quantum metrology protocol based on indefinite evolution (IE), in which two noncommuting signal generators are effectively separated by routing one signal component into an auxiliary qubit and leaving the other in the probe. The authors claim that for a single-qubit probe with orthogonal signal generators, IE achieves a diagonal classical Fisher information matrix with Heisenberg scaling for both parameters without signal reversal; for non-parallel generators the same scaling is achieved with a non-diagonal but invertible CFIM; and for parallel generators signal reversal is required. The protocol is extended to noisy qubit probes, multi-qubit probes, and high-dimensional probes via subspace projection, and a general theorem is stated for non-parallel projected generators. The central claim is that definite evolution cannot match this performance under compatible optimal measurements.","tokens_in":11240,"tokens_out":3201,"duration_ms":29874,"significance":"If the central mechanism is valid, the result is significant: it proposes an operational resolution of both encoding incompatibility and measurement incompatibility in multiparameter estimation, with a concrete resource (indefinite evolution) and with falsifiable CFIM predictions. The paper contains no fitted parameters, and the comparison with definite evolution is well posed. The multi-qubit and high-dimensional extensions are natural and, if supported, would broaden the impact. However, the core derivation of the separated evolution, Eq. (7), is not actually carried out in the main text, and the most obvious parameter-independent compensation fails from the second time slice onward; this makes the central claim currently unsupported. The value of the paper therefore depends entirely on whether the missing construction exists and is provided in a complete form.","major_comments":[{"comment":"The derivation of the central separated evolution (7) hinges on the 'history-dependent compensation operations C+ and C−', but these operations are never specified. This is not a mere exposition gap: for the orthogonal qubit case G1=X, G2=Z, the first-slice condition (6) forces C+ = I and C− = G2 to first order in Δt. With that choice, the second slice leaves a residual −iβΔt(G2−I)|φ> on the |−> branch, whereas the ideal evolution requires −2iβΔt|φ>. Eliminating this residual with a parameter-independent controlled unitary would require U_−(I+G2)=2I on all probe states, which is impossible because I+G2 is singular for G2^2=I. The error is O(βΔt) per slice and accumulates to O(βt), so it does not vanish in the continuous limit. Either the C± operations depend on α and β, which would make the protocol circular, or a non-obvious parameter-independent construction must be supplied in the Supplementary Material. As written, Eq. (7) is not established and the central claim is unsupported.","section":"Indefinite-evolution sensing protocol, Eqs. (5)-(7)"},{"comment":"For non-parallel generators H=αX+β(X sinθ+Z cosθ), the paper states that IE 'can still extract the Z component of the β signal into the auxiliary qubit' and then quotes F≃4t^2 [[1, sinθ],[sinθ,1]] in Eq. (9), but no explicit protocol, compensation operations, or measurement procedure is given for this case. Since Eq. (4) cannot be satisfied for 0<θ<π/2, the mechanism is not a minor modification of the orthogonal case, and the claimed F matrix needs a derivation. The effective CFI values F_α^eff=F_β^eff=4t^2 cos^2θ also require specifying how the non-diagonal CFIM is inverted in practice. This is a load-bearing gap for the claim that non-parallel generators achieve simultaneous Heisenberg scaling without signal reversal.","section":"Noiseless qubit probe, non-parallel case"},{"comment":"Theorem 1 is stated without proof. The statement that any pair of nonparallel traceless projected generators retains simultaneous HL scaling is central to the high-dimensional extension, and the proof is not a routine consequence of the two-level example in §6. The proof should be given in the main text or a clearly labeled appendix, including the construction of the effective IE protocol and the derivation of Eq. (15). Without it, the general claim is unsupported.","section":"High-dimensional probe, Theorem 1"},{"comment":"The noisy qubit section presents results only through figures and references to 'the Supplementary Materials'. The main text does not define the noise channels, the syndrome-extraction operations, the recovery operations, or the resulting CFIM expressions. In particular, the claims for the X- and Z-noise case with three auxiliary qubits are not verifiable from the manuscript. The authors should either include the full derivation in the main text or ensure that the Supplementary Material is self-contained and provided with the submission.","section":"Noisy qubit probe"}],"minor_comments":[{"comment":"There is a typo in the phrase 'under a compatible optimal measurement scheme', which reads 'udner' in the manuscript.","section":"Introduction"},{"comment":"The notation G_i is used for both the bare generator in H=Σ h_i G_i and the dressed generator defined in Eq. (2). These are different objects; please use a distinct symbol, e.g. \\bar{G}_i, for the dressed generator.","section":"Multiparameter quantum sensing, Eq. (2)"},{"comment":"The figure panels would benefit from labeled axes and a legend identifying FE, QEC, and IE; currently the reader must infer which curve corresponds to which strategy from the text.","section":"Figure 2"},{"comment":"Reference [54] contains a typo: 'high-effciency' should read 'high-efficiency'.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"The reliability of this paper rests entirely on the existence of the parameter-independent history-dependent compensation operations C±. The stress-test argument shows that the natural choice fails already at the second slice and that no fixed controlled-unitary correction can satisfy the first two slice conditions; this is exactly the kind of gap that a referee cannot fill. I would ask the authors to either give the explicit construction or, if none exists, substantially weaken the claims. The paper also states Theorem 1 and the noisy-case results without proof, and the Supplementary Material was not part of the reviewable text. I would not recommend acceptance until the missing construction is provided, because the current version's central claim is not established."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know. The central idea is genuinely new: instead of fighting multiparameter incompatibility with error correction or sequential schemes, the authors put signal reversal and control operations in a coherent superposition and route the two signals into separate degrees of freedom. That is a nice move, and for a single short-time slice it works. The problem is the continuous-time version. The paper never defines the 'history-dependent compensation operations' C+ and C−, and these are not a minor detail. The stress-test note shows that a parameter-independent choice can’t even correct the first slice properly, and the naive choice C+=I, C−=G2 accumulates an O(βt) error. I haven’t independently confirmed every line of that calculation, but the burden is on the authors to exhibit a construction that does not depend on the unknown α and β. Without it, Eq. (7) is an assumption, not a derivation.\n\nWhat the paper does well: it clearly separates encoding incompatibility from measurement incompatibility, gives a clean mechanism for orthogonal generators, includes numerical support, and is honest about the scope of its claims. The extension to multi-qubit probes via parallel local encoding is natural, and the subspace-projection route for high-dimensional probes is plausible. The comparison with definite evolution—sequential signal reversal erases one signal rather than transferring it—is correct. The citation pattern is normal, and there are no fitted parameters.\n\nSoft spots beyond the central gap: the non-parallel case and Theorem 1 are in an inaccessible supplement, so I can't check them. The noisy section is a sketch. These are secondary to the C± issue.\n\nWho is this for? Anyone working on indefinite causal order or multiparameter quantum metrology. The concept is worth discussing, but the current preprint shouldn't be cited as evidence that the simultaneous Heisenberg limit is achievable. I'd send it to peer review with a request for major revision. If the stress-test obstruction is fatal, the paper can be reworked into a proposal with the short-time result; if the authors find a parameter-independent compensation, this becomes an important result.","headline":"New and promising idea for multiparameter metrology, but the missing compensation operations C± are a load-bearing gap that leaves the central equation unproven.","tokens_in":11754,"tokens_out":11173,"would_cite":false,"duration_ms":94972,"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":"Coherently superposing different evolution channels can separate noncommuting signals into distinct subsystems, so one common measurement reaches the simultaneous Heisenberg limit for multiparameter estimation.","keywords":["indefinite evolution","multiparameter quantum metrology","Heisenberg limit","quantum Fisher information matrix","signal reversal","auxiliary qubit","subspace projection","quantum error correction"],"falsifier":"Build the two-slice IE circuit for $H=\\alpha X+\\beta Z$ with explicit gate definitions for $C_+$ and $C_-$ that contain no dependence on $\\alpha$ or $\\beta$, and measure the resulting classical Fisher information matrix; if the cross term $F_{\\alpha\\beta}$ fails to vanish or $F_\\beta$ does not scale as $4t^2$, the product-state separation of Eq. (7) does not hold.","tokens_in":10798,"feed_emoji":"⚛️","tokens_out":14234,"duration_ms":108081,"temperature":0.7,"pith_summary":"Multiparameter quantum sensing usually runs into two obstacles: noncommuting signal generators prevent optimal encoding for all parameters at once, and the optimal measurements for different parameters are mutually incompatible. The paper claims that indefinite evolution—coherently superposing different sensing channels via auxiliary control—removes both obstacles. For a qubit probe with orthogonal generators ($H=\\alpha X+\\beta Z$), the protocol maps the $\\beta$ signal into an auxiliary qubit and leaves $\\alpha$ in the probe; the two are then read out by compatible measurements, and the classical Fisher information matrix is diagonal with Heisenberg scaling ($\\sim 4t^2$) for each parameter. The same separation is achieved for non-parallel and parallel generators (the latter requiring signal reversal), and extended to noisy, many-qubit, and high-dimensional probes. If correct, indefinite evolution is an operational resource for attaining the simultaneous Heisenberg limit in multiparameter sensing.","feed_headline":"Superposed evolutions hit the simultaneous Heisenberg limit","feed_subtitle":"Routing each signal to its own subsystem lets one common measurement reach the precision limit for all parameters.","key_machinery":"The load-bearing object is the indefinite-evolution encoding block $\\mathcal{E}$: a controlled superposition of the sensing channel $U(\\alpha,\\beta)$ and a conjugated channel $D U(\\alpha,\\nu\\beta) D^\\dagger$, where $D$ is an auxiliary gate satisfying $[D,G_1]=0$ and $\\{D,G_2\\}=0$, and $\\nu=\\pm1$ selects whether the $\\beta$ signal is reversed. In the short-time limit, this block (followed by decoding) gives $e^{-i\\beta Z_a \\Delta t}|+\\rangle_a \\otimes e^{-i\\alpha G_1 \\Delta t}|\\varphi\\rangle_p$. The history-dependent compensation operations $C_+$ and $C_-$, applied in the $\\{|+\\rangle_a,|-\\rangle_a\\}$ basis between time slices, are what make the two branches accumulate the same effective auxiliary evolution, so that repeated slicing integrates to the product state of Eq. (7). That product state is the reason the CFIM is diagonal and simultaneously Heisenberg-limited.","core_discovery":"The central discovery is that indefinite evolution converts a multiparameter estimation problem into two effectively single-parameter problems by assigning each signal to its own subsystem. For $H=\\alpha X+\\beta Z$, a short-time encoding block superposes $U(\\alpha,\\beta)$ with $D U(\\alpha,\\beta) D^\\dagger$ for $D=X$; because $X$ commutes with the $\\alpha$ generator $X$ and anticommutes with the $\\beta$ generator $Z$, the $\\beta$ phase appears on the auxiliary control qubit as $e^{-i\\beta Z_a \\Delta t}$ while the probe accumulates $e^{-i\\alpha X \\Delta t}$. Repeating with history-dependent compensation $C_+,C_-$ yields, in the continuous limit, $e^{-i\\beta Z_a t}|+\\rangle_a \\otimes e^{-i\\alpha X t}|\\varphi\\rangle_p$. The parameters are thus separated into different degrees of freedom, the quantum and classical Fisher information matrices become diagonal, and the simultaneous Heisenberg limit is attained under compatible optimal measurements. The paper proves this for orthogonal generators without signal reversal, for non-parallel generators without signal reversal (with effective CFI $4t^2 \\cos^2\\theta$), and for parallel generators only when signal reversal is available, while definite evolution with temporal signal reversal reaches at most a single-parameter Heisenberg limit. A general theorem extends the result to any pair of nonparallel traceless projected generators in a two-dimensional encoding subspace.","pith_inferences":["A natural extension is to more than two parameters: assigning each additional signal to its own auxiliary qubit should, under the right commutation relations, yield a diagonal CFIM with simultaneous Heisenberg scaling; the paper does not address this case.","The short-time Trotter decomposition suggests a finite-slice tradeoff: shorter slices reduce the separation error but require more compensation rounds, and the optimal slice length under decoherence is not analyzed here.","The subspace-projection result hints that high-dimensional sensing can be reduced to qubit-like encoding whenever a two-level block diagonalizes the signal; testing this on random three-level generators would quantify how much leakage degrades the simultaneous limit.","In noisy settings, the protocol relies on syndrome extraction; combining indefinite evolution with other quantum error-correcting codes may extend simultaneous Heisenberg scaling to non-Markovian or correlated noise, but that is not demonstrated."],"forward_implications":["For a single-qubit probe with orthogonal generators, the simultaneous Heisenberg limit is attainable without signal reversal, recovering the full single-parameter scaling $4t^2$ for both parameters.","For parallel generators, signal reversal becomes a necessary resource for indefinite evolution to achieve simultaneous Heisenberg scaling; without it the CFIM is singular and the parameters are not identifiable.","Definite evolution, even with signal reversal, can only cancel one signal component via temporal reversal and thus remains at single-parameter Heisenberg scaling.","Under Markovian X-type or X-plus-Z noise, indefinite evolution with additional syndrome-extraction auxiliary qubits restores simultaneous Heisenberg scaling for both parameters.","For an n-qubit probe, local IE blocks achieve spatial Heisenberg scaling ($n^2$) for both parameters at once, and high-dimensional probes reach the same via subspace projection when the projected generators are nonparallel."],"supporting_citations":[{"why":"Defines the quantum Fisher information matrix and the multiparameter Cramér–Rao bound that set the precision benchmark.","marker":"[40]"},{"why":"Supplies the Holevo bound formula that characterizes the best precision attainable after optimizing over measurements.","marker":"[42]"},{"why":"Establishes the compatibility conditions that make multiparameter measurement incompatibility the problem IE addresses.","marker":"[43]"},{"why":"Provides multiparameter sensitivity bounds used to define the simultaneous Heisenberg limit as a target.","marker":"[44]"},{"why":"Introduces indefinite causal order as a metrological resource, the basis for the superposed-channel IE construction.","marker":"[71]"},{"why":"Reports an experimental demonstration of superposed gate order with super-Heisenberg precision, supporting the physical feasibility of IE superpositions.","marker":"[72]"},{"why":"Shows how quantum error correction achieves the Heisenberg limit, the definite-evolution baseline IE is compared against.","marker":"[26]"},{"why":"Demonstrates indefinite-causal-order error correction that restores Heisenberg scaling under noise, extended here to noisy multiparameter sensing.","marker":"[30]"}],"fun_headline_variants":["Indefinite evolution hits multiparameter Heisenberg limit","Superposed controls reach simultaneous Heisenberg precision","One measurement, all parameters: Heisenberg limit achieved","Quantum superposition breaks multiparameter metrology barrier","Indefinite evolution: key to simultaneous Heisenberg scaling"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The compensation operations $C_+$ and $C_-$ must be physically implementable without knowing the signal parameters $\\alpha$ and $\\beta$; the paper does not specify their gate-level construction.","fun_headline_variants_meta":{"raw":{"variants":["Indefinite evolution hits multiparameter Heisenberg limit","Superposed controls reach simultaneous Heisenberg precision","One measurement, all parameters: Heisenberg limit achieved","Quantum superposition breaks multiparameter metrology barrier","Indefinite evolution: key to simultaneous Heisenberg scaling"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000539,"raw_usage":{"total_tokens":2643,"prompt_tokens":1057,"completion_tokens":1586,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":673,"completion_tokens_details":{"reasoning_tokens":1511}},"tokens_in":673,"tokens_out":1586,"duration_ms":8782,"temperature":1.0,"reasoning_tokens":1511,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T14:20:41.700665+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Build the two-slice IE circuit for $H=\\alpha X+\\beta Z$ with explicit gate definitions for $C_+$ and $C_-$ that contain no dependence on $\\alpha$ or $\\beta$, and measure the resulting classical Fisher information matrix; if the cross term $F_{\\alpha\\beta}$ fails to vanish or $F_\\beta$ does not scale as $4t^2$, the product-state separation of Eq. (7) does not hold.","supporting_citations":[{"cited_title":"Suzuki, Explicit formula for the holevo bound for two-parameter qubit-state estimation problem, Journal of Mathematical Physics57(2016)","cited_arxiv_id":null,"evidence_quote":"Supplies the Holevo bound formula that characterizes the best precision attainable after optimizing over measurements."},{"cited_title":"Gessner, L","cited_arxiv_id":null,"evidence_quote":"Provides multiparameter sensitivity bounds used to define the simultaneous Heisenberg limit as a target."}],"review_version":1}