{"id":"6cf26fc1-6944-490b-b07f-ad8fecb1737f","arxiv_id":"2608.13521","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"Coupling one controllable qubit to a bosonic sensor gives a provable exponential reduction in the number of measurements needed to learn Fourier coefficients and temporal correlations of classical signals, with a 10^7-fold improvement demonstrated experimentally.","lead":"A single controllable qubit coupled to a conventional sensor can learn high-frequency Fourier components of a classical signal with exponentially fewer measurements than Gaussian-only sensors. The authors prove this, demonstrate a 10^7-fold reduction on a superconducting cavity-qubit device, and simulate gains for dark matter searches and wireless decoding.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 1's 'any conventional protocol' is proven only for Gaussian sensing; non-Gaussian receivers are not ruled out, so the advertised scope exceeds the proof.","rationale":"Read in good faith, the paper's formal hierarchy (Appendix C4) defines conventional quantum sensing as Gaussian sensing (Definition 18), and the QFS protocol, AFI framework, and experimental demonstrations are coherent and independently supportive. The concern is not internal inconsistency: within the Gaussian class the lower bound is plausible and Lemma 9 is a reasonable mechanism. The soft spot is that the main-text theorem and the abstract drop the qualifier 'Gaussian', so the advertised exponential advantage is broader than the proof. The reader's weakest assumption identifies this same scope gap; I partially agree. The check I propose would settle whether the gap is only in presentation (non-Gaussian receivers also fail, as I suspect from energy-vs-Fourier-support arguments) or substantive (some non-Gaussian receiver breaks the separation). Until that check or an added proof, keeping the verdict conditional is appropriate.","tokens_in":50962,"tokens_out":40924,"duration_ms":460871,"concrete_test":"Simulate the Fig. 2 binary task (kth x-Fourier amplitude 1.0, total probe energy E=k) for a non-Gaussian receiver: e.g., photon-number-resolving or parity measurement on a squeezed-vacuum or single-photon-subtracted probe, and the optimal Helstrom measurement between the two displaced-channel states. Compute the one-shot success probability (or the number of shots for 70% accuracy) for k=2,4,...,40. If it decays as exp(-Omega(k)), the Gaussian-only proof probably extends; if it decays only polynomially or saturates, the 'any conventional protocol' claim is overbroad.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The load-bearing gap is between the advertised and the proven scope of the central lower bound. Theorem 1 and the abstract state that 'any conventional protocol/sensor' needs exp(Omega(k)) queries, but the hardness proof in Appendix E2 (paired with Definition 18 and Lemma 9) is established only for Gaussian sensing protocols: Gaussian probe states, generaldyne measurements, and classical adaptivity. Lemma 9's bound C >= (8E+4)^{-1} A A^T is specifically a Gaussian-measurement statement; it does not by itself constrain non-Gaussian receivers such as photon-number-resolving or Helstrom measurements on non-Gaussian probes of the same O(k) energy. If such a receiver could estimate the kth x-quadrature Fourier coefficient in poly(k) queries, the central claim would fail as stated; if not, the paper still needs an explicit qualifier or an additional argument to back the plain-language 'any conventional' wording. This is a proof-scope concern, not an observed contradiction.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper claims that a single controllable qubit coupled to a conventional single-mode sensor yields exponential reductions in the number of signal queries needed to learn classical-signal features. Theorem 1 asserts an O(k)-query protocol with O(k) probe energy plus one ancilla qubit and one control operation for learning the kth Fourier coefficient of a displacement-channel distribution, against an exp(Omega(k)) lower bound for any conventional protocol of the same energy. Theorems 2 and 3 extend the hierarchy to Gaussian-versus-classical probes and to a single qubit of quantum memory for temporal correlations. The authors introduce the QPsi/AFI framework, report a superconducting cavity-qubit demonstration with claimed 10^7-fold query reduction, and present numerical simulations for axion dark-matter and wireless-communication applications.","tokens_in":51107,"tokens_out":8426,"duration_ms":90543,"significance":"If the theorem were established for the stated class, this would be a notable advance: an exponential learning advantage from one qubit and one control gate for elementary signal-learning tasks, backed by a concrete experimental platform and an explicit framework. Strengths include a clear energy-accounting convention, an open data/code repository, detailed experimental parameters, and a unified AFI formalism that produces matching upper and lower bounds. The main caveat is that the central lower bound is proved only for Gaussian sensing protocols, while the abstract and Theorem 1 advertise 'any conventional protocol.' The paper also leaves the proof of Theorem 1 to appendices that are absent from the version under review, which prevents full verification.","major_comments":[{"comment":"The central lower bound is proved for the Gaussian sensing class defined in Definition 18 (Gaussian probe states, generaldyne measurements, classical adaptivity), and the energy-Fisher constraint in Lemma 9 is specifically a Gaussian-measurement statement. The abstract and Theorem 1 state the hardness for 'any conventional protocol/sensor' without this qualifier. A non-Gaussian receiver, such as photon-number-resolving detection on a coherent or squeezed probe, is not included in the proof, so the advertised scope exceeds the proven one. Please either extend the lower bound to the larger class or consistently qualify the central claims as applying to Gaussian conventional protocols; the same qualifier should appear in the abstract, Theorem 1, and Figure 2(c).","section":"Theorem 1, Definition 18, Lemma 9"},{"comment":"The proof of Theorem 1 is deferred to paired sections E2 (lower bound) and F2 (upper bound) of the appendices, but these sections are not present in the manuscript version under review; only the roadmap and the definitions and lemmas in Appendix C appear. Consequently, the central exponential lower bound—in particular the Fourier-suppression argument that converts the Gaussian outcome lemma into an exp(Omega(k)) query bound for adaptive protocols—cannot be checked. The published version must include these proofs as complete theorems with all constants stated, or the version under review is not self-contained enough for acceptance.","section":"Appendix E2/F2"},{"comment":"The experimental demonstration of the 10^7-fold query reduction mixes directly measured data with a simulated, noiseless Gaussian baseline. In Figure 2(c) and Appendix B3, the QFS sample count at roughly 100 shots is experimental, but the conventional baseline is an idealized simulation, and the factor 10^7 is an extrapolated projection rather than a directly measured number of queries. Moreover, in the quantum-memory demonstration (Appendix B4, step 3), the qubit rotation angle is set using nu = 0.41 with the statement that the parameters 'were chosen to maximize the quantum advantage over the baseline'; a parameter chosen after seeing the comparison is not a fixed prediction. Please state clearly which numbers are measured, which are simulated, and how the free parameters were selected.","section":"Appendix B4, Figure 2(c)"}],"minor_comments":[{"comment":"The abstract renders '10^7-fold reductions' as '107-fold reductions'; this is a formatting error that should be corrected.","section":"Abstract"},{"comment":"The phrase 'any conventional Gaussian approach' in the figure caption already carries the Gaussian qualifier, while Theorem 1 and the abstract use 'any conventional protocol' without it; the terminology should be made consistent throughout.","section":"Section II, Figure 2(c)"},{"comment":"The equivalence C >= (8E+4)^-1 A A^T if and only if A^T C^+ A <= (8E+4) I is asserted without proof for general matrix A; please justify via the Schur complement or state it as a separate lemma.","section":"Appendix C2, Lemma 9"},{"comment":"The energy constraint in the definition of Gaussian sensing should explicitly state whether E includes all ancilla modes and the measurement dilation, since Lemma 9's proof uses the energy of the joint state.","section":"Appendix C4, Definition 18"},{"comment":"The text says QPsi extends beyond quantum Fisher information, but the central lower bound for Theorem 1, as presented, is derived via quantum Fisher information in Lemma 9; clarify that the AFI framework generalizes QFI-based bounds while the specific application still uses a Fisher-information step.","section":"Section III.A"}],"recommendation":"major_revision","confidential_remarks":"The manuscript relies heavily on companion papers from the same groups (QSL, QCDS) for core modeling tools and experimental primitives; if those papers are not published, the present manuscript's self-containedness is further reduced. The truncated appendices are also a serious concern: I would not send this to production without the full proofs of E2 and F2, and I would request a scope revision that aligns the theorem statements with the Gaussian class actually proved."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The core result is worth taking seriously: a single qubit plus one echoed conditional displacement can exponentially reduce the number of queries needed to learn a Fourier coefficient of a classical displacement signal, and the paper backs it with a coherent framework (QΨ/AFI) plus a superconducting-cavity demonstration of a 10^7-fold sample reduction. The Gaussian-outcome lemma, the Fisher-information energy bound, and the Fourier-suppression argument are clean and standard, and the authors are honest about several experimental limitations—they explicitly note that their memory demonstration does not operate in the regime where the memory advantage matters. Data and code are released, which is real evidence and should count in the paper's favor.\n\nThe soft spots are real but mostly addressable. The main lower-bound proofs live in Appendices E and F, which were truncated in the review copy, so I could not verify the central deductions. More importantly, the advertised scope of Theorem 1 exceeds what is proven: the exponential lower bound is established for Gaussian sensing protocols (Gaussian probes, generaldyne measurements, classical adaptivity), not for arbitrary 'conventional' receivers. The abstract's phrase 'conventional sensor' does not carry that qualifier, so a reader could reasonably think photon-number-resolving or other non-Gaussian receivers are also ruled out. That is a proof-scope mismatch, not an observed contradiction, but it should be fixed by either broadening the proof or narrowing the language. The experimental shot-count extrapolations also lack error bars, and the wireless and axion simulations depend on several free parameters and deliver only quadratic advantages in those applications, so the 'orders-of-magnitude' framing there should be read accordingly.\n\nWho is this for? Quantum sensing theorists, quantum learning theorists, and experimentalists working with cavity–qubit platforms. A serious referee should be assigned, but the referee needs the full appendices and should push the authors to qualify every 'conventional' statement to 'Gaussian' where that is what is proven. If the appendices check out, this is a solid paper with a new framework and a genuine experiment. My recommendation: engage with it, but require the scope fix before publication.","headline":"A strong central idea and a real experimental proof-of-principle, but the advertised 'conventional' lower bound is proven only for Gaussian protocols; worth refereeing carefully with full appendices.","tokens_in":51700,"tokens_out":2112,"would_cite":true,"duration_ms":24762,"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":"A single controllable qubit, coupled to an otherwise conventional sensor, exponentially reduces the number of measurements needed to learn classical signals.","keywords":["quantum sensing","exponential quantum advantage","sample complexity","Fourier coefficient estimation","single-qubit control","phase-space inference","temporal correlations","superconducting circuit"],"falsifier":"Take the binary discrimination task behind Figure 2 at equal energy and replace the qubit-assisted protocol with a squeezed probe followed by photon-number-resolving detection. If the sample complexity needed to reach $70\\%$ accuracy grows polynomially in $k$ rather than as $\\exp(\\Omega(k))$, the exponential-advantage claim fails outside the Gaussian class. Equivalently, compute the AFI of a photon-number-resolving receiver for the Fourier-coefficient observable: an AFI that is only polynomially small in $k$ would contradict the theorem's universality for conventional protocols.","tokens_in":50757,"feed_emoji":"📡","tokens_out":7641,"duration_ms":77960,"temperature":0.7,"pith_summary":"The paper sets out to show that one controllable qubit, added to an otherwise conventional quantum sensor, can exponentially reduce the number of measurements needed to learn properties of a classical signal. It establishes this for learning Fourier coefficients of a displacement distribution, for multi-time correlations of a time-varying signal, and for polynomial functionals of a signal. If true, near-term devices with a single coherent qubit and modest control would gain an exponential sample-complexity edge over equal-energy Gaussian sensing across tasks such as dark-matter-stream characterization, radar, and wireless decoding. The paper also reports a proof-of-principle superconducting experiment in which the qubit-assisted sensor learns Fourier amplitudes with roughly $10^7$ times fewer measurements than an idealized squeezed-homodyne baseline.","feed_headline":"One qubit turns sensors into exponential learners","feed_subtitle":"Coupling one control qubit to a sensor learns a kth Fourier feature in O(k) shots; Gaussian sensors need exponentially many.","key_machinery":"The carrying object is the accessible feature information (AFI), defined by Q$\\Psi$ as the squared $L^2(P_0)$ norm of the conditional expectation $\\mathbb{E}[h(X)|Y]$ of the target feature given the measurement outcome. A small AFI for an entire architecture yields a matching lower bound on the number of queries (the semiclassical measurement lemma), and the same quantity is saturated by an optimal estimator, so a ratio $\\exp(\\Omega(k))$ between the AFI of the qubit-assisted and Gaussian families certifies exponential advantage. Mechanically, the lower bound uses the fact that an energy-$E$ Gaussian experiment produces a Gaussian outcome whose covariance must satisfy $C \\succeq (8E+4)^{-1}AA^T$, so its response to the $k$th Fourier character is suppressed by $\\exp(-\\Omega(k^2/E))$; the qubit protocol imprints the character $\\chi_k$ directly through echoed conditional displacements, $ECD(\\beta)=D(\\beta/2)|g\\rangle\\langle e|+D(-\\beta/2)|e\\rangle\\langle g|$, whose response amplitude decays only polynomially in $k$.","core_discovery":"The paper's central claim is that a small amount of quantum coherence and control is a qualitatively new sensing resource. For a classical signal that acts on a bosonic probe as a random displacement $D(\\alpha)$, a protocol using probe energy $O(k)$, one ancilla qubit, and one control operation can learn the $k$th Fourier coefficient with $O(k)$ signal queries, while any conventional protocol in the class the paper formalizes (Gaussian probes with generaldyne measurements, plus adaptivity) requires $\\exp(\\Omega(k))$ queries. The same mechanism gives a single qubit of quantum memory an exponential advantage for estimating $m$-point temporal correlations, and gives ordinary Gaussian probes an exponential advantage over fully classical coherent-state probes for angular Fourier moments. The experiments, using a transmon qubit dispersively coupled to a microwave cavity, support the claimed scaling and show a factor of about $10^7$ in shot count at $k=20$ relative to an idealized squeezed-homodyne benchmark. All of these separations are derived from a single framework, Quantum Phase-Space Inference (Q$\\Psi$), whose central quantity, the accessible feature information (AFI), simultaneously certifies a lower bound for an experimental family and an optimal protocol that attains it.","pith_inferences":["The theorem-level lower bounds are proved within the formal class of Gaussian sensing protocols, so the plain-language phrase 'conventional sensor' should be read as 'Gaussian sensor' until non-Gaussian receivers are ruled out; the natural next test is a photon-number-resolving receiver at equal energy.","Because Q$\\Psi$ recasts experimental design as optimization over real-valued phase-space overlaps, the framework may lend itself to automated search for single-qubit protocols targeting other observables, such as higher-order spectra, rather than hand-built circuits.","Theorem 3's memory advantage suggests that stationary, separable sensors with one coherent ancilla could replace entangled multi-mode probes in noise spectroscopy, radar, and temporal-correlation tasks, an extrapolation to hardware the paper does not test.","If the single-qubit advantage persists under realistic photon loss at larger $k$, combining these protocols with error mitigation could open a route to practical quantum-enhanced sensing; the paper does not analyze this regime."],"forward_implications":["Learning the $k$th Fourier coefficient of a classical signal takes $O(k)$ queries with a single control qubit, where the paper proves Gaussian sensors require $\\exp(\\Omega(k))$ queries (Theorem 1).","Adding one qubit of quantum memory makes $m$-point temporal correlation estimation an $O(1)$-query task, while memoryless protocols with equal energy need $\\exp(\\Omega(m))$ queries (Theorem 3).","Gaussian quantum probes already give an exponential advantage over fully classical coherent-state probes: $O(1)$ versus $\\exp(\\Omega(k))$ queries for angular Fourier coefficients (Theorem 2).","In the superconducting experiment, discriminating two signals with $70\\%$ accuracy at $k=20$ takes about $10^2$ shots with the qubit-assisted protocol versus about $10^9$ for an idealized equal-energy Gaussian protocol, a seven-order-of-magnitude reduction.","In numerical simulations, a shallow qubit-based receiver identifies a 64-QAM symbol with about $100$ times fewer shots than a two-mode squeezed receiver and roughly $10^4$ times fewer than heterodyne detection at the same field strength."],"supporting_citations":[{"why":"Supplies the Quantum Signal Learning phase-space representation that formalizes classical signals as displacement channels with property kernels.","marker":"[55]"},{"why":"Provides the experimental platform and coherent echoed-conditional-displacement control used in the proof-of-principle demonstrations.","marker":"[67]"},{"why":"Gives the pulse-level realization of the echoed conditional displacement gate that imprints the target Fourier character on the qubit.","marker":"[69]"},{"why":"Documents limitations of quantum-Fisher-information metrology that motivate the Q$\\Psi$/AFI formulation.","marker":"[48]"},{"why":"Provides the axion-stream physical model and parameters used in the dark-matter application simulation.","marker":"[70]"},{"why":"Defines the two-mode squeezed-vacuum interferometer baseline against which the simulated wireless-receiver speedup is measured.","marker":"[44]"}],"fun_headline_variants":["One qubit exponentially shrinks measurement counts","Single qubit yields exponential sensing advantage","Exponential learning gain from a single qubit","One qubit cuts measurements exponentially"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The exponential lower bounds are proven only for Gaussian sensing protocols (states and measurements built from coherent states, squeezing, and homodyne or heterodyne readouts); if 'conventional sensor' includes non-Gaussian receivers such as photon-number-resolving detectors, the advertised scope is not established by the paper's proofs.","fun_headline_variants_meta":{"raw":{"variants":["One qubit exponentially shrinks measurement counts","Single qubit yields exponential sensing advantage","Exponential learning gain from a single qubit","One qubit cuts measurements exponentially"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000385,"raw_usage":{"total_tokens":2075,"prompt_tokens":1024,"completion_tokens":1051,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":640,"completion_tokens_details":{"reasoning_tokens":998}},"tokens_in":640,"tokens_out":1051,"duration_ms":9654,"temperature":1.0,"reasoning_tokens":998,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T05:11:12.301775+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take the binary discrimination task behind Figure 2 at equal energy and replace the qubit-assisted protocol with a squeezed probe followed by photon-number-resolving detection. If the sample complexity needed to reach $70\\%$ accuracy grows polynomially in $k$ rather than as $\\exp(\\Omega(k))$, the exponential-advantage claim fails outside the Gaussian class. Equivalently, compute the AFI of a photon-number-resolving receiver for the Fourier-coefficient observable: an AFI that is only polynomially small in $k$ would contradict the theorem's universality for conventional protocols.","supporting_citations":[{"cited_title":"Demkowicz-Dobrzański, J","cited_arxiv_id":null,"evidence_quote":"Supplies the Quantum Signal Learning phase-space representation that formalizes classical signals as displacement channels with property kernels."},{"cited_title":"Malnou, D","cited_arxiv_id":null,"evidence_quote":"Provides the experimental platform and coherent echoed-conditional-displacement control used in the proof-of-principle demonstrations."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Defines the two-mode squeezed-vacuum interferometer baseline against which the simulated wireless-receiver speedup is measured."}],"review_version":1}