{"id":"c52ed00e-c69e-4001-8dd3-58482a16651c","arxiv_id":"2411.10744","paper_version":1,"verdict":"UNVERDICTED","confidence":"HIGH","novelty_score":2.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A perspective proposing the term 'digital-analog quantum machine learning' (DAQML) and arguing, from a selective review, that combining analog blocks with digital gates may enable NISQ-era quantum machine learning advantages.","lead":"This paper is a short perspective that reviews recent work combining digital-analog quantum computing with machine learning, and proposes the name 'DAQML' for the area. It argues that this hybrid approach may let noisy intermediate-scale quantum devices perform useful machine learning tasks without full fault tolerance.","discovery_kind":"review","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The central claim of fruitfulness/advantage for DAQML rests on extrapolation from subroutine-level resource gains and from analog simulation experiments, without evidence connecting these to end-to-end QML advantage.","rationale":"The reader's verdict of UNVERDICTED is appropriate because the manuscript is a Perspective with no new derivations, experiments, or code. I agree with the reader that the central claim depends on an unsupported assumption, but I locate the weak point more specifically: the paper jumps from resource gains in isolated subroutines, such as the digital-analog QFT, and from successful digital-analog quantum simulation experiments, to an expectation of advantage in full QML pipelines. The decisive missing piece is not only hardware fidelity, as the reader emphasizes, but also the absence of any end-to-end complexity or benchmark evidence that analog blocks reduce the dominant cost of a learning task or evade classical simulability. A concrete resource benchmark on a specific cited proposal would settle whether the extrapolation has quantitative support. Since the paper is a perspective and its claim is explicitly hedged, the correct disposition remains UNVERDICTED; my concern does not change the reader's verdict but sharpens the reason for withholding validation.","tokens_in":3597,"tokens_out":3530,"duration_ms":41141,"concrete_test":"Select one cited DAQML proposal, e.g., the digital-analog QCNN of Ref. [7] or the digital-analog VQE of Ref. [3], and benchmark it end-to-end against (a) the fully digital compiled version and (b) a classical baseline such as a tensor-network or kernel method, at fixed problem size and target accuracy. For each version, count total two-qubit-gate depth, total single-qubit gates, and estimated success probability under a hardware-specific noise model. If the digital-analog version does not reduce the cost metric that dominates NISQ execution (typically two-qubit-gate depth or accumulated error), or if the classical baseline matches accuracy in polynomial resources, then the Section 3 assertion that these works 'show evidence' of fruitfulness and advantage fails for that instance.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central assertion in Section 3, that DAQML 'can be fruitful' and 'may provide a quantum advantage', is supported only by citing first-instance proposals [3,4,6–10] and by analogy to digital-analog quantum simulation experiments in Section 4. The load-bearing step is the transfer of resource reductions demonstrated for a subroutine (digital-analog QFT in Ref. [8], HHL co-design in Ref. [9]) or for simulation platforms to the full QML pipeline. That transfer is asserted, not argued: none of the cited works is reported as outperforming a classical ML baseline or as providing an end-to-end resource count that includes data loading, training, measurement overhead, and noise accumulation in long analog blocks. The perspective also does not address known obstacles to near-term QML advantage, such as barren plateaus or classical simulability of circuits built from global, low-depth native Hamiltonians. Without a concrete mechanism by which analog blocks reduce the dominant cost of a learning task, the fruitfulness claim remains an extrapolation. The paper's own 'perhaps' and 'may' language signals this, but a perspective still needs at least one quantitative anchor for its central claim.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This manuscript is a Perspective that proposes the name \"Digital-Analog Quantum Machine Learning\" (DAQML) for the combination of digital-analog quantum protocols with quantum machine learning algorithms. It reviews a small set of recent theoretical and experimental works (Refs. [3,4,6-10]) that employ large analog blocks of native Hamiltonian evolution together with single-qubit digital gates, and it argues that this combination may be fruitful and may enable better scalability of QML algorithms on NISQ devices. The paper also briefly surveys digital-analog quantum simulation experiments in trapped ions, superconducting circuits, and Rydberg atom arrays, and speculates that DAQML could become one of the first approaches to impact industry.","tokens_in":3801,"tokens_out":2514,"duration_ms":35664,"significance":"If the promise of DAQML materializes, the direction could provide a practical route toward near-term quantum advantage in machine learning by leveraging the scalability of analog simulation and the flexibility of digital gates. The paper is clearly written and serves as a concise entry point to a young subfield, and it performs the useful service of naming the area and collecting the relevant first-instance references. The author is appropriately cautious, using hedged language such as 'perhaps' and 'may'. However, the central claim of fruitfulness and potential advantage is not quantitatively anchored, and the evidence cited is mostly from first-instance proposals and simulation experiments rather than from demonstrations that DAQML outperforms classical baselines or reduces end-to-end resources in a learning task. The manuscript would be significantly strengthened by a more critical discussion of the conditions under which analog blocks could actually reduce the dominant costs of QML.","major_comments":[{"comment":"The central assertion that previous works 'show evidence that this combination can be fruitful' is supported only by citing first-instance proposals (Refs. [3,4,6,7,10]) and subroutine-level resource gains (Refs. [8,9]). None of these cited works is reported to outperform a classical machine learning baseline, and none provides an end-to-end resource count for a learning task that includes data loading, training, measurement overhead, and noise accumulation. Since this assertion is the main claim of the Perspective, it needs at least one quantitative anchor or a clearly stated condition under which DAQML would provably reduce the cost of a QML pipeline.","section":"Section 3, final paragraph"},{"comment":"The resource-gain evidence is drawn from the digital-analog quantum Fourier transform and from a co-design of the HHL algorithm, which are subroutines, not complete QML algorithms. The transfer of subroutine-level savings to the full QML setting is asserted rather than argued. For example, a faster QFT does not by itself imply faster training of a variational quantum classifier, because the training cost is dominated by repeated circuit evaluations and classical optimization. This is a load-bearing gap in the 'better scalability' claim.","section":"Section 3, Refs. [8] and [9]"},{"comment":"The extrapolation from digital-analog quantum simulation experiments to quantum machine learning tasks ignores known obstacles to near-term QML advantage. The manuscript does not discuss barren plateaus in variational circuits, the possibility that circuits built from global low-depth native Hamiltonians are classically simulable, or the accumulation of errors in long analog blocks. Without addressing these issues, the Outlook's vision that DAQML 'may represent a significant step forward' remains an unsupported extrapolation. Adding a paragraph that either connects to known no-go results or explains why analog blocks circumvent them would substantially strengthen the Perspective.","section":"Sections 4 and 5"}],"minor_comments":[{"comment":"The word 'accesible' should be 'accessible' in the first paragraph of the abstract.","section":"Abstract"},{"comment":"The figure contains a typographical artifact 'R1,,2' and the labels are not fully legible; please provide a cleaner version with consistent subscript formatting.","section":"Figure 1"},{"comment":"The phrase 'which is significantly impacting society at large, by enabling a plethora of possibilities, as well as challenges' is a run-on sentence; consider splitting it for clarity.","section":"Section 1"},{"comment":"The term 'extreme digital-analog quantum protocol (EDAQP)' is introduced in the text and figure but is not defined elsewhere; a one-sentence explicit definition in the main text would help readers.","section":"Section 2"},{"comment":"The sentence 'This could, in turn, impact subsequently society at large, as well as the scientific enterprise as well' contains a redundant 'as well'; please revise.","section":"Section 5"}],"recommendation":"major_revision","confidential_remarks":"The author's heavy reliance on their own prior work and that of close collaborators (Refs. [2,4,8,9]) is noticeable, though not by itself disqualifying for a Perspective. The editor may wish to consider whether the manuscript's scope fits the journal; it is a short perspective with no original technical results, and the main claim could be strengthened by an external viewpoint or a more balanced citation of possible limitations. If the author can add the requested quantitative or analytical discussion, the revised version could be acceptable as a Perspective."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this is a perspective, not a research paper. The only genuinely new element is the label 'DAQML' and a compact diagram of an 'extreme' digital-analog protocol. As a review it is readable and the descriptions of the cited papers look faithful. I have no reason to doubt the characterizations of Refs. [3]–[10].\n\nWhat it does well: it makes a fair, low-key case that a hybrid digital-analog mode of execution is worth testing for QML, and it points to concrete first instances. The author is appropriately hedged with 'may' and 'perhaps' in several places, and doesn't try to dress the perspective up as a derivation.\n\nWhere it is soft: the central claim in Section 3 is supported by extrapolation. Resource gains in a QFT subroutine or an HHL co-design are not evidence that an end-to-end QML training loop will beat a classical baseline. There is no account of data loading, measurement overhead, noise accumulation in long analog blocks, barren plateaus, or classical simulability. The stress-test note is right on this. Also, the citation base is heavily self-referential: most of the technical content traces to the author's group or close collaborators, and the Outlook refers to 'the paradigm that we created more than 10 years ago'. That is fine as history but not as independent grounding. The final paragraph about industry and society is boosterish and doesn't belong in the analytic part.\n\nNone of this makes the paper incoherent. It is a reasonable short perspective for a venue that wants a quick overview of a niche direction. But it is not a serious refereeable research contribution. If the journal publishes perspectives, it could be accepted after the central promise is tempered and a limitations paragraph is added. If it arrived as a regular research paper, I would desk reject.\n\nI would not spend referee time on it; I also wouldn't cite it as a technical source. It is a mild pointer to other people's work.","headline":"A coherent but thin perspective whose central promise is extrapolated from subroutine resource gains and self-cited prior work; fine as a taxonomy, not as evidence of quantum machine learning advantage.","tokens_in":4325,"tokens_out":3302,"would_cite":false,"duration_ms":33092,"reading_group":"no","serious_thinker":"yes","would_accept_peer_review":false},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"A Perspective argues that mixing large analog evolution blocks with small digital gates can make quantum machine learning scale on near-term NISQ devices, and proposes the name DAQML for this research direction.","keywords":["quantum machine learning","digital-analog quantum protocols","NISQ devices","quantum advantage","variational quantum algorithms","neutral-atom arrays","quantum Fourier transform"],"falsifier":"A head-to-head experiment on a single platform—running a DAQML variational circuit and a purely digital variational circuit with the same expressibility on the same learning task—would falsify the central claim if the digital-analog version shows no reduction in circuit depth, gate count, or measured error.","tokens_in":3381,"feed_emoji":"⚛️","tokens_out":7859,"duration_ms":70811,"temperature":0.7,"pith_summary":"This Perspective argues that a hybrid quantum computing recipe—large global analog evolution blocks stitched together by small single-qubit digital gates—can make machine learning calculations more scalable on today's noisy intermediate-scale quantum devices. The author proposes the name \"digital-analog quantum machine learning\" (DAQML) for this research avenue and reviews recent first instances: variational eigensolvers for molecules, a genetic algorithm, a quantum approximate optimization algorithm, quantum kernels for image classification, a digital-analog quantum Fourier transform, and neutral-atom learning algorithms. The combined evidence, in the author's reading, supports the idea that DAQML is fruitful and that better scalability of quantum machine learning may eventually yield a quantum advantage over classical machine learning. A reader should care because it offers a concrete path to useful quantum computation before fault-tolerant machines exist.","feed_headline":"Digital-analog quantum protocols could give machine learning an edge","feed_subtitle":"Combining big analog blocks with tiny digital gates may let quantum ML scale on today's hardware.","key_machinery":"The central object is the extreme digital-analog quantum protocol (EDAQP): a circuit made of global native-interaction blocks $U_k(t_k)$, each a unitary applied to all qubits by the platform's own Hamiltonian, separated by tiny single-qubit rotations $R_{i,j}$. This construction carries the argument because it concentrates the expensive work in large analog pieces that are natural to the hardware, while restricting the digital part to the operations a platform performs best. The result is a reduction in circuit depth and number of discrete gates, which is what the paper identifies as the route to scalability on NISQ devices.","core_discovery":"The central claim is that the digital-analog quantum paradigm, previously developed for quantum simulation, carries over to machine learning and is already showing promise in proof-of-principle works. On the paper's terms, the extreme digital-analog quantum protocol (EDAQP) is the generic template: the largest possible analog blocks—global unitary gates generated by the platform's native Hamiltonian—are interleaved with the smallest possible digital operations, single-qubit gates, which have the best fidelities on most platforms. The reviewed works are taken as evidence that this combination is fruitful, and the paper's outlook states that combining large analog blocks with digital steps 'may represent a significant step forward' in quantum technologies, possibly the first approach to impact industry.","pith_inferences":["My inference: if DAQML matures, the hardware's native Hamiltonian effectively becomes a trainable layer, so choosing a platform becomes part of the learning model rather than a fixed constraint.","My inference: the resource gains claimed for individual primitives such as the quantum Fourier transform only become an end-to-end advantage if the analog blocks can be calibrated and controlled as reliably as digital gates; this is a testable assumption that the paper leaves open.","My inference: the same digital-analog trick might apply to quantum generative models, where global analog evolution could act as a fast mixer or sampler between digital readout and feedforward steps.","My inference: naming the area DAQML is a community-forming proposal, and its success will depend on whether independent groups adopt the term and build benchmarks around it."],"forward_implications":["If DAQML is right, QML algorithms such as variational eigensolvers and quantum kernels can be implemented on NISQ hardware with fewer total gates than fully digital decompositions, because the global analog blocks replace sequences of two-qubit gates.","The digital-analog quantum Fourier transform, with its reported resource gains over the purely digital version, can be slotted into larger QML pipelines, cheapening a widely used primitive.","Neutral-atom arrays, with strong native interactions and recently demonstrated scalability, become a natural platform for DAQML and could be where a favourable machine-learning task first shows a practical quantum advantage.","Near-term industrial adoption may arrive through the co-design of algorithms with the native Hamiltonians of specific quantum platforms, before fault-tolerant error correction is available."],"supporting_citations":[{"why":"Foundational reference defining the digital-analog quantum simulation paradigm that the paper extends to machine learning.","marker":"[2]"},{"why":"First proposal of a digital-analog variational quantum eigensolver on neutral-atom arrays, a key instance of DAQML.","marker":"[3]"},{"why":"Extends the digital-analog approach to a quantum genetic algorithm for molecular ground-state energies in a scalable way.","marker":"[4]"},{"why":"Applies digital-analog interactions to approximate the quantum approximate optimization algorithm, a task relevant to machine learning.","marker":"[6]"},{"why":"Implements digital-analog quantum convolutional neural networks for image classification, applying the paradigm to a core QML task.","marker":"[7]"},{"why":"Shows possible resource gains in implementing the quantum Fourier transform with digital-analog protocols, a primitive that underlies QML algorithms.","marker":"[8]"},{"why":"Co-designs a linear-systems quantum algorithm in digital-analog form, with implications for QML pipelines involving linear algebra.","marker":"[9]"},{"why":"Proposes digital-analog quantum learning algorithms for neutral atoms, combining near-term QML utility with efficient scalability.","marker":"[10]"}],"fun_headline_variants":["Digital-analog blend could scale quantum machine learning","Quantum ML benefits from analog-digital mix","Digital-analog quantum ML: a path to near-term advantage","Analog blocks plus digital gates: recipe for practical quantum ML","Quantum ML without fault tolerance via digital-analog design"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole promise depends on the untested premise that NISQ devices with tens to hundreds of qubits can run digital-analog protocols with high enough fidelity and low enough overhead to beat classical machine learning, which the cited demonstrations do not yet establish.","fun_headline_variants_meta":{"raw":{"variants":["Digital-analog blend could scale quantum machine learning","Quantum ML benefits from analog-digital mix","Digital-analog quantum ML: a path to near-term advantage","Analog blocks plus digital gates: recipe for practical quantum ML","Quantum ML without fault tolerance via digital-analog design"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000712,"raw_usage":{"total_tokens":3160,"prompt_tokens":855,"completion_tokens":2305,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":471,"completion_tokens_details":{"reasoning_tokens":2229}},"tokens_in":471,"tokens_out":2305,"duration_ms":16498,"temperature":1.0,"reasoning_tokens":2229,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T19:20:15.458104+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A head-to-head experiment on a single platform—running a DAQML variational circuit and a purely digital variational circuit with the same expressibility on the same learning task—would falsify the central claim if the digital-analog version shows no reduction in circuit depth, gate count, or measured error.","supporting_citations":[],"review_version":1}