{"id":"83186b34-3d93-4f3e-9fa9-3f0eb0f9b6e6","arxiv_id":"2507.19153","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"Adaptively segmented pulse-based VQE prepares ground states of small Heisenberg and mixed-field Ising chains in simulated Rydberg arrays, and a hybrid gate scheme is proposed for measuring the energy.","lead":"This paper shows, in simulation, that pulse-level variational optimization can prepare ground states of Heisenberg and mixed-field Ising spin chains in Rydberg atom arrays with up to ten qubits. It tests whether analog pulse control, rather than digital gates, can run useful near-term quantum simulations.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Heisenberg N=6,10 results hinge on an undemonstrated q=pi initial-state preparation; Appendix B's GHZ circuit exceeds the ~6 µs coherence time.","rationale":"After reading the paper and the reader's verdict, I find the same load-bearing concern: the feasibility of the q=pi initial state for odd N/2 Heisenberg rings. The numerical PVQE results themselves are credible—they match exact diagonalization and the algorithm is standard—but the abstract's claim 'up to ten qubits' includes N=6 and N=10 Heisenberg cases that require an initial state outside the natural product-state manifold. The paper's own Appendix B admits the naive GHZ circuit is limited by coherence, and the alternative string unitary is only a plausibility argument. This is not a mathematical contradiction; it is a gap between the numerics and the experimental claim. The concern is load-bearing because without a feasible q=pi preparation, the central claim overstates what has been demonstrated. I concur with the reader's CONDITIONAL verdict: the paper should be accepted for its numerical evidence, but the claims should be narrowed or supplemented with a concrete preparation protocol. I did not find a more fundamental issue; the fixed-radius selection in Fig. 9 is a secondary weakness, but the q=pi preparation is the primary barrier to the abstract's headline.","tokens_in":44438,"tokens_out":5421,"duration_ms":52448,"concrete_test":"Simulate the Appendix B GHZ-based q=pi preparation in Pulser using the parameters of the paper (n=70, Jnn/h~2.3 MHz, coherence ~6 µs), including both the linear-depth and logarithmic-depth circuits, with CNOT gate times taken from Ref. [35]. Record the total wall-clock time and the fidelity of the prepared state relative to Eq. (15). If the time exceeds ~6 µs or the fidelity is below ~0.99, the N=6 and N=10 Heisenberg PVQE results are not hardware-feasible, and the abstract's 'up to ten qubits' claim must be narrowed to even N/2 cases or accompanied by a demonstrated preparation protocol.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central claim includes accurate preparation of the antiferromagnetic Heisenberg ground state for N=10. For rings with odd N/2 (N=6,10), the ground state lies in the momentum q=pi sector, while the all-down initial state is q=0. Because the PVQE Hamiltonian in Eq. (3) is invariant under the translation operator, the q quantum number is conserved, so no pulse sequence starting from the product state can reach the target. The paper therefore assumes the q=pi state of Eq. (15) can be prepared. Appendix B proposes two routes: a GHZ circuit whose CNOT gates each take ~6.2 µs at Jnn/h ~2.3 MHz (from Ref. [35]), and a many-body string unitary U_flip(pi/2) that is stated to be realizable 'in principle' without a demonstrated pulse sequence. The full GHZ circuit for N=6 requires at least five sequential CNOTs (~31 µs), far beyond the ~6 µs coherence time cited in Sec. VI; the logarithmic-depth version is not quantified. The string unitary acts with a Y1 X2 ... XN term, which is not a native interaction of the Rydberg Hamiltonian Eq. (1), and no construction is given. The paper itself acknowledges that the GHZ implementation 'may be limited by the short coherence time' (Appendix B). Thus the Heisenberg results at N=6 and N=10, and hence the 'up to ten qubits' claim, rest on an unverified experimental capability. This is the load-bearing weak point: a reader cannot tell from the manuscript whether the demonstrated numerics correspond to any physically realizable experiment.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a pulse-based variational quantum eigensolver (PVQE) for Rydberg-atom arrays, in which the global Rabi frequency and detuning are linearly ramped pulses that are adaptively refined by random time segmentation, and the array radius is treated as an additional variational parameter. The authors test this ansatz on the one-dimensional antiferromagnetic Heisenberg model (N=4, 6, 8, 10) and the mixed-field Ising model (N=10), reporting relative energy errors and spin correlation functions against exact diagonalization. They also introduce a hybrid measurement scheme in which global rotations synthesized by variational quantum gates are used to measure the XX, YY, and ZZ components of the Heisenberg Hamiltonian. The central claim is that ground states of these models can be accurately prepared for systems of up to ten qubits.","tokens_in":44709,"tokens_out":6041,"duration_ms":60526,"significance":"If fully supported, the work would provide a hardware-realistic, pulse-level variational method for preparing small many-body ground states on Rydberg platforms. A notable strength is that all numerical results are benchmarked against independent exact diagonalization, and the optimized states are shown to reproduce spin correlations, not only energies. The linearly varying, clock-constrained pulse model is also more experimentally credible than piecewise-constant ansätze, and the proposed variational-rotation measurement scheme addresses a practical bottleneck for evaluating generic Hamiltonians. However, the support for the headline claim is incomplete: the Heisenberg results for N=6 and N=10 rely on a q=pi initial state whose preparation within the quoted coherence time is not demonstrated, and for the larger systems the reported high accuracies are often achieved only by the best of many optimization runs rather than by typical runs.","major_comments":[{"comment":"The Heisenberg results for N=6 and N=10 depend on the unverified assumption that the q=pi state of Eq. (15) can be prepared on hardware. Because the PVQE Hamiltonian in Eq. (1) commutes with the translation operator, the momentum quantum number is conserved, and the all-down product state (q=0) cannot reach the ground-state sector with q=pi; the paper therefore initializes with |Psi_{q=pi}>. Appendix B proposes a GHZ circuit (Eq. B4) whose CNOT gates each require approximately 6.2 microseconds at Jnn/h~2.3 MHz [35], so the sequential N=6 implementation is roughly 31 microseconds, far exceeding the ~6 microsecond coherence time cited in Sec. VI; the logarithmic-depth variant is not quantified. The alternative U_flip(pi/2) in Eq. (B5) is a non-native N-body Pauli string for which no pulse sequence is given. The authors themselves concede that the GHZ implementation may be limited by the short coherence time (Appendix B). Without a demonstrated or quantified preparation protocol within coherence time, the abstract's \"up to ten qubits\" claim for the Heisenberg model is not supported.","section":"Sec. IV, Eq. (15), and Appendix B"},{"comment":"The reported high accuracy for larger systems is based on the best-performing runs rather than on typical behavior. For N=8, the text states that over 90 of the 100 PVQE runs converge to states with rather higher errors, and the ensemble-averaged final relative error is 2.14 +/- 0.12%, while only the best run reaches about 0.1%. For N=10 with the q=pi initial state, the best result among 92 runs is on the order of 0.1%, with an ensemble average of 1.65 +/- 0.17%. The manuscript does not report success probabilities, medians, or the full error distribution. If PVQE is claimed to accurately prepare ground states, the typical success rate must be quantified; as written, the central claim overstates the reliability of the method.","section":"Sec. IV, Fig. 3(a) and Fig. 4(b)"},{"comment":"The fixed-radius hybrid scheme intended for measurement of the Heisenberg Hamiltonian also shows poor typical convergence: the Fig. 9 caption states that only 10 representative runs out of 100 achieved good convergence. No success probability or ensemble statistics are reported for this configuration. Since the hybrid protocol is presented as experimentally feasible, the same typical-performance concern applies here as in Fig. 3, and the manuscript should report how often the full T*=3.3 microsecond protocol actually prepares the ground state to the claimed accuracy.","section":"Sec. VI, Fig. 9"}],"minor_comments":[{"comment":"There is a typographical error in the Conclusions: \"antiferromangnetic\" should be \"antiferromagnetic.\"","section":"Sec. VII"},{"comment":"The text says the optimization achieves high accuracy with \"as few as nine time segments\" and then states that the highlighted run uses \"only three pulse segments\"; this is confusing and should be reworded to distinguish the best run from the threshold-crossing run.","section":"Sec. IV, Fig. 2(a)"},{"comment":"References [27] and [34] refer to the same paper (Sherbert et al., Phys. Rev. Appl. 23, 024036); if both citations are kept, this duplication should be noted or the numbering adjusted.","section":"References [27] and [34]"},{"comment":"The logarithmic-depth GHZ circuit is referenced but not described; a few sentences or an explicit gate-count estimate for N=6 and N=10 would allow the reader to check whether that optimized version fits within the quoted coherence time.","section":"Appendix B, Fig. 11"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is honest about the coherence-time limitation in Appendix B, but the abstract overstates the result. The q=pi initial-state preparation is the more fundamental correctness risk for the Heisenberg N=6 and N=10 claims. Additionally, the typical-performance statistics (N=8: about 90% of runs fail to reach high accuracy; fixed-R: only 10/100 runs achieve good convergence) suggest the method is not yet robust enough to support the headline claim without further reporting or qualification. A revision that either demonstrates the q=pi preparation, restricts the claims, and reports success probabilities would be appropriate."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Read it. The core is a clean numerical demonstration that adaptive segmentation of linear-ramp Rydberg pulses can approximate ground states of small Heisenberg and mixed-field Ising chains, plus a sensible hybrid scheme for measuring non-native observables via variational quantum gates. The numerics are honest: energy errors and spin correlations are checked against exact diagonalization, and the paper is upfront about the momentum-sector obstruction and the short coherence time.\n\nWhat's actually new: the linear-ramp implementation of the ctrl-VQE segmentation idea is more hardware-realistic than piecewise-constant pulses; treating the array radius as a variational parameter is a nice twist; the analysis of the q=pi obstruction is correct; and the fixed-radius hybrid measurement proposal with global rotations is a practical addition. The MFI results are credible though somewhat expected, since the target Hamiltonian is close to the Rydberg ansatz Hamiltonian—the authors say this themselves.\n\nThe soft spots are real. The abstract says 'up to ten qubits' but for Heisenberg with odd N/2 (N=6,10), the ground state lies in the q=pi sector, and the all-down initial state can't reach it. The paper assumes a q=pi initial state, and Appendix B's preparation routes don't close the deal: the GHZ circuit takes roughly 31 microseconds for N=6 at their quoted gate time, against a ~6 microsecond coherence time, and the string unitary is 'in principle' with no pulse sequence. This is load-bearing for the ten-qubit Heisenberg claim. The manuscript acknowledges the difficulty but doesn't downgrade the abstract accordingly. Also, the N=8 ensemble average is 2.14% with only the best runs showcased, and the fixed-radius hybrid figure shows 10 selected runs out of 100. Those statistics should be reported honestly. The imported gate fidelities from Ref. [35] aren't re-checked, which is a minor point.\n\nNone of this is a mathematical error. The central algorithm demonstration for N=4, N=6/10 from the q=pi state, and the MFI results hold up as numerics. The paper should be sent to referees, with the clear instruction to push on the q=pi preparation: either demonstrate a feasible protocol, or rewrite the abstract and conclusions to present the odd-N/2 Heisenberg results as conditional on a separate capability. The measurement scheme also deserves scrutiny on whether the variational gate approach actually preserves accuracy at scale.\n\nWho reads this? People working on pulse-level variational algorithms for Rydberg arrays and near-term analog simulators. It's a useful contribution, not a breakthrough. I'd cite the linear-ramp segmentation and the hybrid measurement protocol if I were working in that area. Send it to peer review with requests for honest statistics and narrowed claims.","headline":"A genuine numerical study of adaptive linear-ramp PVQE with a useful hybrid measurement scheme, but the Heisenberg N=6,10 claims lean on an undemonstrated q=pi initial state; referee should require the claim to be narrowed.","tokens_in":45287,"tokens_out":2671,"would_cite":true,"duration_ms":27311,"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":"Randomly segmented laser pulses can prepare ground states of Rydberg spin models with up to ten qubits.","keywords":["pulse-based variational quantum eigensolver","Rydberg atoms","optical tweezer arrays","Heisenberg model","mixed-field Ising model","analog quantum simulation","adaptive pulse segmentation","quantum many-body state preparation"],"falsifier":"Measure the fidelity of the $q=\\pi$ state produced by the GHZ-type circuit or by a pulse realizing $U_{\\mathrm{flip}}(\\pi/2)$ on a Rydberg array; if the achieved overlap is too low to keep the reported energy errors, or if the preparation takes longer than the coherence time, the odd-$N/2$ Heisenberg results do not transfer to hardware.","tokens_in":44187,"feed_emoji":"⚛️","tokens_out":12007,"duration_ms":104952,"temperature":0.7,"pith_summary":"This paper argues that a variational quantum eigensolver can run entirely at the pulse level on Rydberg atoms in optical tweezer arrays, without decomposing the computation into digital gates. The algorithm represents the candidate state as the output of a time-dependent laser pulse and improves that state by randomly splitting the pulse into more segments and re-optimizing the Rabi frequency and detuning at each new boundary. Numerical simulations show this prepares accurate ground states for the one-dimensional antiferromagnetic Heisenberg model and the mixed-field Ising model with up to ten qubits, and that the optimized states reproduce the models' spin correlations. A hybrid measurement scheme, in which digital gates are approximated by optimized analog pulses, keeps the entire protocol within the hardware coherence time. If correct, this offers a practical route to small many-body ground-state preparation and energy measurement on current neutral-atom devices.","feed_headline":"Pulse shaping alone prepares spin ground states on Rydberg atoms","feed_subtitle":"A variational algorithm that splits laser pulses reproduces Heisenberg and Ising ground states on up to ten qubits.","key_machinery":"The key machinery is the adaptive random time-splitting of a single global pulse. The variational state is $|\\Psi[\\Omega,\\Delta,R]\\rangle = \\mathcal{T}\\exp\\left(-\\frac{i}{\\hbar}\\int_0^T dt\\,\\hat H(t)\\right)|\\psi_0\\rangle$, where $\\hat H(t)$ is the Rydberg Hamiltonian with time-dependent Rabi frequency $\\Omega(t)$, detuning $\\Delta(t)$, and a circular geometry of radius $R$. Starting from a one-segment linear ramp, each iteration randomly splits one time interval, interpolates the pulse values at the new boundary, and re-optimizes all boundary values to minimize the expectation value of the target Hamiltonian. This turns a small initial parameter set into a progressively richer variational ansatz while avoiding piecewise-constant discontinuities; a minimum segment duration and a hardware clock period keep the schedules physically realizable. A second component is the hybrid measurement scheme, in which the global $\\pi/2$ rotations needed to measure $\\hat X$ and $\\hat Y$ correlations are themselves synthesized from optimized analog pulses.","core_discovery":"The central discovery is that the ground states of two prototypical spin models can be prepared by globally optimizing a continuous, piecewise-linear pulse of laser detuning and Rabi frequency, with the atom array radius as the only geometrical parameter. For Heisenberg rings with an even number of spin pairs, starting from the all-down product state, the random time-splitting procedure converges to the target ground state: the best $N=4$ run reaches a relative energy error below $1\\%$ with three segments, while $N=8$ needs about fifty segments and reaches roughly $0.1\\%$. For $N=6$ and $N=10$, where the ground state lives in the momentum $q=\\pi$ sector, the all-down state is trapped in the $q=0$ sector; the authors show that initializing in a $q=\\pi$ superposition repairs convergence, with $N=6$ errors around $0.005\\%$ and $N=10$ errors around $1.7\\%$. The same algorithm prepares the mixed-field Ising ground state for $N=10$ with average errors below $0.02\\%$ across the tested transverse-field range. The optimized states match exact diagonalization for both energy and spin correlation functions, including the SU(2)-symmetric correlations of the Heisenberg model, even though the pulse ansatz does not enforce that symmetry.","pith_inferences":["Editorial inference: because the ansatz is a generic time-ordered unitary, the same random-segmentation optimizer could be pointed at excited-state or time-evolution cost functions, not just ground states.","Editorial inference: the $q=\\pi$ preparation bottleneck suggests a direct experimental benchmark: synthesize the string unitary $U_{\\mathrm{flip}}(\\pi/2)$ as an analog pulse; success would remove the main obstacle for odd-$N/2$ rings.","Editorial inference: the observed restoration of SU(2) symmetry in the Heisenberg correlations may be a finite-size effect of the optimizer; enforcing the symmetry explicitly in the ansatz or measuring the symmetry violation as $N$ grows would test how the method scales."],"forward_implications":["If the numerical results carry over to hardware, Rydberg tweezer arrays can prepare and measure the ground-state energy of the one-dimensional Heisenberg and mixed-field Ising models at up to ten qubits using only global pulses and one geometric parameter.","The optimized variational states reproduce spin correlations, so the protocol can be used to extract correlation functions of the target model, not just energies.","Because the optimized pulses resemble quasi-adiabatic schedules for Ising-type ground states, the method offers a way to initialize analog Rydberg simulators for subsequent quench dynamics.","The hybrid rotation scheme keeps the total experimental duration around $3.3\\,\\mu\\mathrm{s}$, below the roughly $6\\,\\mu\\mathrm{s}$ coherence time quoted for current arrays, making the measurement step feasible now.","For odd-$N/2$ Heisenberg rings, the results depend on an initial $q=\\pi$ state, so the proposed GHZ-type or string-unitary preparation is an integral part of the claim rather than a peripheral detail."],"supporting_citations":[{"why":"It supplies the gate-free ctrl-VQE idea that motivates starting from a pulse-level variational state and optimizing pulse parameters directly.","marker":"[25]"},{"why":"It establishes the digital-analog VQE blueprint on Rydberg arrays and the measurement toolbox that this protocol extends.","marker":"[26]"},{"why":"It provides the variational quantum gate synthesis whose optimized analog pulses implement the global rotations used in the hybrid measurement scheme.","marker":"[35]"},{"why":"It provides the virtual device-level simulation platform and the hardware parameter bounds, including pulse amplitudes, clock period, and minimum segment duration, used in all numerical experiments.","marker":"[36]"},{"why":"It identifies the momentum-sector obstruction ($q=\\pi$ versus $q=0$) for Heisenberg rings and motivates a symmetry-adapted initial state.","marker":"[40]"},{"why":"It establishes the sign rule for antiferromagnetic Heisenberg ground states that places the odd-$N/2$ ground state in the $q=\\pi$ sector, explaining why the all-down product state fails.","marker":"[47]"},{"why":"It provides the experimental context of optical tweezer arrays with measured coherence times and positional disorder that set the feasibility bounds for the protocol.","marker":"[10]"}],"fun_headline_variants":["Pulse splits prepare Heisenberg and Ising ground states","Adaptive pulse splitting nails spin ground states on Rydberg arrays","Ten qubits, one laser pulse: ground states solved","Pulse-based VQE reaches Heisenberg and Ising ground states","Analog pulse algorithm prepares spin ground states on ten qubits"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the $q=\\pi$ starting state required for odd-size Heisenberg rings can be prepared with high fidelity inside the roughly six-microsecond coherence time of current Rydberg arrays.","fun_headline_variants_meta":{"raw":{"variants":["Pulse splits prepare Heisenberg and Ising ground states","Adaptive pulse splitting nails spin ground states on Rydberg arrays","Ten qubits, one laser pulse: ground states solved","Pulse-based VQE reaches Heisenberg and Ising ground states","Analog pulse algorithm prepares spin ground states on ten qubits"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.0006,"raw_usage":{"total_tokens":2800,"prompt_tokens":940,"completion_tokens":1860,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":556,"completion_tokens_details":{"reasoning_tokens":1775}},"tokens_in":556,"tokens_out":1860,"duration_ms":13939,"temperature":1.0,"reasoning_tokens":1775,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T17:59:36.913570+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the fidelity of the $q=\\pi$ state produced by the GHZ-type circuit or by a pulse realizing $U_{\\mathrm{flip}}(\\pi/2)$ on a Rydberg array; if the achieved overlap is too low to keep the reported energy errors, or if the preparation takes longer than the coherence time, the odd-$N/2$ Heisenberg results do not transfer to hardware.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"It supplies the gate-free ctrl-VQE idea that motivates starting from a pulse-level variational state and optimizing pulse parameters directly."},{"cited_title":"Michel, S","cited_arxiv_id":null,"evidence_quote":"It establishes the digital-analog VQE blueprint on Rydberg arrays and the measurement toolbox that this protocol extends."},{"cited_title":"Chevallier, J","cited_arxiv_id":null,"evidence_quote":"It provides the variational quantum gate synthesis whose optimized analog pulses implement the global rotations used in the hybrid measurement scheme."},{"cited_title":"Silv´ erio, S","cited_arxiv_id":null,"evidence_quote":"It provides the virtual device-level simulation platform and the hardware parameter bounds, including pulse amplitudes, clock period, and minimum segment duration, used in all numerical experiments."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"It identifies the momentum-sector obstruction ($q=\\pi$ versus $q=0$) for Heisenberg rings and motivates a symmetry-adapted initial state."},{"cited_title":"Marshall, Antiferromagnetism, Proceedings of the Royal Society of London","cited_arxiv_id":null,"evidence_quote":"It establishes the sign rule for antiferromagnetic Heisenberg ground states that places the odd-$N/2$ ground state in the $q=\\pi$ sector, explaining why the all-down product state fails."},{"cited_title":"Scholl, M","cited_arxiv_id":null,"evidence_quote":"It provides the experimental context of optical tweezer arrays with measured coherence times and positional disorder that set the feasibility bounds for the protocol."}],"review_version":2}