{"id":"b703d7cb-6cfc-41e8-8853-23f0cb9dc08c","arxiv_id":"2601.08137","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":8,"one_line_summary":"A trapped-ion experiment prepared low-energy states of a 19-spin Ising chain by engineered dissipation, backed by an exact finite-step Kraus form of the cooling channel.","lead":"The paper runs a 'cooling' protocol on a trapped-ion quantum computer: an auxiliary qubit repeatedly nudges a 19-spin chain toward its lowest-energy state instead of computing it directly. Even with thousands of imperfect two-qubit gates the measured energy stayed far below the random-state value, and noise mitigation brought it into line with noiseless simulation.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Load-bearing premise unmet: the implemented K̂ has positive-frequency support (finite β plus OFT truncation), so K̂|E0⟩≠0 and uniqueness for Γ_H∘Γ_K∘Γ_H is only cited, not shown; noiseless N=19 plateau above E0 confirms the guarantee does not apply.","rationale":"I read the paper in good faith: the Kraus derivation (Eqs. 7–11 and Appendix A) is correct and is a genuine extension beyond the Lindblad regime, and the fidelity monotonicity statement (Eq. 12) is a correct conditional theorem. The experimental data credibly show a steady low-energy state and ZNE improvement. However, the theorem's antecedent is not verified for the implemented channel. The paper itself acknowledges that truncation of the OFT produces positive-frequency support (Sec. III.C, Fig. 1(d)) and that even noiseless N=19 simulation does not converge to E0 (Sec. III.F, Fig. 5). This is not a negligible numerical detail: it means Eq. (5) fails, so the proof of monotone approach to |E0⟩ does not apply to the protocol as implemented. In addition, uniqueness of the fixed point of the discrete interleaved map is simply cited from Ref. [42] rather than established here. The experimental demonstration remains valuable as evidence of noise-robust convergence to a low-energy fixed point, but it does not establish ground-state preparation. Because the theoretical contribution stands and the experimental claims are already hedged in the body, a conditional verdict is appropriate rather than rejection; the reader already identified this as the weakest assumption, and the proposed numerical fixed-point test would settle whether the concern is quantitative or structural. Hence no change to the reader's verdict is needed.","tokens_in":23272,"tokens_out":15636,"duration_ms":153807,"concrete_test":"Numerically construct the full matrix of the implemented channel Γ_H∘Γ_K∘Γ_H for N=4 and N=6 exactly as in the experiments: K̂ from Eq. (23) with S_s=4π/(b-a), M_s=4, Δs=π/(b-a), Trotterized Ŵ, τ=4, Δt=0.25, N_t=4, and the finite-β Fermi filter. Diagonalize this channel. Compute (i) ‖K̂|E0⟩‖ with the actual filter parameters, (ii) the spectral gap of the channel, and (iii) all fixed points ρ_ss, then evaluate F(ρ_ss,|E0⟩) and E(ρ_ss). If the unique fixed point has F<1−10^{-2} or if there are multiple fixed points, the monotonicity guarantee does not apply to the implemented protocol and the claims must be narrowed to preparation of a low-energy fixed point. This is a noiseless numerical check requiring no quantum hardware.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central guarantee (Eq. 12) is conditional on |E0⟩ being the unique steady state of the implemented map Γ_H∘Γ_K∘Γ_H (Eq. 31). Uniqueness is inherited from Ref. [42] for the continuous-time ideal channel, but the implemented K̂ does not satisfy the sufficient condition K̂|E0⟩=0 (Eq. 5). With the finite-β Fermi filter (Eq. 29), f(ω)>0 for ω>0; at the gap β=8/Δ gives f(Δ)∼e^{-10}≈4.5×10^{-5}, while truncation and discretization of the OFT (Sec. III.C, Fig. 1(d)-(e)) broaden the filter edges by ∼π/S_s, producing larger positive-frequency support. Hence |E0⟩ is not a dark state of the implemented Γ_K, and the noiseless map has a shifted fixed point—directly visible as the N=19 noiseless plateau above E0 in Fig. 5. Moreover, uniqueness of the fixed point of the Trotterized, interleaved channel is never checked; if a dark subspace exists, Eq. (12) guarantees nothing about the limit. The experimental convergence to a low-energy state therefore demonstrates a low-energy fixed point of a different, noisy map, not ground-state preparation.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents a dissipative ground-state preparation protocol based on the single-ancilla construction of Ding et al. [42], implemented on the Quantinuum Reimei trapped-ion device for transverse-field Ising chains of up to 19 spins. The main theoretical contribution is an exact Kraus representation of the dissipation channel, Γ_K[ρ] = M_0 ρ M_0^† + M_1 ρ M_1^† with M_0 = cos(√(τ K†K)) and M_1 = −i√τ K sinc(√(τ K†K)) (Eqs. 7–10), valid for arbitrary step size τ. The paper further proves that the ground-state fidelity is monotonically non-decreasing under repeated application of Γ_K, provided the ground state is the unique steady state (Eq. 12). The experiments show that, despite hardware noise, the energy converges to a low-energy state far from the maximally mixed state even for circuits with up to 4110 entangling gates, and that ZNE improves agreement with noiseless simulations.","tokens_in":23521,"tokens_out":3094,"duration_ms":31766,"significance":"If the central claim holds, the paper provides a useful finite-step generalization of the Lindblad-based single-ancilla dissipation protocol and a NISQ-relevant experimental demonstration of noise-robust dissipative state preparation. The exact Kraus derivation (Appendix A) is algebraically sound and the completeness relation (Eq. 11) is correctly verified; the small-τ reduction to Lindblad dynamics is standard. The experimental data are presented carefully, with explicit gate counts, noise parameters, and ZNE procedures. However, the claim is conditional on a load-bearing assumption—unique steady state of the implemented channel—that is not established for the actual Trotterized, interleaved map. The paper itself acknowledges in Sec. III.C that the implemented filter violates Eq. (14), so the ground state is not an exact steady state of the implemented channel. Thus the theoretical guarantee does not directly apply to the experiment; the demonstrated result is preparation of a low-energy state, not exact ground-state preparation. This is a significant gap but one that can be addressed by either proving/quantifying the steady-state shift or by reframing the claims as approximate pr","major_comments":[{"comment":"The monotonicity theorem is stated for a CPTP map with a unique steady state. The map actually implemented in the experiment is the interleaved channel Γ_H ∘ Γ_K ∘ Γ_H of Eq. (31), not the ideal Γ_K. Uniqueness of the fixed point for this Trotterized, interleaved channel is not proved; it is inherited only by citation from Ref. [42], which concerns the continuous-time Lindblad generator. Since the whole ground-state preparation claim rests on this premise, the manuscript should either prove uniqueness for the implemented map under stated parameter conditions or explicitly list it as an unverified assumption with consequences.","section":"Sec. II.A, Eq. (12), Eq. (31)"},{"comment":"The implemented filter function has positive-frequency support: finite β rounds the Fermi-Dirac edges, and OFT truncation at S_s further broadens them by ∼π/S_s, as shown in Fig. 1(d)–(e). The paper itself states that 'A nonvanishing value of f̃(ω) at ω > 0 ... implies that the jump operator K̂ contains energy-increasing transitions, in which case the ground state is no longer the steady state.' Therefore the sufficient condition K̂|E0⟩=0 (Eq. 5) is violated, and the fixed point of Γ_K is not |E0⟩. The Appendix C bounds are operator-norm bounds on aliasing/truncation errors; they do not remove this positive-frequency support or bound the resulting shift of the steady state. This is load-bearing because Eq. (12) guarantees nothing about the limit when the uniqueness premise fails.","section":"Sec. III.C, Eq. (29), Eq. (14)"},{"comment":"The noiseless N=19 simulation levels off noticeably above the exact ground-state energy E0, confirming that the implemented noiseless map has a fixed point different from |E0⟩. The abstract's claim that ZNE brings energies into agreement with noiseless simulations is therefore not equivalent to ground-state preparation: it is agreement with a systematically shifted noiseless value. The manuscript should quantify the deviation between the noiseless fixed point and E0, or provide a rigorous error bound in terms of β, S_s, Δ_s, and the interleaving parameters, and adjust the wording of the central claim accordingly.","section":"Sec. III.F, Fig. 5"}],"minor_comments":[{"comment":"Typo: 'lager system sizes' should be 'larger system sizes'.","section":"Sec. IV"},{"comment":"Refs. [47] and [49] are the same work (Sivarajah et al., t|ket⟩); one duplicate should be removed.","section":"References"},{"comment":"The caption states 'Reimei-E for G=1 at m≥30 are also shown as orange open triangles,' but the main text says these are noisy simulations using Reimei-E with G=1 for m≥30. Please clarify whether these are emulator data and mark them consistently in the figure legend.","section":"Fig. 5 caption"},{"comment":"The choice M_s=4 is very small; the paper motivates it solely by resource constraints. A brief discussion of the resulting practical trade-off (circuit depth vs. systematic error) would help the reader judge the validity of the fixed point at larger m.","section":"Sec. III.C"},{"comment":"The statement 'Reference [42] shows that the ground state is the unique steady state of Γ_K under appropriate assumptions' should be more explicit about which assumptions are needed for the finite-τ Kraus map presented here, since the original proof may target the Lindblad generator.","section":"Sec. II.A"}],"recommendation":"major_revision","confidential_remarks":"The manuscript's strongest contribution is the exact finite-step Kraus representation and its careful experimental implementation; both are solid and worth publishing after revision. The main concern is that the central 'ground-state preparation' claim is not supported by the implemented channel's actual fixed point. The authors are transparent about the violation of Eq. (14) and the noiseless plateau, so this is a fixable rigor gap rather than a fatal error. I would not recommend rejection, but the claims need to be either weakened to 'low-energy state preparation with quantified systematic error' or supplemented by an analysis of the perturbed fixed point and its distance from the ground state."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The finite-τ Kraus representation (Eqs. 7–11) is the real new contribution. Ding et al. worked in the Lindblad limit; here the channel is written explicitly for arbitrary step size, and the algebra checks out—I verified the completeness relation and the small-τ reduction to Lindblad form. The fidelity-monotonicity argument is standard but correctly applied. The hardware result is also legitimately useful: on Reimei, energies stay far from the maximally mixed state even at 4110 entangling gates, and ZNE with the exponential fit brings the N=6 data into line with the noiseless simulation. That is a genuine demonstration of robustness, not a simulation.\n\nThe soft spot is not the math but the distance between the theorem and the implemented map. The monotonicity guarantee needs the ground state to be the unique steady state of the actual channel Γ_H∘Γ_K∘Γ_H. Uniqueness is cited from Ref. [42] for the ideal continuous-time map, but the implemented K has positive-frequency support because β is finite and the OFT is truncated and discretized (Fig. 1d). So K|E0⟩ ≠ 0, and the noiseless N=19 simulation indeed levels off above E0 (Fig. 5). The paper acknowledges this in Sec. III.C and III.F, but the abstract's “agreement with noiseless simulations within statistical uncertainties” concerns ZNE versus the noiseless simulation, not convergence to the true ground state. So “ground-state preparation” overstates what is demonstrated: it is preparation of a low-energy state that sits systematically above E0, with the offset growing with system size.\n\nSmaller issues: the hardware error bars are shot noise only; the ZNE fits use two or three noise points; and the filter parameters β, b, a are set using exact diagonalization of the target Hamiltonian, so the demonstration is tuned to the problem. No code or data artifacts are shipped, which is a real barrier for a hardware paper. None of these are fatal, and the paper is unusually candid about several of them.\n\nOverall: the theoretical result appears correct and worth having, and the experiment is a solid step toward dissipative preparation on NISQ hardware. I would send this to peer review. A referee should ask for a check of uniqueness for the Trotterized interleaved map (or at least a bound on the dark-space leakage), systematic error bars on the noiseless offset, and release of the circuits and data. Those are conditions, not grounds for rejection.","headline":"Solid theory, honest experiment, but the ground-state guarantee is conditional on uniqueness that is cited, not checked, and the noiseless benchmark itself misses E0.","tokens_in":24184,"tokens_out":1850,"would_cite":true,"duration_ms":19298,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["81P68","81S22"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper demonstrates that a dissipation channel with an exact Kraus form can prepare a spin-chain ground state on a trapped-ion quantum computer, tolerating thousands of noisy entangling gates.","keywords":["dissipative ground-state preparation","Kraus representation","Lindblad dynamics","trapped-ion quantum computer","transverse-field Ising model","zero-noise extrapolation","operator Fourier transform","monotonic fidelity"],"falsifier":"A decisive test is to compute the spectrum of the implemented interleaved channel Γ_H∘Γ_K∘Γ_H with the discretized filter and check whether any eigenoperator other than the ground-state projector has eigenvalue 1; if one does, the monotonic-fidelity premise fails. Experimentally, run the noiseless protocol for N=19 to much larger m and check whether the energy approaches the exact E0 as S_s and Δs are refined; the existing plateau above E0 would, if persistent, be evidence of additional steady states.","tokens_in":23015,"feed_emoji":"⚛️","tokens_out":9408,"duration_ms":85230,"temperature":0.7,"pith_summary":"The paper establishes that ground-state preparation via engineered dissipation works as a discrete, step-size-independent quantum channel and survives real hardware noise. The authors first write the dissipation map in an exact Kraus form valid for any time step, going beyond the small-step Lindblad limit; they then prove that fidelity with the ground state is non-decreasing under every application, so any initial state is driven to the ground state provided that state is the channel's unique steady state. On a trapped-ion quantum computer they prepare a transverse-field Ising chain of up to 19 spins, observing convergence to a low-energy state even with circuits containing 4110 entangling gates, and zero-noise extrapolation makes the measured energies agree with noiseless simulation. The point is that dissipative state preparation may be a practical route to ground states on noisy near-term hardware, using only one ancilla qubit and mid-circuit reset.","feed_headline":"Dissipation prepares a 19-spin ground state on a quantum device","feed_subtitle":"One ancilla qubit plus mid-circuit reset turns hardware noise into a tool for ground-state preparation.","key_machinery":"The workhorse is the jump operator K̂, built by Fourier filtering the time-evolved coupling Â(s)=e^{iHs} Â e^{-iHs} with a filter whose frequency support lies only on negative energy differences, so K̂ moves population downward in energy and annihilates the ground state. A single ancilla qubit couples to the system through a dilated operator K̂_dil = |1⟩⟨0|⊗K̂† + |0⟩⟨1|⊗K̂; the evolution exp(−iK̂_dil√τ), followed by discarding the ancilla, defines the channel Γ_K. The exact Kraus form uses the operator functions cos and sinc of √(τK†K), with completeness following from the operator identity cos² + sin² = I. In the limit τ→0 the channel reduces to Lindblad evolution with jump operator K̂, an","core_discovery":"The central discovery is that the dissipative channel Γ_K has an exact two-operator Kraus form for any step τ: Γ_K[ρ] = M0ρM0† + M1ρM1†, with M0 = cos√(τK†K) and M1 = −i√τK sinc√(τK†K). The completeness identity M0†M0 + M1†M1 = I makes it a completely positive, trace-preserving map, and when the jump operator K̂ annihilates the ground state, fidelity with that fixed point is non-decreasing under each application. This turns engineered dissipation into deterministic ground-state preparation. The hardware demonstration on a trapped-ion computer uses one ancilla qubit and mid-circuit reset; circuits with thousands of entangling gates still converge to low-energy states, and zero-noise extrapola","pith_inferences":["A natural next test is to benchmark convergence speed against the spectral gap of the implemented channel; measuring that gap directly could replace the paper's exact-diagonalization parameter choices.","The fidelity-monotonicity theorem does not imply monotone energy, as the paper itself notes in its comparison with imaginary-time evolution; designing jump operators that also make energy decrease monotonically is an open direction.","The observed noise robustness suggests the dissipative channel may actively suppress coherent gate errors rather than merely tolerating them; embedding Γ_K in a depolarizing error model could quantify when this protection breaks.","Because filter parameters can be estimated classically, the protocol could extend to models without exact solutions, such as frustrated or mixed-field spin chains, where ground-state preparation is currently harder."],"forward_implications":["Ground-state preparation no longer needs variational optimization or post-selection: repeated application of the dissipative channel increases fidelity from any initial state.","Because the Kraus form is exact for arbitrary step τ, large steps can be chosen to converge faster while remaining a legitimate quantum channel; only filter and Trotter approximations contribute error.","The protocol survives realistic noise: circuits with up to 4110 native two-qubit gates yield energies far below the maximally-mixed value, so engineered dissipation acts as a built-in error suppressor.","Zero-noise extrapolation corrects the residual noise: exponential ZNE makes the measured N=6 and N=19 energies agree with noiseless simulations within statistical uncertainties.","The resource footprint is small—one ancilla qubit, mid-circuit reset, and a Trotterized operator Fourier transform—so the method fits on current hardware."],"fun_headline_variants":["Dissipation cools 19 spins to ground state on ion trap","Exact Kraus map tames noise for quantum ground-state prep","Trapped-ion chain uses dissipation to reach low-energy state","Robust ground-state prep via engineered dissipation on qubits"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The whole guarantee rests on the ground state being the unique steady state of the channel actually implemented; the paper inherits that uniqueness from a prior analysis, and in practice the finite filter width plus truncation and discretization of the time integral give the filter a small positive-frequency component, so K̂ no longer exactly annihilates the ground state and even the noiseless simulation plateaus slightly above the exact ground-state energy.","fun_headline_variants_meta":{"raw":{"variants":["Dissipation cools 19 spins to ground state on ion trap","Exact Kraus map tames noise for quantum ground-state prep","Trapped-ion chain uses dissipation to reach low-energy state","Robust ground-state prep via engineered dissipation on qubits"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000164,"raw_usage":{"total_tokens":1100,"prompt_tokens":781,"completion_tokens":319,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":525,"completion_tokens_details":{"reasoning_tokens":249}},"tokens_in":525,"tokens_out":319,"duration_ms":3927,"temperature":1.0,"reasoning_tokens":249,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-03T10:55:03.476113+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A decisive test is to compute the spectrum of the implemented interleaved channel Γ_H∘Γ_K∘Γ_H with the discretized filter and check whether any eigenoperator other than the ground-state projector has eigenvalue 1; if one does, the monotonic-fidelity premise fails. Experimentally, run the noiseless protocol for N=19 to much larger m and check whether the energy approaches the exact E0 as S_s and Δs are refined; the existing plateau above E0 would, if persistent, be evidence of additional steady states.","supporting_citations":[],"review_version":1}