{"id":"1f8f1707-179d-40df-a82a-2b79dbbfea1c","arxiv_id":"2412.13711","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"Amplitude damping noise, channeled through an ancilla-based encoding, reproduces the fermionic bath dynamics of impurity models with an order-of-magnitude qubit reduction.","lead":"This paper shows that qubit decay, normally a nuisance, can be used as the dissipation that impurity-model simulations need. If it works, quantum simulations of strongly correlated materials would need far fewer qubits and no costly ground-state preparation.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The method's central claim depends on T1-limited qubits whose noise is essentially pure amplitude damping; the numerical demonstration omits pure dephasing, which would corrupt the waiting-time dissipation and the claimed error scaling.","rationale":"The reader's weakest assumption is the same one that, in my reading, bears the most weight: the method rests on a noise model that is not currently typical of superconducting hardware. I agree with the CONDITIONAL verdict: the mathematical construction (fermionic QRT, ancilla reduction, noise encoding, PM fitting) is argued carefully, and the RLM emulation supports the claim as stated for an amplitude-damping-only device. However, the numerical validation does not include dephasing or other error channels, and the scaling analysis in App. H explicitly assumes no pure dephasing. A dephasing channel would not only add a constant error floor; because Twait grows with T1, the dephasing error during waiting time can grow or persist as T1 is increased, changing the central scaling advantage. This is a testable, quantitative issue rather than a philosophical objection. I do not see circular reasoning or fabrication; the paper is transparent about its limitations. The appropriate verdict remains CONDITIONAL, pending a realistic noise-model check and, ideally, a hardware demonstration.","tokens_in":24888,"tokens_out":17585,"duration_ms":159492,"concrete_test":"Repeat the Section IV emulation with the same circuit and parameters (Nb=8, Nanc=1, τ=0.3, T1=10^5) but add a pure-dephasing channel (jump operator Z/√Tφ) on every noisy qubit, for Tφ = T1 and Tφ = 10 T1. Recompute the greater-Green's-function error (Eq. B9) at the optimal τ. If the error increases by more than ~2× relative to the amplitude-damping-only case, the hardware assumption is load-bearing; if the result is insensitive for Tφ ≥ 10 T1, the concern is mitigated.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The most load-bearing condition for the central claim is the assumed noise model: the noisy bath qubits must undergo amplitude damping as their dominant, controllable decoherence channel, and the impurity/ancilla qubits must be effectively noiseless (Sec. IV, App. H). The waiting-time calibration makes the total damping probability per Trotter step equal Λτ, and the error analysis takes the gate-time noise as the only T1-dependent error. If pure dephasing is present, the waiting intervals---which scale as Twait ≈ T1 Λτ - TTrotter---accumulate an additional dephasing error Twait/Tφ that cannot be removed by recalibrating Twait. In the encoded frame, dephasing maps to E† S_z E, which is not one of the target fermionic jump operators, so it directly contaminates the simulated Green's function. The numerical evidence in Fig. 5 includes only amplitude damping, so it does not quantify this contamination. The paper acknowledges that standard architectures have significant pure dephasing and only suggests mitigation, but no mitigation is demonstrated. Consequently, the claimed qubit-count and long-time advantages are not yet established for currently available hardware.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a quantum algorithm for impurity-model dynamics that uses qubit amplitude damping as a resource. The bath is first represented by dissipative pseudomodes, whose Lindblad jump operators are then mapped onto qubit amplitude damping through ancilla fermionic modes and a 'noise encoding' unitary. Waiting times are used to tune the effective dissipation rate. The authors argue that this approach reduces qubit count, reaches longer times than closed-bath representations, and avoids ground-state preparation because the open system relaxes to its steady state. The central demonstration is a numerical simulation of the resonant level model with Nb=8 noisy bath qubits and one ancilla, compared with exact and pseudomode dynamics.","tokens_in":25090,"tokens_out":10481,"duration_ms":98691,"significance":"If the proposal works as stated, it is a timely and conceptually interesting inversion of the usual treatment of hardware noise: instead of mitigating T1 decay, the algorithm converts it into the dissipative bath of a DMFT impurity model. The paper contains several careful supporting pieces: the fermionic quantum regression derivation in App. A2, the dilation argument justifying Markovian bath replacement in Apps. A3-A4, and the ancilla-invariance proof in App. C, all of which appear technically sound. The convex fitting of the hybridization function is also a solid, reproducible element. The main significance is conditional, however, on a hardware regime in which amplitude damping is the dominant decoherence channel on the bath qubits and the impurity/ancilla qubits are effectively noiseless, a regime that the paper acknowledges but does not demonstrate or quantify.","major_comments":[{"comment":"The waiting-time calibration is stated in two incompatible ways. The main text requires (TTrotter + Twait)/T1 = Λτ, with Λmin = TTrotter/(T1τ), while App. H derives Twait = T1Λτ, which would give a total damping probability of Λτ + TTrotter/T1 per Trotter step. This distinction matters because App. H treats the TTrotter/T1 contribution as a separate noise error in Eq. (H15), whereas the main-text calibration appears to absorb it into the target dissipation rate. The numerical demonstration in Sec. IV does not state which convention was used. Please reconcile the two formulas and specify whether gate-time amplitude damping is counted as desired dissipation or as error; if it is error, the calibration in Sec. III.B should be Twait/T1 = Λτ, and if it is dissipation, Eq. (H15) overcounts the gate-time contribution.","section":"Sec. III.B and App. H, Eq. (H11)"},{"comment":"The central numerical evidence and all error scalings assume that the noisy bath qubits suffer only amplitude damping, with no pure dephasing, and that impurity and ancilla qubits are noiseless. The paper acknowledges in Sec. I and Sec. V that standard architectures have significant pure dephasing and suggests mitigation, but no mitigation is demonstrated or quantified. This is load-bearing because the waiting intervals are long, Twait ≈ T1Λτ - TTrotter, and a dephasing channel with time Tφ adds an error proportional to Twait/Tφ that cannot be removed by recalibrating Twait; in the encoded frame the dephasing operator is not one of the target fermionic jump operators. Please add a quantitative robustness analysis, for example a noise model with finite Tφ, or an explicit error-mitigation protocol, and state clearly that the claimed resource advantages hold only in the amplitude-damping-dominated regime.","section":"Sec. IV and App. H"},{"comment":"The asymptotic claims in Sec. III.C and Table I rely on assumed error scalings: ε_Trotter = ατt, ε_noise = βT/T1, and especially ε_fit = γ/Nb. The first two are standard heuristic forms, but ε_fit = γ/Nb is introduced without derivation or numerical verification, and the constants α, β, γ are not estimated. The fixed-(Nb,τ) linear-in-t runtime and gate-count advantage is more robust, but the optimized t^{2/3} error scaling and the corresponding resource table should be presented as heuristic rather than established results unless the scalings are checked numerically for the models studied.","section":"App. H, Eqs. (H2)-(H3), (H16), Table I"}],"minor_comments":[{"comment":"The text says T1 = 10^5 in units of the single-qubit gate duration and T2qb = 10, but it is unclear whether T1 applies to both single- and two-qubit gate durations and whether the waiting-time calibration uses the same T1; please clarify.","section":"Sec. IV"},{"comment":"The axis labels in Fig. 4(c) are not fully specified; please state the units of the number of two-qubit gates per Trotter step and the definition of K on the horizontal axis.","section":"Fig. 4(c)"},{"comment":"The sentence 'This derivation extends naturally to time-dependent Lindbladians' at the end of App. A2 is not obvious; a brief justification or reference would help the reader assess the scope of the fermionic quantum regression theorem used in the main text.","section":"App. A2"},{"comment":"There are several typographical errors, including 'Specificaly' in App. F, 'operaotrs' in App. A4, and 'Kramers-Kroenig' in App. B1; these should be corrected in a final revision.","section":"Throughout"}],"recommendation":"major_revision","confidential_remarks":"I see no concern about novelty or citation practice. The paper is honest about its idealized noise assumption, but that assumption is central to the claimed hardware advantage, so the requested robustness analysis should be a condition of acceptance rather than left as future work."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Here's my read on arXiv:2412.13711. This is a genuine step beyond Leppäkangas et al.: it extends the noise-harvesting idea to fermionic impurity models, which is not trivial. The authors give a concrete circuit construction—fermionic pseudomode representation, ancilla-assisted mapping to local jump operators, a noise-encoding unitary that turns c and c† into S−, and a Trotter-with-waiting protocol. The appendix derivations (fermionic quantum regression, bath-switching via dilation, ancilla-invariance proof) are careful and check out as far as the text shows. The resource analysis in App. H is a real attempt to quantify when this beats a closed bath, and the claimed scalings are plausible.\n\nThe numerical demonstration on the resonant level model is a decent proof of principle: an 11-qubit noisy circuit reproduces the exact greater Green's function, and the pseudomode fitting is convex, so nothing is fitted into existence.\n\nThe soft spots are real but not fatal. The central assumption is that the noisy bath qubits are pure amplitude dampers (T1-limited) and the impurity/ancilla qubits are noiseless. The paper says this and suggests mitigation, but the emulation in Fig. 5 includes only amplitude damping—no pure dephasing, leakage, crosstalk, or shot noise. So the reported qubit and error numbers are not yet hardware-validated. Also, the test is a single non-interacting resonant level; the extension to interacting impurity models is argued, not demonstrated. The scaling analysis uses uncalibrated constants α, β, γ, Tg, so treat the asymptotic claims as directional.\n\nThese are scoping issues, not logical holes. The paper is honest about its idealizations. The main open question is empirical: is there a device where amplitude damping is the dominant, controllable noise channel and the other qubits are clean enough? If yes, this could be an important resource reduction for DMFT.\n\nWho should read this: people working on quantum algorithms for condensed matter, and experimental groups with tunable T1. It deserves serious peer review. I'd send it out. The strongest improvement would be a hardware demonstration or at least a dephasing-inclusive numerical study.","headline":"A genuine fermionic extension of noise-harvesting for impurity models, with sound derivations and a credible resource advantage, but the practical case rests on an idealized T1-only noise model and a single non-interacting test.","tokens_in":25693,"tokens_out":3216,"would_cite":true,"duration_ms":27972,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["81P68"],"pacs":["03.67.Lx","03.65.Yz"],"model":"deepseek-v4-flash","headline":"Natural qubit decay can be repurposed as the dissipative bath in impurity-model simulations, cutting the qubit count by an order of magnitude and removing ground-state preparation.","keywords":["noise harvesting","amplitude damping","impurity model","dynamical mean field theory","pseudomode","quantum simulation","Lindblad dynamics","Green's functions"],"falsifier":"Run the resonant level model circuit on a device where the noisy qubits have a dephasing time T2 comparable to their relaxation time T1, with no mitigation. If the measured greater Green's function deviates from the exact result by more than the paper's predicted Trotter, fitting, and 1/T1 noise error, the core identification of waiting-time noise with the pseudomode jump operators is disproved for that hardware. A more direct test is quantum process tomography of a noisy qubit during the waiting window: if the channel contains a significant Z-type (dephasing) component, the construction breaks down.","tokens_in":24563,"feed_emoji":"⚛️","tokens_out":6006,"duration_ms":50376,"temperature":0.7,"pith_summary":"The paper proposes a quantum algorithm that turns the dominant decoherence channel of T1-limited qubits, amplitude damping, into a computational resource. In dynamical mean field theory, an impurity model is a small system coupled to a dissipative bath; the authors engineer a circuit in which the bath's fermionic emission and absorption are mimicked by the natural relaxation of noisy qubits, while only a few noiseless qubits hold the impurity and ancilla modes. They claim three advantages: a reduction in the number of qubits by about an order of magnitude in their test case, access to longer-time dynamics without revivals, and automatic preparation of the steady state by relaxation. If the method works on hardware, it would make impurity-model Green's functions, the bottleneck of DMFT, accessible to near-term quantum processors with mixed-quality qubits.","feed_headline":"Noise-harvesting circuit simulates impurity models with 11 qubits","feed_subtitle":"Amplitude damping is repurposed as a dissipative bath, cutting qubit count tenfold and reaching long-time dynamics.","key_machinery":"The pseudomode representation of the bath, in which fermionic bath modes are each coupled to a Markovian reservoir so that the hybridization function is fitted by Lorentzians instead of Dirac peaks, is the first step. The Lindblad jump operators are then mapped to qubit amplitude damping through three transformations: addition of ancilla fermionic modes to make jumps local, a fermion-to-qubit encoding, and a noise-encoding unitary that rotates each jump operator to the qubit operator S−. The dissipation rate is tuned by inserting a waiting time in the Trotter loop, satisfying a relation that matches the physical noise rate to the target dissipation rate. This combination of pseudomode fitting, noise encoding, and waiting-time dissipation carries the argument.","core_discovery":"The central claim is that the dissipative dynamics of a fermionic pseudomode bath can be identified exactly with the amplitude-damping channel of physical qubits, provided a waiting time is inserted after each Trotter step to match the dissipation rate. The paper constructs a noise-encoding unitary that transforms the fermionic jump operators, dressed by Jordan-Wigner strings, into the qubit amplitude-damping operator, so that during the waiting window the qubits' natural T1 decay implements precisely the Lindblad jumps of the pseudomode model. Because the bath is open, its steady state is reached automatically from simple product initial states, and the absence of revivals means a fixed, small number of bath modes suffices for arbitrarily long evolution. Numerically, the greater Green's function of the resonant level model computed with amplitude damping, using 8 bath modes plus one ancilla, agrees with the exact result while using only 11 qubits where a closed bath would need more than 120.","pith_inferences":["Because the construction works for any hybridization function and for interacting impurities, the same encoding should extend to the Anderson impurity model and multi-orbital DMFT baths by applying the pseudomode fit per spin species.","On hardware with appreciable pure dephasing, the waiting-time window would implement a mixture of amplitude damping and dephasing; the paper's own suggestion is zero-noise extrapolation, but a more direct design implication is to engineer qubits with T1-dominated noise so that the harvested channel is clean.","The minimum dissipation rate sets a resolution floor on bath features, suggesting a hardware-software co-design metric: faster gates or longer T1 both sharpen the accessible spectral features.","The revival-free nature of the open-bath representation implies that the qubit advantage grows with the longest time one wants to simulate, so the benefit is largest in weakly coupled or low-temperature regimes where closed baths need the most sites."],"forward_implications":["For a T1-limited processor with a few clean ancilla qubits, computing impurity Green's functions consumes an order of magnitude fewer qubits than a closed-bath simulation at the same target time.","The gate count and total runtime grow only linearly with simulation time rather than quadratically, so long-time dynamics that would require more than 150 closed bath sites are captured with 8 to 32 pseudomodes.","State preparation is automatic: the open bath relaxes to its steady state exponentially fast, eliminating ground-state or Gibbs-state preparation circuits.","The method composes naturally with partial quantum error correction, since noisy qubits are used as a resource while impurity and ancilla qubits are protected.","With optimized Trotter step and bath size, the total error scales as t to the two-thirds, compared with t to the three-halves for closed baths, and the optimal bath size actually shrinks as the target time grows."],"supporting_citations":[{"why":"Extends this prior noise-harvesting approach from bosonic and spin systems to fermionic impurity models.","marker":"[8]"},{"why":"Supplies the pseudomode method of representing non-Markovian baths by modes coupled to a Markovian reservoir.","marker":"[23]"},{"why":"Provides the fermionic quantum regression theorem and Lindblad-discretized leads used to derive Green's functions in pseudomode models.","marker":"[45]"},{"why":"Establishes the fermionic bath-switching equivalence that justifies replacing the physical bath by a pseudomode model with the same hybridization function.","marker":"[49]"},{"why":"Documents the revival effect that forces closed baths to grow linearly with time, the baseline the paper improves on.","marker":"[57]"},{"why":"Sets up dynamical mean field theory and defines the impurity-model bottleneck that the algorithm addresses.","marker":"[11]"}],"fun_headline_variants":["Noise as feature: 11-qubit circuit hits long-time impurity dynamics","Amplitude damping: from noise to automatic state prep in 11 qubits","Quit ground-state search: noise-based bath gives 11-qubit advantage","Turn qubit decay into physics: 11 qubits outdo 120 for impurity models"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The harvested channel is assumed to be dominated by amplitude damping, with impurity and ancilla qubits essentially noiseless, so that the waiting-time evolution is exactly the desired Lindblad dissipation; if pure dephasing or other decoherence is comparable on the noisy qubits, the identification fails unless mitigated.","fun_headline_variants_meta":{"raw":{"variants":["Noise as feature: 11-qubit circuit hits long-time impurity dynamics","Amplitude damping: from noise to automatic state prep in 11 qubits","Quit ground-state search: noise-based bath gives 11-qubit advantage","Turn qubit decay into physics: 11 qubits outdo 120 for impurity models"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000733,"raw_usage":{"total_tokens":3251,"prompt_tokens":890,"completion_tokens":2361,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":506,"completion_tokens_details":{"reasoning_tokens":2274}},"tokens_in":506,"tokens_out":2361,"duration_ms":18448,"temperature":1.0,"reasoning_tokens":2274,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T12:51:52.287329+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the resonant level model circuit on a device where the noisy qubits have a dephasing time T2 comparable to their relaxation time T1, with no mitigation. If the measured greater Green's function deviates from the exact result by more than the paper's predicted Trotter, fitting, and 1/T1 noise error, the core identification of waiting-time noise with the pseudomode jump operators is disproved for that hardware. A more direct test is quantum process tomography of a noisy qubit during the waiting window: if the channel contains a significant Z-type (dephasing) component, the construction breaks down.","supporting_citations":[{"cited_title":"These can be used to derive the form of HFs we use in our work, i.e","cited_arxiv_id":null,"evidence_quote":"Extends this prior noise-harvesting approach from bosonic and spin systems to fermionic impurity models."},{"cited_title":"Gramsch, K","cited_arxiv_id":null,"evidence_quote":"Provides the fermionic quantum regression theorem and Lindblad-discretized leads used to derive Green's functions in pseudomode models."},{"cited_title":"Pleasance, B","cited_arxiv_id":null,"evidence_quote":"Establishes the fermionic bath-switching equivalence that justifies replacing the physical bath by a pseudomode model with the same hybridization function."},{"cited_title":"Dorda, M","cited_arxiv_id":null,"evidence_quote":"Documents the revival effect that forces closed baths to grow linearly with time, the baseline the paper improves on."},{"cited_title":"2 (c) is defined as sZ +∞ 0 dt G> tprep (t) − G> trelax=100 (t) 2 (B8) 16 18 20 22 24 26 28 30 32 Nb 0.85 0.90 0.95 1.00 1.05 1.10 1.15relax / 2 = 0.6 = 0.2 Figure 6","cited_arxiv_id":null,"evidence_quote":"Sets up dynamical mean field theory and defines the impurity-model bottleneck that the algorithm addresses."}],"review_version":1}