{"id":"c724c0b4-f602-4640-9de9-b52c8dd51602","arxiv_id":"2504.21575","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"Tadpole improvement factors in real-time lattice gauge theory simulations are state- and time-dependent and should be updated self-consistently at each time step.","lead":"This paper shows that the standard 'tadpole improvement' correction used in lattice gauge theory simulations must become time- and position-dependent in real-time dynamics, not a fixed constant. The authors demonstrate with small SU(2) lattice simulations that an entangled initial state makes the correction sizable, which matters for future quantum simulations of particle collisions.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The numerical evidence shows differences among evolutions, but no continuum-limit benchmark demonstrates that the state-dependent u0,i actually reduces lattice-spacing artifacts; 'improvement' is asserted, not shown.","rationale":"The reader's weakest assumption correctly identifies that Eq. (6) is an unproven extrapolation to non-equilibrium states. My concern is closely related but more direct: even granting that identification, the paper never shows that the proposed scheme reduces lattice-spacing artifacts. Sections III.A and III.B present only fixed-a=1 simulations, so the observed differences could be a redefinition of the lattice Hamiltonian rather than an improvement. The reader's conditional verdict is appropriate because the idea is plausible and the proof-of-principle is suggestive; however, the conditions for acceptance should explicitly include a continuum-spacing test. I therefore keep the verdict unchanged (CONDITIONAL) while adding this requirement. The reader's focus on truncation sensitivity and reproducibility is also valid, but the continuum test is the most decisive check of the central claim.","tokens_in":10796,"tokens_out":6182,"duration_ms":67087,"concrete_test":"Repeat the L=10 plaquette-chain and 7x3 honeycomb simulations at two lattice spacings, e.g., a=1 and a=0.5, keeping the physical size (La and Lx a, Ly a) and physical coupling fixed by rescaling g appropriately in Eqs. (1), (9), (13), (14). Compute the time-dependent electric energy density under the unimproved, ground-state-tadpole, and dynamical-tadpole Hamiltonians. If the dynamical-tadpole results at a=1 and a=0.5 agree significantly better than the unimproved or constant-tadpole results (i.e., explicit a-dependence is reduced), the central claim is supported. If the difference between the two lattice spacings is comparable in all three schemes, the 'improvement' is not demonstrated and the verdict should be reconsidered.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is that using a constant (vacuum) tadpole factor leads to unquantified errors and that the time-dependent local factor from Eq. (6) provides the correct improvement. Sections III.A and III.B compare three evolutions (unimproved, ground-state-improved, dynamical) at a single lattice spacing (a=1 in Eqs. (9) and (14)). The three differ, but difference is not improvement: there is no continuum extrapolation, no physical benchmark, and no demonstration that the dynamical-tadpole results have reduced O(a) sensitivity. The instantaneous-mean-field identification in Eq. (6), extrapolated from the Euclidean-vacuum Lepage-Mackenzie construction (Ref. [18]), is also not derived for non-equilibrium states; the observed deviation could be a truncation artifact of jmax=1/2 (which the authors acknowledge) or a mis-specified renormalization. Thus the paper establishes a phenomenon (state-dependent tadpole factors) but not that the proposed protocol is an improvement in the sense of reducing discretization errors.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper argues that in real-time (Minkowski) Hamiltonian lattice gauge theory simulations, the tadpole-improvement factor relating lattice link variables to continuum fields is not a constant per configuration but is generically space- and time-dependent, because it is fixed by the expectation value of the plaquette operator in the instantaneous state. The authors propose an iterative protocol in which u0,i(t) is recomputed self-consistently at each time step from local plaquette expectation values, Eq. (6), and they demonstrate the resulting dynamics for a jmax=1/2-truncated SU(2) plaquette chain and a 2+1D honeycomb lattice, comparing unimproved, ground-state-improved, and dynamically-improved evolutions for two classes of initial states. They also note that the improved Hamiltonian becomes non-translationally invariant during evolution and discuss algorithmic challenges for measuring the off-diagonal plaquette operators.","tokens_in":10989,"tokens_out":5734,"duration_ms":63384,"significance":"If the central claim is correct, the paper identifies a genuinely new systematic element for quantum simulations of non-equilibrium gauge theories: the standard constant-vacuum tadpole factor is not adequate for states with local energy-density inhomogeneities, and an instantaneous mean-field update is needed. This would matter for simulations of scattering, fragmentation, and other processes with spatially varying energy density. The paper's strengths are the clear algorithmic statement in Fig. 1, the explicit two-system numerical demonstration that u0,i(t) is state- and position-dependent, the use of two different initial-state structures (product and entangled), and the honest identification of truncation and implementation issues in Sec. IV. What is not yet shown is that the proposed protocol actually reduces lattice-spacing artifacts; the numerical evidence is at a single lattice spacing, a single truncation, and small volumes. The conceptual point is plausible and the paper is a useful contribution, but the central 'improvement' claim still requires validation.","major_comments":[{"comment":"The central claim that the dynamical local scheme is an 'improvement' is not established by the numerical evidence. All simulations are performed at a=1, jmax=1/2, and fixed small volumes, so the differences among unimproved, ground-state-improved, and dynamically-improved evolutions demonstrate state dependence, but difference is not improvement: there is no continuum extrapolation, no variation of the gauge-field truncation, and no benchmark showing that the dynamical-tadpole results have reduced O(a) sensitivity. I request a concrete quantitative test: repeat at least one of the two systems at two or three lattice spacings with the appropriate coupling rescaling, and with jmax=1/2 versus a higher truncation, and show that the dynamical local scheme reduces the lattice-spacing dependence of a chosen observable such as the time-dependent electric energy density.","section":"III.A, III.B, Figs. 3–7"},{"comment":"The identification u0,i = (1 + (1/(2Nc))⟨ψ|□i + □i†|ψ⟩)^(1/4) is taken directly from the Euclidean vacuum Lepage-Mackenzie construction and applied to arbitrary instantaneous states without derivation. In Ref. [18] the plaquette expectation value is a proxy for the link renormalization in the vacuum ensemble generated by the improved action; for a generic out-of-equilibrium state there is no first-principles argument that the same relation removes the lattice-spacing artifacts from the Hamiltonian. The paper shows that the iterative procedure converges, but convergence of a self-consistent mean-field equation does not imply that the fixed point is the correct renormalization. This is a load-bearing assumption for the paper's 'improvement' claim. I request either a derivation in a controlled setting (e.g., perturbative evaluation in a time-dependent background) or a numerical check against a known continuum-limit result.","section":"II, Eq. (6)"},{"comment":"The observed magnitude of the dynamical-tadpole effect could be enhanced by the jmax=1/2 truncation and by the small-system/boundary choices; the authors acknowledge this possibility in Sec. IV. Since all quantitative statements about 'significant' deviations (Figs. 4 and 7) are made in this truncated Hilbert space, the relevance for the full SU(2) theory is not yet quantified. I ask that the revision include a systematic assessment of how the effect changes with truncation and system size, at least for one of the two setups, so that the reader can separate genuine tadpole physics from truncation artifacts.","section":"III.A, IV"}],"minor_comments":[{"comment":"The state |ψ2⟩ is written as N □0 |ψGS⟩_{u0}; the subscript notation on the ground-state ket is not defined and should be explained (presumably the ground state of the Hamiltonian with the converged tadpole factors).","section":"III.A, Eq. (12)"},{"comment":"The sentence 'the coefficients of the four terms in this controlled-plaquette operator can be found in Ref. [20]' is confusing because explicit coefficients already appear in Eq. (7); please state exactly which coefficients are meant.","section":"III.A, Eq. (7) and footnote 5"},{"comment":"The captions say 'The solid line corresponds to...' but each figure shows multiple solid curves distinguished by color; please use 'solid curves' and clarify the color coding in the captions.","section":"Figs. 3, 4, 6, 7"},{"comment":"The sentence 'In the continuum, it is UV divergent' has an ambiguous antecedent; please rephrase to refer to the tadpole diagram or the two-point contraction explicitly.","section":"II, after Eq. (5)"},{"comment":"The statement that 'the number of steps required to measure all plaquettes are independent of system size' needs grammatical correction to 'is independent' and, more importantly, should be supported by the promised simple protocols rather than left as an assertion.","section":"IV"}],"recommendation":"major_revision","confidential_remarks":null},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The new idea here is real and worth taking seriously. Prior work on tadpole improvement in Hamiltonian simulations assumed a constant, uniform factor. Illa, Savage, and Yao point out that in a non-equilibrium state, the mean-field value of the plaquette—and hence the improvement factor—should be local and time-dependent. That observation is correct as far as the standard renormalization argument goes, and the iterative self-consistent protocol in Fig. 1 is a practical way to implement it. The numerical demonstrations on plaquette chains and honeycomb lattices show the expected effect: the dynamical factors differ from the vacuum or unimproved cases, and the difference is larger when the initial state is entangled. The paper is also honest about its main limitation, jmax = 1/2 truncation, and lists higher truncation as necessary follow-up.\n\nThe soft spot is that the paper calls this an \"improvement\" without showing that it reduces lattice-spacing artifacts. All the numerics are at a single lattice spacing, a = 1, and the comparisons are between three unimproved/improved Hamiltonians. Three different evolutions demonstrate sensitivity, but they do not tell you which one is closer to the continuum. There is no continuum extrapolation, no known benchmark observable, and no evidence that the dynamical prescription goes in the right direction. The instantaneous-mean-field identification in Eq. (6) is also an extrapolation of the Euclidean-vacuum Lepage-Mackenzie construction to arbitrary out-of-equilibrium states; it is plausible, but not derived. These are not fatal flaws for a protocol paper, but they do mean the central quantitative claim is not yet established. The lack of code or data and the absence of error bars or iteration tolerances also make the numerics hard to verify, though for small exact simulations that is a minor issue.\n\nWho is this for? Anyone doing real-time Hamiltonian simulations of lattice gauge theories. The paper identifies a previously unquantified source of systematic error and offers a concrete way to handle it. As a proof of principle, it works. The referee should ask the authors to either provide a continuum-limit test in a small system or soften \"improvement\" to \"state-dependent renormalization,\" and to make the numerical details reproducible. I would send it to peer review.","headline":"The genuinely new observation—that tadpole improvement factors are spacetime-dependent in real-time gauge-theory simulations—is right, but the paper demonstrates sensitivity, not yet improvement; still worth a serious referee.","tokens_in":740,"tokens_out":2303,"would_cite":true,"duration_ms":49732,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["11.15.Ha"],"model":"deepseek-v4-flash","headline":"In real-time lattice gauge theory simulations, the tadpole improvement factor is generically different at each plaquette and changes with time.","keywords":["tadpole improvement","lattice gauge theory","quantum simulation","Kogut-Susskind Hamiltonian","SU(2) Yang-Mills","plaquette operator","real-time evolution","mean-field renormalization"],"falsifier":"A direct check would be to run the same small SU(2) systems at a sequence of smaller lattice spacings (or larger volumes) and compare continuum extrapolations of observables such as electric energy density with and without dynamical local tadpole improvement. If the dynamically improved evolution does not reduce the lattice-spacing dependence relative to vacuum-improved or unimproved evolution, the central claim fails. A more analytic falsifier would be a weak-coupling calculation of the exact time-dependent tadpole coefficient for a simple out-of-equilibrium state, compared against Eq. (6).","tokens_in":10568,"feed_emoji":"⚛️","tokens_out":6656,"duration_ms":65592,"temperature":0.7,"pith_summary":"Quantum simulations of lattice gauge theories aim to extract real-time dynamics from a discretized Hamiltonian, but the link variables appearing in that Hamiltonian are not bare: quantum fluctuations renormalize them through tadpole diagrams. This paper's central claim is that in real-time (Minkowski) simulations this renormalization factor, the tadpole-improvement factor $u_{0,i}$, depends on the local gluonic environment and on time, so it is different for every plaquette and must be updated self-consistently as the wavefunction evolves. Using a constant value, such as the vacuum value, or omitting the improvement altogether, leaves unquantified lattice-spacing errors in the predicted time evolution. The authors demonstrate the effect numerically for small truncated SU(2) gauge systems, finding that the choice matters most when the initial state is entangled and has localized energy density. If this is right, tadpole improvement in quantum simulation is a dynamical, measurement-fed step rather than a fixed preprocessing constant.","feed_headline":"Tadpole fixes in gauge-theory simulations are not constant","feed_subtitle":"When every plaquette's coefficient is updated from the local state, SU(2) time evolution shifts noticeably.","key_machinery":"The load-bearing object is equation (6), the local mean-field tadpole factor $u_{0,i}$, defined from the instantaneous expectation value of the plaquette operator plus its Hermitian conjugate in the current quantum state. At each Trotter step it is inserted into the Kogut-Susskind Hamiltonian's magnetic term as $1/u_{0,i}^4$ (or $1/u_{0,i}^6$ for the honeycomb cells), and the iteration is repeated until convergence. This converts a fixed constant of lattice perturbation theory into a state-dependent, space-time-dependent coupling, and it is the mechanism by which ultraviolet link self-energy corrections are supposed to be removed from real-time dynamics.","core_discovery":"On the paper's own terms, the discovery is that the mean-field tadpole renormalization is a property of the quantum state, not of the lattice action. The standard tadpole-improvement factor, $u_{0,i} = \\bigl(1 + \\frac{1}{2N_c}\\langle\\psi|\\hat{\\square}_i+\\hat{\\square}_i^\\dagger|\\psi\\rangle\\bigr)^{1/4}$, is a local, instantaneous observable of the evolving state. In Euclidean Monte Carlo simulations the ensemble average makes this a single global constant per configuration, but in real-time evolution the expectation value of the plaquette operator depends on the surrounding interaction environment, so each plaquette has its own $u_{0,i}(t)$. The paper argues that the Hamiltonian must therefore be updated at every time step with these local factors, iterated to convergence, and that using either the unrenormalized Hamiltonian or a fixed vacuum value introduces errors that are not controlled as the lattice spacing is reduced. Numerical experiments on 10-plaquette SU(2) chains and $7\\times3$ honeycomb lattices, truncated to $j_{\\max}=1/2$, show that for a product-state initial condition the local factors remain near unity, while for an entangled initial state obtained by applying a plaquette operator to the interacting vacuum, the time evolution with dynamical local tadpole improvement differs substantially from both unimproved and vacuum-improved evolution. The paper further notes that the resulting Hamiltonian is generically not translationally invariant during evolution, even though the interacting vacuum is.","pith_inferences":["Beyond the paper's numerics, the self-consistent iteration of $u_{0,i}(t)$ is a nonlinear feedback into the evolution operator; one could view it as a time-dependent mean-field approximation whose back-reaction on entanglement and thermalization in larger systems is untested.","A natural next test is to compare dynamically improved evolution against exact continuum-limit results in integrable or weak-coupling regimes where perturbation theory provides a benchmark; the paper does not perform such a comparison.","The protocol's practical cost, measuring every plaquette every Trotter step, could be reduced by exploiting smoothness of $u_{0,i}(t)$ and sampling sparsely in time, but the paper notes only that the measurement count can be kept system-size-independent in special bases, not generally."],"forward_implications":["Real-time gauge-theory algorithms must measure, at least in principle, the local plaquette expectation values at each time step and insert them into the Hamiltonian; a fixed vacuum-improvement factor is not a controlled approximation.","During evolution from generic initial states, the improved Hamiltonian is no longer translationally invariant, so circuit constructions or error analyses that assume Hamiltonian symmetries must be revisited.","The effect is amplified when the initial state is entangled and has localized excess energy density, pointing to hadronization, heavy-ion collisions, and strong-field QED as settings where this correction matters.","Tadpole terms are the dominant lattice-spacing artifacts to the energies, while higher-order Symanzik and Hamiltonian improvements are parametrically smaller by comparison."],"supporting_citations":[{"why":"Supplies the Symanzik improvement framework that separates classical lattice-spacing artifacts from the quantum tadpole renormalization considered here.","marker":"[9]"},{"why":"Provides the improved honeycomb-lattice Hamiltonian used as the starting point for the 2+1D demonstration.","marker":"[16]"},{"why":"Defines the tadpole-improvement coefficient from the plaquette expectation value, the formula this paper generalizes to state- and time-dependent local values.","marker":"[18]"},{"why":"Introduces the Kogut-Susskind Hamiltonian whose magnetic term carries the plaquette operators being renormalized.","marker":"[19]"},{"why":"Gives the qubit mapping of SU(2) plaquette chains and the controlled-plaquette operators used in the simulations.","marker":"[20]"},{"why":"Introduce the honeycomb-lattice spin mapping and hex-plaquette operator used for the 2+1D simulations.","marker":"[24, 25]"}],"fun_headline_variants":["Tadpole improvement is not constant—it's local and dynamical","Real-time gauge sims require state-dependent tadpole updates","Local tadpole factors evolve with the quantum state","Entangled initial states reveal dynamical tadpole effects","In real time, tadpole improvement is a local moving target"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the plaquette expectation value in the instantaneous out-of-equilibrium state gives the correct mean-field renormalization of the Hamiltonian at that same time step, an extrapolation of the Euclidean-vacuum Lepage-Mackenzie formula to arbitrary non-equilibrium states; the paper does not derive this from first principles.","fun_headline_variants_meta":{"raw":{"variants":["Tadpole improvement is not constant—it's local and dynamical","Real-time gauge sims require state-dependent tadpole updates","Local tadpole factors evolve with the quantum state","Entangled initial states reveal dynamical tadpole effects","In real time, tadpole improvement is a local moving target"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000944,"raw_usage":{"total_tokens":4051,"prompt_tokens":983,"completion_tokens":3068,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":599,"completion_tokens_details":{"reasoning_tokens":2988}},"tokens_in":599,"tokens_out":3068,"duration_ms":25186,"temperature":1.0,"reasoning_tokens":2988,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-16T04:59:24.793332+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A direct check would be to run the same small SU(2) systems at a sequence of smaller lattice spacings (or larger volumes) and compare continuum extrapolations of observables such as electric energy density with and without dynamical local tadpole improvement. If the dynamically improved evolution does not reduce the lattice-spacing dependence relative to vacuum-improved or unimproved evolution, the central claim fails. A more analytic falsifier would be a weak-coupling calculation of the exact time-dependent tadpole coefficient for a simple out-of-equilibrium state, compared against Eq. (6).","supporting_citations":[{"cited_title":"L¨ uscher and P","cited_arxiv_id":null,"evidence_quote":"Supplies the Symanzik improvement framework that separates classical lattice-spacing artifacts from the quantum tadpole renormalization considered here."}],"review_version":1}