{"id":"a94126d7-64e0-4be0-9d49-87ce99accfde","arxiv_id":"2412.07764","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"A 2-local commuting penalty Hamiltonian can stabilize a distance-2 quantum error-detecting code in an excited subspace, enabling scalable error-resilient analog simulation of many-body spin models.","lead":"This paper proposes a method to make analog quantum simulators resilient to single-qubit noise by encoding the simulation into an excited subspace protected by a two-local commuting penalty Hamiltonian. If correct, it provides a scalable, hardware-friendly error-suppression scheme that previous no-go results suggested was impossible.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Scheme's explicit 1D/2D encodings rely on unproven gadget-coexistence Lemmas D.2–D.4; if any is false, the construction fails even for 1-local noise.","rationale":"The reader's CONDITIONAL verdict is appropriate. The most load-bearing concern I find is not the 1-local error model (which is explicit in the claim and acknowledged by the authors) but the unproven gadget-coexistence lemmas. These lemmas are essential to the correctness of the concrete constructions: the 1D TFI and XY encodings and the 2D generalized graph recipe all rely on them to show that multiple perturbative gadgets combine additively. The paper states the proofs are 'almost the same' as Lemma D.1, which is plausible but not demonstrated. Because a single counterexample would invalidate the central claim, this is a genuine risk. The proposed computation directly tests the lemmas and would settle the concern. If the lemmas hold, the remaining issues (omitted full proofs, missing code/data, 1-local noise scope) are addressable limitations rather than fatal flaws, so the CONDITIONAL verdict should stand.","tokens_in":33490,"tokens_out":49820,"duration_ms":414012,"concrete_test":"Numerically verify Lemmas D.2–D.4 for random block parameters (gx, gz satisfying |gx|≠|gz| and any distinctness conditions) and random zero-energy eigenstates. For each allowed operator pair (A,B) in the lemma, compute P0 A H_pen^{-1} B P0 + P0 B H_pen^{-1} A P0 on a Haar-random basis of S0 and check that the spectral norm is below 1e-10. Run at least 1000 instances covering all (i,k) pairs and parameter ranges. A counterexample would break the 1D/2D encodings; if all pass, the lemmas are almost certainly correct and the omitted proofs are a minor exposition gap.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central construction depends on combining many perturbative gadgets without unwanted interference. Appendix D proves Lemma D.1 and then states Lemmas D.2–D.4 with 'proof omitted since it is almost the same as for Lemma D.1.' These lemmas are used explicitly in Appendix E to cancel two-gadget interference for the 1D TFI and XY models (e.g., Eq. E7) and in Appendix F for the 2D constructions. If any lemma is false, the effective Hamiltonian of H_enc^(2) acquires cross terms between distinct gadgets, violating the encoding conditions (14b)/(14d) and causing leakage from S_enc into S0 \\ S_enc. This would invalidate the central claim that the scheme robustly simulates the target Hamiltonians, independent of the error model. The gap is therefore load-bearing for correctness, not merely an issue of exposition.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a general framework for error-resilient analog Hamiltonian simulation using an excited eigenspace of a 2-local commuting penalty Hamiltonian as a quantum error-detecting code. The basic building block is a [[4,2,2]] Hamiltonian code stabilized by commuting 2-local terms; this is extended to a [[4n,2n,2]] code family by block composition. Inner-block logical operations are implemented directly, while cross-block logical operations use perturbative gadgets analyzed via second-order perturbation theory. The main rigorous claim, Theorem III.1, is an infidelity bound of the form (M/λ)(6+13Mt)^2 for 1-local coherent noise. The scheme is applied to 1D XY and TFI models and to 2D TFI and compass models, with numerical evidence showing the predicted λ^{-1} scaling. Appendix C proves a no-go result that no eigenspace of a 2-local commuting Hamiltonian can have code distance greater than 2, which complements the construction.","tokens_in":33653,"tokens_out":25521,"duration_ms":212112,"significance":"If fully established, the scheme is a significant advance: it circumvents the earlier ground-space no-go theorem of Ref. [36] by moving to an excited code space, uses only 2-local physical terms, and requires only a polynomial number of penalty terms for bounded-degree interaction graphs. The concrete code tables and the numerical benchmarks (Figs. 3 and 5) support the central scaling claim. The paper also contains a useful generalized no-go theorem (Appendix C) showing that distance-2 is the best possible for eigenspaces of 2-local commuting Hamiltonians. However, two load-bearing technical points currently prevent the manuscript from being fully convincing: the coexistence lemmas for perturbative gadgets are stated without complete proofs, and the statement of Theorem III.1 appears inconsistent with its proof regarding the definition of M. Both issues are likely fixable, but they affect the central formal claims rather than merely the presentation.","major_comments":[{"comment":"The proofs of Lemmas D.2, D.3, and D.4 are omitted with the statement that they are 'almost the same' as Lemma D.1, but these lemmas are used directly in Appendix E (Eq. (E7)) to cancel two-gadget interference in the 1D TFI and XY encodings, and in Appendix F for the 2D constructions. If any of these lemmas fails, the effective Hamiltonian H_eff^(2) acquires cross terms between distinct gadgets, violating condition (14d) and causing leakage from the encoding subspace S_enc into S0 \\ S_enc; this would invalidate the central claim even for the 1-local noise model. The scalar argument for Lemma D.1 does not transparently cover the overlapping-support cases in D.2–D.4, so the verification gap is load-bearing. Please provide complete proofs or a rigorous, explicit case analysis for these lemmas.","section":"Appendix D (Lemmas D.2–D.4) and Appendices E–F"},{"comment":"The proof of Theorem B.3 uses the inequalities ||W||/λ ≤ sqrt(M/λ) + M/λ and ||A||/λ + ||B||/√λ ≤ (M/λ + 2√(M/λ)) M. With the theorem's stated definition M := max{||H_enc^(2)||_2, ||H_enc^(1)+V||}, the first step would require ||H_enc^(2)||/√λ ≤ sqrt(M/λ), i.e. ||H_enc^(2)|| ≤ √M, which is not guaranteed; the correct bound is only M/√λ + M/λ. Similarly, under the stated definition ||B|| can be as large as 2M^2, not 2M^{3/2}. Thus the condition λ ≥ 25M does not imply the convergence condition κ ≤ 1/4 required by Theorem B.1, and the main bound (16) is not established as stated. The definition of M and the proof must be made consistent, for example by defining M := max{||H_enc^(2)||_2^2, ||H_enc^(1)+V||} and adjusting the bound accordingly, or by adding an explicit assumption on ||H_enc^(2)||.","section":"Theorem III.1 and Theorem B.3 (Appendix B)"}],"minor_comments":[{"comment":"The captions should clarify whether the quoted lattice dimensions (e.g., '2×3 square lattice') refer to logical sites or physical qubits; the main text suggests logical sites, but an explicit statement would prevent confusion.","section":"Figures 3 and 5"},{"comment":"The claimed local-unitary equivalence to the CSS [[4,2,2]] stabilizer code would be easier to verify with an explicit mapping of the logical Pauli operators or a direct reference to the standard code; the stabilizer eigenvalue check alone does not fully determine the logical operator correspondence.","section":"Appendix A"},{"comment":"The block parameters g_x = 2^{2C}, g_z = 2^{2C+1} depend on the color C; for bounded-degree graphs the number of colors is constant, so these are constants independent of system size, but the main text should state this assumption explicitly when discussing scalability.","section":"Appendix F, Eq. (F1)"},{"comment":"The statement that the scheme is 'hardware-efficient' should be qualified by the requirement of generic 2-local couplings (including X–Y type interactions) and by the potential need for Floquet engineering in platforms with native ZZ couplings; the final paragraph acknowledges this, but the earlier claims should be tempered accordingly.","section":"Section V (Discussion)"},{"comment":"The proof of Lemma D.6 is given 'for concreteness' for a particular choice of A and B; the general case is said to follow similarly, but a brief explanation of how the energy-denominator argument extends to all listed operator types would improve verifiability.","section":"Appendix D, Lemma D.6"}],"recommendation":"major_revision","confidential_remarks":"The core idea—using an excited eigenspace of a 2-local commuting Hamiltonian to obtain a distance-2 code and thereby circumvent the ground-space no-go—is novel and likely publishable after the technical gaps are repaired. The omitted gadget-coexistence proofs are the principal review risk; the inconsistency in the definition of M in Theorem III.1 is a more straightforward correction. I would encourage the authors to supply the missing proofs and to double-check the consistency of the theorem statement with the proof before resubmission."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nWorth your attention: this is the first construction I know of that gets error-resilient analog simulation from 2-local commuting penalties by encoding into an excited subspace of the penalty Hamiltonian instead of the ground space. That genuinely sidesteps the Marvian-Lidar no-go theorem. The authors also prove a generalized no-go showing distance 2 is the best any eigenspace of a commuting 2-local Hamiltonian can do. The main bound (Theorem III.1) is proved in detail, and the numerics match the predicted 1/λ scaling. The [[4n,2n,2]] code family and the gadget tables are explicit and usable.\n\nThe main soft spot: the coexisting-gadget lemmas, D.2–D.4, are stated without proof, with a note that they are almost the same as Lemma D.1. Those lemmas are what cancel interference between neighboring gadgets in the 1D and 2D encodings, so the construction really does depend on them. The omitted proofs may well be routine—D.1 and D.5 are proved carefully and the pattern is consistent—but a referee needs to see them before the central claim is fully established. This is a verification gap, not a demonstrated flaw.\n\nSecond, the error model is restricted to 1-local coherent noise. Appendix C proves that 2-local commuting Hamiltonians cannot protect against 2-local errors, so the scheme cannot be pushed further without new ideas. The paper states this clearly, and the Floquet discussion for ZZ-only hardware is appropriately cautious.\n\nThe paper is honest about what it does and does not do. It accurately describes the prior no-go result as ground-space-specific, and the claim to overcome it is fair. No code or data is shipped, but the simulations are small and reproducible.\n\nThis is a solid, novel theoretical construction with a rigorous bound. Send it to referees, with the instruction that the omitted gadget-coexistence proofs be supplied. I would bring it to a reading group.\n\nRecommendation: accept for peer review, with the unproven lemmas as the required revision.","headline":"Genuinely new excited-subspace encoding that bypasses the Marvian-Lidar no-go theorem with 2-local commuting penalties, but the construction leans on unproven gadget-coexistence lemmas that need to be checked.","tokens_in":34250,"tokens_out":2853,"would_cite":true,"duration_ms":24344,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Excited-state quantum codes beat a no-go for analog simulator noise protection","keywords":["analog quantum simulation","energy-gap protection","quantum error-detecting codes","commuting 2-local Hamiltonians","excited encoding subspace","perturbative gadgets","transverse-field Ising model","infidelity bound"],"falsifier":"With a device that supports the required 2-local couplings, encode the 2D transverse-field Ising model, inject calibrated single-qubit Pauli errors of fixed strength, and compare average infidelity at two penalty strengths (for example, $\\lambda$ and $4\\lambda$) at a fixed time; Theorem III.1 predicts a fourfold drop in infidelity, so a measured scaling flatter than $1/\\lambda$ would falsify the error-suppression claim.","tokens_in":33259,"feed_emoji":"🛡️","tokens_out":6716,"duration_ms":60813,"temperature":0.7,"pith_summary":"This paper claims a general recipe for making analog Hamiltonian simulations resilient to single-qubit coherent noise using only two-local, mutually commuting penalty terms. The key move is to encode the computation in an excited eigenspace of the penalty Hamiltonian rather than its ground space, which sidesteps a no-go theorem that ruled out ground-space encodings with such 2-local commuting penalties. The authors prove a rigorous infidelity bound for the resulting scheme and show numerically that the protected simulation time extends by orders of magnitude for 1D XY, 1D and 2D transverse-field Ising, and 2D compass models. If correct, the result turns a previously forbidden regime into a scalable, hardware-friendly route to longer-lived analog quantum simulation.","feed_headline":"Excited-state quantum codes beat a no-go for analog simulator noise protection","feed_subtitle":"Encoding in an excited subspace of 2-local commuting penalties suppresses single-qubit noise with polynomial overhead.","key_machinery":"The load-bearing object is the [[4,2,2]] Hamiltonian code: four physical qubits grouped into two Bell pairs, with the logical subspace spanned by $|0+\\rangle=|\\Phi_{00}\\rangle|\\Phi_{11}\\rangle$, $|0-\\rangle=|\\Phi_{10}\\rangle|\\Phi_{01}\\rangle$, $|1+\\rangle=|\\Phi_{01}\\rangle|\\Phi_{10}\\rangle$, $|1-\\rangle=|\\Phi_{11}\\rangle|\\Phi_{00}\\rangle$, stabilized by the commuting 2-local penalty $H_{\\rm pen}(g)$ with $|g_x/g_z|\\neq1$. The code satisfies the error suppression condition $P_{\\rm enc}V_iP_{\\rm enc}=c_iP_{\\rm enc}$ for every single-qubit Pauli $V_i$, and its tensor-product extension $[[4n,2n,2]]$ keeps the penalty 2-local with a constant gap. Perturbative gadgets of the form $\\alpha Z^{(a)}_2X^{(b)}_3+\\beta Z^{(a)}_4X^{(b)}_3$ convert 2-local cross-block physical terms into logical Ising and XY interactions through second-order perturbation theory, and the proof of Theorem III.1 runs through a generalized timescale-separation bound that controls the difference between the noisy and effective evolutions.","core_discovery":"The central discovery is that the [[4,2,2]] quantum error-detecting code can be realized as an excited, zero-energy eigenspace of a purely 2-local commuting penalty Hamiltonian $H_{\\rm pen}(g)=g_xX_1X_2+g_zZ_1Z_2+g_xX_3X_4+g_zZ_3Z_4$ with $|g_x/g_z|\\neq1$, and that tensor products of these blocks give a $[[4n,2n,2]]$ code protected by a constant energy gap. Because the encoding lives in an excited subspace, the construction evades the no-go theorem of Ref. [36] that prohibits ground-space protection by commuting 2-local Hamiltonians. The paper proves (Theorem III.1) that for 1-local coherent errors $V=\\sum_i\\epsilon_i V_i$, the simulation infidelity is bounded by $1-|\\langle\\psi_{\\rm tar}(t)|\\psi(t)\\rangle|^2 \\le (M/\\lambda)(6+13Mt)^2$, so the error can be made arbitrarily small with a penalty strength that grows only polynomially with system size. Cross-block logical couplings are generated by perturbative gadgets whose second-order effective interactions are computed explicitly, and a generalized no-go result (Appendix C) shows distance-2 protection is the best any eigenspace of a commuting 2-local Hamiltonian can achieve.","pith_inferences":["Beyond the paper: the same excited-subspace idea might protect against biased noise (for example, only $Z$-type errors) with distance-2 codes tuned to that bias, since the obstruction in Appendix C concerns full 2-local error sets rather than structured subsets.","Beyond the paper: if hardware noise is dominated by 2-local crosstalk, the scheme as stated offers no protection; a testable extension would be to interpose a dynamical-decoupling or Floquet layer that converts 2-local errors into effectively 1-local ones before the penalty acts.","Beyond the paper: the graph-coloring parameter assignment used for 2D lattices suggests a systematic route to other finite-degree interaction graphs, and one could test whether the number of distinct block parameters stays bounded for, say, random regular graphs.","Beyond the paper: the bound's linear growth in $t$ hints that the protection window scales as $\\lambda/M^2$; measuring the crossover time as a function of $\\lambda$ would give a clean experimental signature."],"forward_implications":["A simulator built from 2-local commuting penalty terms can keep single-qubit coherent errors under control with penalty strength growing polynomially, not exponentially, in system size.","The same 2-local gadget construction covers 1D XY, 1D and 2D transverse-field Ising, and 2D compass models, with numerics showing protected evolution time extended by orders of magnitude at fixed noise strength.","Because the no-go extended in Appendix C limits any eigenspace of a 2-local commuting Hamiltonian to distance 2, this construction saturates what that class of penalties can achieve.","The infidelity bound $\\propto 1/\\lambda$ (with perturbative gadgets) is concrete enough to specify the penalty strength needed for a target accuracy at a given simulation time."],"supporting_citations":[{"why":"Proves the no-go theorem for ground-space protection by 2-local commuting Hamiltonians, the barrier this work overcomes by moving to an excited subspace.","marker":"[36]"},{"why":"Previous two-body robust Hamiltonian construction that the authors contrast with their polynomial-overhead, commuting-penalty scheme.","marker":"[42]"},{"why":"Supplies the timescale-separation identity used in the proof of the infidelity bound.","marker":"[61]"},{"why":"Structure lemma for 2-local commuting Hamiltonians used to extend the no-go to any eigenspace (distance at most 2).","marker":"[62]"},{"why":"Documents single-qubit coherent errors as the dominant noise in current analog simulators, motivating the error model.","marker":"[49]"},{"why":"Identifies the code as equivalent to the standard CSS [[4,2,2]] stabilizer code up to a single-qubit unitary.","marker":"[51]"},{"why":"Numerical simulation package used for the reported scaling checks of the infidelity bound.","marker":"[55]"}],"fun_headline_variants":["Excited-state codes beat no-go for analog sim noise","2-local commuting code shields analog simulators","Constant-gap protection from 2-local terms","No-go defied: excited codes for robust analog sims"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The scheme's protection is proven only for single-qubit coherent Pauli noise and assumes controlled access to generic two-local couplings; if two-qubit or stochastic errors dominate, the error-suppression condition fails and the paper's own no-go extension blocks any 2-local commuting penalty from protecting against them.","fun_headline_variants_meta":{"raw":{"variants":["Excited-state codes beat no-go for analog sim noise","2-local commuting code shields analog simulators","Constant-gap protection from 2-local terms","No-go defied: excited codes for robust analog sims"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000319,"raw_usage":{"total_tokens":1838,"prompt_tokens":1024,"completion_tokens":814,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":640,"completion_tokens_details":{"reasoning_tokens":751}},"tokens_in":640,"tokens_out":814,"duration_ms":8397,"temperature":1.0,"reasoning_tokens":751,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T18:33:11.926729+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"With a device that supports the required 2-local couplings, encode the 2D transverse-field Ising model, inject calibrated single-qubit Pauli errors of fixed strength, and compare average infidelity at two penalty strengths (for example, $\\lambda$ and $4\\lambda$) at a fixed time; Theorem III.1 predicts a fourfold drop in infidelity, so a measured scaling flatter than $1/\\lambda$ would falsify the error-suppression claim.","supporting_citations":[{"cited_title":"Marvian and D","cited_arxiv_id":null,"evidence_quote":"Proves the no-go theorem for ground-space protection by 2-local commuting Hamiltonians, the barrier this work overcomes by moving to an excited subspace."},{"cited_title":"Robust universal Hamiltonian quantum computing using two-body interactions","cited_arxiv_id":"1911.01354","evidence_quote":"Previous two-body robust Hamiltonian construction that the authors contrast with their polynomial-overhead, commuting-penalty scheme."},{"cited_title":"Bravyi and M","cited_arxiv_id":null,"evidence_quote":"Structure lemma for 2-local commuting Hamiltonians used to extend the no-go to any eigenspace (distance at most 2)."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Documents single-qubit coherent errors as the dominant noise in current analog simulators, motivating the error model."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Identifies the code as equivalent to the standard CSS [[4,2,2]] stabilizer code up to a single-qubit unitary."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Numerical simulation package used for the reported scaling checks of the infidelity bound."}],"review_version":1}