{"id":"998bc2e8-7f58-4426-a8b6-30c5beb0e3ec","arxiv_id":"2501.00443","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Weakly interacting fermionic lattice systems have a constant spectral gap in a Gibbs sampler Lindbladian, giving O(n) mixing time and efficient quantum Gibbs state preparation at any fixed temperature.","lead":"This paper proves that when particles in a fermionic lattice interact only weakly, a specially designed quantum cooling process can reach thermal equilibrium quickly at any temperature. The result becomes a mathematical guarantee that such thermal states can be prepared efficiently on a quantum computer.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Application of Hastings' stability theorem to the third-quantized parent Hamiltonian is the load-bearing step; its hypotheses must be verified directly.","rationale":"The reader's weakest assumption correctly identifies the dependence on the external stability theorem (Theorem 2 of [30]) as the least secured element of the proof. The internal propositions are asserted with proofs, but the actual application of Hastings' theorem to the enlarged third-quantized geometry is not independently justified beyond citing the theorem. The main result would collapse if the constants in Propositions 2 or 3 failed to be uniform in n, or if the support of the local approximants in Proposition 3 did not match the lattice geometry assumed by the theorem. This is a genuine load-bearing concern, but it is a standard black-box reliance and the paper gives a coherent route to verifying it. I therefore keep the reader's CONDITIONAL verdict unchanged: the proof should be revised to spell out the verification of Theorem 2's hypotheses, and the terse calculations and typos should be cleaned up, but there is no demonstrated internal contradiction that would justify rejection.","tokens_in":38408,"tokens_out":31215,"duration_ms":335866,"concrete_test":"Independently recompute the decay constants in Propositions 2 and 3 for the a-fermion setting, tracking every step of Lemmas 12-22 and Definition 8 for a sequence of system sizes n = 8, 16, 32 in dimension D = 1. In particular, verify that the (C U, mu)-decay of V_parent and the [K, nu]-decay of H_parent0's coefficient matrix are uniform in n, and that C U is below the threshold C0(Delta0, K, nu, mu, D) required by Theorem 2. If a single constant grows with n, the gap bound fails; if all constants stay bounded, the black-box application is sound.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central gap lower bound in Theorem 3, and hence Theorem 1, is obtained by applying Theorem 2 (from [30]) to H_parent = H_parent0 + V_parent. For that theorem to give a constant gap, three conditions must all hold with constants uniform in the system size: (i) H_parent0 is a quadratic free-fermion Hamiltonian with spectral gap lower bounded by a constant independent of n; (ii) its coefficient matrix has [K, nu]-decay with K, nu independent of n; and (iii) V_parent has (C U, mu)-decay with C, mu independent of n and C U below a threshold depending only on beta, r0, D. Propositions 1, 2, and 3 assert these conditions, but Proposition 3's proof is long and ends with a decay statement in the operator norm (Definition 6 / Definition 7) whose constants come from a chain of Lieb-Robinson approximations (Lemmas 13-22). The third-quantized geometry is not the same as the geometry in [30]: each physical site carries four a-Majorana modes, parity sectors must be handled separately, and the mapping Phi is not an algebra homomorphism. If any of the constants C, mu, K, nu implicitly acquires an n-dependence through the 4n x 4n enlarged system, or if the support of the local approximants W_{j,r} is not contained in balls of the original lattice, then Theorem 2 does not apply and the claimed n-independent mixing time collapses. Since Theorem 2 itself is not proved in this paper, the correctness of the main result inherits all of its hypotheses without an independent check.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the mixing time of the efficiently implementable Lindbladian of Chen, Kastoryano, and Gilyén when applied to weakly interacting fermionic lattice Hamiltonians H = H0 + V, where H0 is quadratic and geometrically local and V is a parity-preserving interaction of strength U. The main claim (Theorem 1) is that for every inverse temperature β > 0 there is a positive Uβ depending only on β, r0, D such that for U < Uβ the even-parity sector of the Lindbladian has spectral gap bounded below by a constant independent of the system size, giving ε-mixing time O(n + log(1/ε)) and hence an O(n^3 polylog(n/ε)) quantum Gibbs-state-preparation algorithm. The proof maps the Lindbladian to a parent Hamiltonian by third quantization, computes the gap and decay of the non-interacting parent Hamiltonian, proves that the interacting correction has (CU, μ)-decay using Lieb-Robinson bounds, and then imports the stability theorem for free-fermion Hamiltonians from [30]. The paper also derives exponential clustering of correlations for the Gibbs state from the gapped parent Hamiltonian. The overall architecture is coherent and the parity/superselection handling is careful, but two load-bearing estimates in the locality proofs need repair.","tokens_in":38714,"tokens_out":35565,"duration_ms":357922,"significance":"If the main theorem is established, it is a significant advance: it gives the first rigorous proof of fast mixing for an efficiently implementable fermionic Gibbs sampler at arbitrary temperature, in a regime of weak interactions, and it yields polynomial-time quantum Gibbs-state preparation for a broad class of non-integrable fermionic lattice models. The paper is commendably explicit about the use of the external stability theorem [30], and it makes a serious effort to verify the hypotheses of that theorem rather than treating it as a black box. The third-quantization construction, the explicit free-parent gap computation, and the detailed Lieb-Robinson estimates are valuable technical contributions. No numerical data or code is expected in a proof paper of this type. The main risk is not the import of [30] itself but whether the paper's Propositions 1–3 really verify its hypotheses with all constants uniform in system size; I identified two specific gaps in those verifications.","major_comments":[{"comment":"The local approximant Z_1^loc(T) is defined using X_1(-βt'), which by Lemma 13 is a truncated version of γ^free_j(-βt'), not of γ^int_j(-βt'). But Z_1(T) contains the product γ^int_j(βt) γ^int_j(-βt'). In the proof, after 'These results allow us ...', the displayed triangle inequality is not a valid bound for AB - CD with A = γ^int_j(βt), B = γ^int_j(-βt'), C = ∫_0^{βt} X_2(s,t) ds, and D = X_1(-βt'), because B is not close to D: D approximates the free evolution, while B is the interacting correction and differs from D by an O(1) quantity. Therefore the claimed estimate ∥Z_1^loc(T) - Z_1(T)∥ ≤ C U (T+1)^{D+1}(r+1)^D e^{-r/2} is not established. This error propagates to Lemma 22, Corollary 10, Proposition 3, and hence to Theorem 3 and Theorem 1. A repair appears feasible: one can use the local approximant for γ^int provided by Corollary 9 for the second factor as well, obtaining an O(U^2 e^{-μr}) error, which is O(U e^{-μr}) for U below a constant. Please provide a corrected proof of this lemma.","section":"Section 8, Lemma 21 (Eqs. (118)–(119))"},{"comment":"The step Σ_j e^{-μ(d(j,k)+d(j,k'))} ≤ C'' Σ_j e^{-μ(d(k,k') + d(j,(k+k')/2))} ≤ C e^{-μ d(k,k')} is not justified. The first inequality is not a consequence of the triangle inequality, and the sum over the lattice can contribute a factor polynomial in d(k,k') (in a D-dimensional lattice, the number of sites at distance R grows polynomially). Since Proposition 2's [K,ν]-decay is a required hypothesis for applying Theorem 2, this estimate needs to be repaired. A standard fix is to prove the decay with a slightly smaller rate ν < μ so that the polynomial prefactor is absorbed, but as written the uniform-in-n claim does not follow.","section":"Section 7.2, Lemma 12 proof, Eq. (93)"}],"minor_comments":[{"comment":"The symbol Ξ_kk is never defined; since H^parent_{C,free} = 0 by Lemma 7, the terms L_{Ξ_kk} + R_{Ξ^†_kk} should be omitted or explicitly set to zero.","section":"Section 6, Eq. (69)"},{"comment":"The sentence 'Moreover, ∥\\tilde B_j∥ ≤ C' U' should presumably read '∥\\tilde B^int_j∥ ≤ C' U'; the full operator \\tilde B_j need not be small, and the subsequent use in Corollary 10 requires the bound on the interacting part.","section":"Lemma 22"},{"comment":"The formula \\tilde B^free_j = σ^{1/4} B^free_j σ^{-1/2} should read σ^{1/4} B^free_j σ^{-1/4}; the displayed exponent is a typo.","section":"Lemma 7 proof"},{"comment":"The estimate for ∥γ^int_j(t) - ∫_0^t X_2(s,t) ds∥ contains an unexplained factor e^{-βω/4}, although the left-hand side is independent of ω; this factor should appear only after integration against \\check F_1(t,ω).","section":"Lemma 15 proof"},{"comment":"The sentence beginning 'Then Xs(s,t) [YZ: two s here?]' contains an unfinished editorial comment and a typo; please clean it up.","section":"Lemma 14 proof"},{"comment":"'We can that H^parent_0 is quadratic' is missing the word 'see'; please correct.","section":"Section 5.1, after Eq. (52)"},{"comment":"Several places use d(j,k) interchangeably for distances between Majorana modes and between their lattice sites; a brief remark in the notation or a footnote would help the reader.","section":"General notation"}],"recommendation":"major_revision","confidential_remarks":"For the editor: the manuscript is a strong technical contribution, and the two proof gaps I identified are local and appear fixable rather than fatal. The reliance on [30] is legitimate, but the verification of its hypotheses must be complete. I recommend major revision rather than rejection: ask the authors to supply a corrected proof of Lemma 21 and a corrected decay estimate in Lemma 12, and to check that all downstream constants remain uniform in n."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Here's my take. The paper proves that for weakly interacting fermionic lattice Hamiltonians, the Lindbladian from Chen-Kastoryano-Gilyen has a spectral gap bounded below by a constant independent of system size, at any temperature, provided the interaction is below a threshold. That gives O(n + log(1/eps)) mixing time and an efficient Gibbs state preparation algorithm. If correct, this is the first such result for non-integrable fermions at low temperature, and it also gives exponential correlation decay. The architecture is coherent: third quantization extended to arbitrary Lindbladians, a parent Hamiltonian whose free part has an explicitly computed gap, and the interacting part shown to be quasi-local with (CU, mu)-decay. The handling of fermionic parity is genuinely careful, and the decision to restrict to the even sector is well justified by superselection.\n\nThe weakest link is the use of Hastings' stability theorem as a black box. The paper verifies the required decay conditions in Propositions 2 and 3, but the proof of Proposition 3 is a long chain of Lieb-Robinson estimates, and an error there could break n-independence. I did not find a specific mistake, but a referee should check that the constants C, mu, K, nu really do not acquire an n-dependence in the enlarged third-quantized space. The fact that each site carries four a-Majorana modes is handled by the geometry, and the support of local approximants is in the original lattice, so the formal application looks right.\n\nThere are some editorial remnants: an unresolved author query 'Xs(s,t) [YZ: two s here?]', a broken sentence after Lemma 19 ('In Lemma 7. Therefore...'), and several 'straightforward' calculations. These don't threaten the result but should be cleaned up.\n\nOverall, this deserves a serious referee. The main theorem is important and plausible; the proof is long but structured. The paper is for quantum information theorists working on Gibbs state preparation and open quantum systems. I would cite it and bring it to a reading group. My verdict is accept-with-revisions, with extra scrutiny on the stability theorem application.","headline":"First rigorous efficient mixing guarantee for fermionic Gibbs samplers at arbitrary temperature, with a sound but dense proof that hinges on a black-box stability theorem.","tokens_in":39242,"tokens_out":2872,"would_cite":true,"duration_ms":30492,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["81P68","82C10","81V70"],"pacs":["03.67.Ac","05.30.-d","71.10.Fd"],"model":"deepseek-v4-flash","headline":"Weakly interacting fermionic Gibbs states can be prepared efficiently at any temperature because the Lindbladian mixing time is linear in the system size.","keywords":["mixing time","Lindbladian spectral gap","Gibbs state preparation","fermionic lattice Hamiltonians","third quantization","weak interaction","quantum Gibbs sampler","Fermi-Hubbard model"],"falsifier":"Take the 2D Fermi-Hubbard model at a fixed small U below the claimed threshold and fixed beta, and compute the spectral gap of the Lindbladian (10) restricted to even-parity observables for increasing system sizes n; if the gap decays with n instead of approaching a positive constant, Theorem 1 is false.","tokens_in":38178,"feed_emoji":"⚛️","tokens_out":4608,"duration_ms":44277,"temperature":0.7,"pith_summary":"This paper aims to prove that Gibbs states of weakly interacting fermionic lattice Hamiltonians can be prepared efficiently on a quantum computer at any temperature. It shows that a recently proposed Lindbladian, when restricted to even-parity observables, has a spectral gap bounded below by a constant independent of system size, provided the interaction strength is below a constant threshold. From this gap, the epsilon-mixing time is at most linear in the system size, giving a polynomial-time quantum Gibbs state preparation algorithm. The proof works by mapping the Lindbladian to a parent Hamiltonian via third quantization and treating the interaction as a quasi-local perturbation of a gapped free-fermion Hamiltonian.","feed_headline":"Fermion Gibbs states mix fast at any temperature","feed_subtitle":"A constant spectral gap for weakly interacting fermionic Lindbladians makes quantum Gibbs state preparation polynomial-time.","key_machinery":"The central object is the parent Hamiltonian H_parent = \\tilde\\Phi(L^\\dagger), constructed by mapping super-operators to operators on a doubled fermionic Hilbert space whose particles are called a-fermions. The third quantization map \\Phi preserves locality and, crucially, preserves the spectrum on the even-parity sector, so a gap lower bound on H_parent transfers to the Lindbladian restricted to even-parity observables. The free part H0_parent decouples into single-mode terms whose spectral gap can be computed exactly, while the interacting part V_parent is shown to have (CU,mu)-decay using Lieb-Robinson bounds. A stability theorem for free-fermion Hamiltonians then converts these structural facts into a constant gap lower bound.","core_discovery":"For any inverse temperature beta>0 there exists a threshold U_beta such that whenever the interaction strength U is below it, the Lindbladian in Eq. (10) has a unique stationary state in the fermionic even-parity sector and mixes in time O(n + log(1/epsilon)). The key structural result is that the parent Hamiltonian splits as H0_parent + V_parent, where H0_parent is a quadratic free-fermion Hamiltonian with a constant spectral gap and exponentially decaying coefficients, while V_parent is a sum of quasi-local terms whose strength is O(U). Applying a stability theorem for free-fermion Hamiltonians then shows that the gap of the parent Hamiltonian, and hence the Lindbladian gap restricted to even parity, remains bounded below by a constant independent of n.","pith_inferences":["If the theorem is correct, weak-interaction fermionic Gibbs states form a concrete family where a provable quantum speedup over rigorous classical algorithms might be demonstrated, since no polynomial-time classical algorithm with arbitrary precision is known in this regime.","The interaction threshold U_beta likely shrinks as beta grows, because the stability theorem's conditions depend on the unperturbed gap of the free parent Hamiltonian, which involves factors like e^{-O(\\beta^2 \\epsilon^2)}; mapping this dependence numerically would delineate the practical regime of the algorithm.","The restriction to even-parity observables is essential in the third-quantization construction; extending the argument to odd-parity observables or to Lindbladians without detailed balance would require a different mechanism than the spectrum-preserving map used here."],"forward_implications":["The Gibbs state e^{-\\beta H}/Z of any weakly interacting fermionic lattice Hamiltonian covered by the theorem can be prepared to trace distance epsilon in time O(n^3\\,\\mathrm{polylog}(n/\\epsilon)).","The result applies directly to the Fermi-Hubbard model on a D-dimensional lattice whenever the on-site interaction strength |U| is below a constant depending only on the lattice geometry, the chemical potential, and the inverse temperature.","The Lindbladian dynamics converges exponentially fast in the KMS norm to the unique stationary state for all even-parity initial states obeying the fermionic superselection rule.","The same spectral gap implies exponential decay of correlations in the Gibbs state at any temperature for sufficiently weak interactions, as stated in Corollary 3."],"supporting_citations":[{"why":"Supplies the Lindbladian in Eq. (10), the KMS detailed balance condition, and the Gibbs state preparation algorithm whose mixing time is the subject of this paper.","marker":"[10]"},{"why":"Establishes the parent-Hamiltonian approach for bounding Lindbladian mixing times, which this paper adapts to fermionic systems.","marker":"[27]"},{"why":"Provides the stability theorem for free-fermion Hamiltonians used to lower bound the spectral gap of the parent Hamiltonian from its free part and quasi-local perturbation.","marker":"[30]"},{"why":"Introduces third quantization, the method used here to map fermionic Lindbladians to operators in an enlarged Hilbert space while controlling locality and spectrum.","marker":"[31]"},{"why":"Supplies the Lieb-Robinson bound used throughout to establish quasi-locality of jump operators and of the interacting part of the parent Hamiltonian.","marker":"[33]"},{"why":"Provides the Hamiltonian simulation algorithm used to convert the mixing time bound into the O(n^3 polylog(n/epsilon)) gate complexity for Gibbs state preparation.","marker":"[35]"}],"fun_headline_variants":["Fermion mixing time at most linear at any temp","Constant spectral gap for weakly interacting fermions","Any-temperature fast fermion Gibbs state prep","Weak interactions ensure linear mixing for fermions","Fermionic Lindbladians mix fast even at high temperature"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole gap bound rests on the imported stability theorem for free-fermion Hamiltonians, which requires the perturbation V_parent to have exponentially decaying local terms with strength below a constant determined by the unperturbed gap; if that theorem does not apply to the constructed a-fermion geometry, the main result collapses.","fun_headline_variants_meta":{"raw":{"variants":["Fermion mixing time at most linear at any temp","Constant spectral gap for weakly interacting fermions","Any-temperature fast fermion Gibbs state prep","Weak interactions ensure linear mixing for fermions","Fermionic Lindbladians mix fast even at high temperature"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000133,"raw_usage":{"total_tokens":1062,"prompt_tokens":796,"completion_tokens":266,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":412,"completion_tokens_details":{"reasoning_tokens":193}},"tokens_in":412,"tokens_out":266,"duration_ms":3745,"temperature":1.0,"reasoning_tokens":193,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T22:51:05.366074+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take the 2D Fermi-Hubbard model at a fixed small U below the claimed threshold and fixed beta, and compute the spectral gap of the Lindbladian (10) restricted to even-parity observables for increasing system sizes n; if the gap decays with n instead of approaching a positive constant, Theorem 1 is false.","supporting_citations":[{"cited_title":"The stability of free fermi Hamiltonians,","cited_arxiv_id":null,"evidence_quote":"Provides the stability theorem for free-fermion Hamiltonians used to lower bound the spectral gap of the parent Hamiltonian from its free part and quasi-local perturbation."},{"cited_title":"Third quantization: a general method to solve master equations for quadratic open fermi systems,","cited_arxiv_id":null,"evidence_quote":"Introduces third quantization, the method used here to map fermionic Lindbladians to operators in an enlarged Hilbert space while controlling locality and spectrum."},{"cited_title":"The finite group velocity of quantum spin systems,","cited_arxiv_id":null,"evidence_quote":"Supplies the Lieb-Robinson bound used throughout to establish quasi-locality of jump operators and of the interacting part of the parent Hamiltonian."}],"review_version":1}