{"id":"47e584a1-8b98-4155-8ca2-e18c632e98fe","arxiv_id":"2511.05490","paper_version":4,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"Exact strong zero modes exist and are boundary-localized in integrable circuits and XXZ chains with non-diagonal (U(1)-breaking) open boundary conditions.","lead":"This paper constructs exact conserved \"boundary clock\" operators (exact strong zero modes) for integrable quantum circuits and the spin-1/2 XXZ chain with generic end fields that break spin-rotation symmetry. It shows these operators are pinned to one boundary, make certain edge spins keep infinite-time memory, and become non-local under the map to a classical exclusion process.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Coherence-time claim depends on unverified decay of the ESZM-orthogonal part C_{O'}(t); only σ^z_1 at N=6,8,10 is checked, leaving Eq. (45) plateaus for σ^x_1σ^z_2 and σ^y_1σ^z_2 untested.","rationale":"The paper's main algebraic result — an exactly conserved, exponentially localized operator for non-diagonal integrable boundaries — is well supported by the transfer-matrix construction and MPO norm computations; I see no reason to doubt [Ψ,H]=0. The weakest link is the step from an exact conserved operator to 'infinite boundary coherence times'. That step requires the non-conserved remainder in Eq. (42) to have a vanishing late-time value in the thermodynamic limit. This is an expectation, not a theorem, and it is not implied by the existence of Ψ. In integrable chains, conserved charges can slow or prevent relaxation of local observables, and finite-size diagonal averages for operators like σ^z_1 are O(1/N), not exponentially small; the paper does not quantify this. The numerical check in Fig. 5 is limited to one operator and three small sizes. The proposed ED test for the σ^x/σ^y_1σ^z_2 operators at larger N directly checks whether Eq. (45) is the correct plateau — exactly the advertised quantity. I do not regard this as fatal to the construction, but the physical claim should be conditional on this check. Since the reader already assigned CONDITIONAL and this is the same concern, no verdict adjustment is needed.","tokens_in":17606,"tokens_out":17658,"duration_ms":153191,"concrete_test":"Perform exact diagonalization for N=10, 12, and 14 (16 if feasible) at the Fig. 5 parameters ∆=2.5, h1=(1,0.1,0), hN=(0.25,0.5,1). Compute infinite-temperature autocorrelators C_{σ^z_1}(t), C_{σ^x_1σ^z_2}(t), and C_{σ^y_1σ^z_2}(t). Evaluate the late-time plateau (or the long-time average after the initial transient) and compare with |cΨ1|² and the two expressions in Eq. (45). Require the difference to decrease with N (e.g., scale as 1/N or faster); if it remains O(1), the assumed decay of C_{O'} fails and the coherence plateau is not determined by the ESZM overlap alone.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The existence of an exactly conserved, exponentially localized Ψ is well supported by the MPO construction and Hilbert–Schmidt norm calculations, so I do not question the algebraic claim. The advertised physical consequence — infinite boundary coherence times — requires more than [Ψ,H]=0. In Eq. (42) the paper decomposes O = c1 Ψ + c2 O' and then asserts (text after Eq. (42)) that C_{O'}(t) 'is expected to decay in time to a value that vanishes as N→∞'. This is a separate dynamical assumption, not a consequence of the exact conservation of Ψ. In an integrable model with many conserved charges, O' can have overlap with other conserved or nearly conserved operators; even for extensive charges those overlaps are O(1/N), and finite-N diagonal contributions of σ^z-like operators are not exponentially small. The paper verifies the plateau only for σ^z_1 at N=6,8,10 (Fig. 5); the other two operators in Eq. (44) are not checked. If the late-time plateau for σ^x_1σ^z_2 or σ^y_1σ^z_2 differs from the |c1|² values in Eq. (45) by an amount that does not vanish with N, the central 'infinite coherence time' claim is quantitatively wrong even though Ψ exists. This is the load-bearing soft spot.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper constructs exact strong zero mode operators Ψ for an integrable brick-wall quantum circuit and, in the Trotter limit, for the spin-1/2 XXZ chain with open boundary terms, under the condition that the left boundary field lies in the x-y plane (h1^z=0) while the right boundary is general. Ψ is built from the derivative of the Sklyanin-type transfer matrix at u*=iπ/2, with the left boundary parameter fixed to ξ^(L)=iπ/2. The authors give an MPO representation (Appendices A–C), prove exponential localization in Hilbert–Schmidt norm with closed-form decay rate d6 (Eqs. (32)–(33)), and verify this numerically. They then argue that the ESZM produces non-decaying infinite-temperature autocorrelation plateaus for boundary observables σ^z_1, σ^x_1σ^z_2, σ^y_1σ^z_2 (Eqs. (42)–(45)), with numerical support shown for σ^z_1. Finally, under the similarity transformation to the asymmetric simple exclusion process, the same operator loses its boundary-localized structure, and they conclude it is not dynamically significant in ASEP.","tokens_in":17959,"tokens_out":14481,"duration_ms":125525,"significance":"The paper's main algebraic claim is well supported and technically impressive: exact ESZM operators are constructed in explicit MPO form for both a Floquet circuit and the XXZ chain with non-diagonal boundaries, the localization length is given by a closed-form d6, and the normalization is controlled in finite volume (Appendices A–C). The numerical data in Fig. 4 confirm the exponential localization. The discussion of the ASEP map is a useful negative result: the ESZM delocalizes under the transformation, so its dynamical role in ASEP is not inherited. If the coherence-time prediction can be placed on a firmer footing, this will be a solid contribution to the strong-zero-mode literature.","major_comments":[{"comment":"The 'infinite boundary coherence times' claim rests on the assertion that C_{O'}(t) 'is expected to decay in time to a value that vanishes as N→∞'. This is not a consequence of [Ψ,H]=0: in an integrable chain O' can overlap with other conserved operators (the transfer-matrix charges Q^(n), and possibly quasilocal conserved operators), and those overlaps control the infinite-time average of C_{O'}. The numerical test in Fig. 5 covers only σ^z_1 for N=6,8,10 and one set of boundary fields; the plateaus predicted for σ^x_1σ^z_2 and σ^y_1σ^z_2 in Eqs. (44)–(45) are not checked. Please either prove the absence of such overlaps or provide data for all three operators at several N demonstrating that the late-time values approach (45). As written the physical conclusion is stronger than the evidence.","section":"§3.1, after Eq. (42)"},{"comment":"The construction of Ψ is not fully self-contained. The key identity (18), the choice u*=iπ/2, the condition ξ^(L)=iπ/2, and the MPO representation are taken from Refs. [26,49], both 'to appear'. Since [Ψ,H]=0 is the basis of the paper, the authors should either derive these steps in the main text/appendix or state precisely which propositions from the forthcoming papers are used and why they apply here. The explicit matrix elements in Appendix A are helpful but do not by themselves show that the operator obtained from (24) commutes with U.","section":"§2.2/Eq. (24)"}],"minor_comments":[{"comment":"The text says the autocorrelation is shown for δ=2.5, but the XXZ model is defined after taking δ→0; the figure caption says ∆=2.5. Please correct the typo.","section":"§3.1, Fig. 5"},{"comment":"'General open boundary conditions' is broader than the proven statement; the left boundary field is required to satisfy h1^z=0 (Eq. (3)). Suggest rewording to avoid overclaim.","section":"Abstract/§1"},{"comment":"The normalization expression has notational artifacts (e.g. λ_L^4, λ_L^6, dangling factors) that make it hard to read; please typeset carefully and check the formula.","section":"Appendix B"},{"comment":"Please state explicitly that each Ψ_j carries a non-identity operator on site j (r_{2j-1}, r_{2j} ∈ {x,y,z}); this justifies the contributions to Eq. (45) and aids the reader.","section":"Eq. (26)"}],"recommendation":"major_revision","confidential_remarks":"The main concern is the gap between the algebraic result and the advertised dynamical consequence. I would ask the authors to strengthen §3.1 before publication. Also, because Refs. [26,49] are forthcoming, the editor may want to verify that those papers are available or that the construction here is sufficiently self-contained."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nThe headline: this is the first construction of an exact strong zero mode operator for an integrable circuit and the XXZ chain when the boundary fields break the bulk U(1) symmetry. The MPO construction is explicit, the Hilbert–Schmidt norms decay with a closed-form rate (Eq. 32), and the numerics in Fig. 4 match. The ASEP result — that the ESZM becomes nonlocal under the mapping — is a nice addition and seems correctly derived.\n\nWhat I actually trust: the algebraic existence and localization of Ψ. The transfer-matrix argument is spelled out in enough detail in the appendices that an independent check is feasible. The normalization constants are given, and the decay rate d6 < 1 is computed. That part holds up.\n\nThe soft spot is the step from [Ψ,H]=0 to 'infinite boundary coherence times.' After Eq. (42) they decompose O = c1 Ψ + c2 O′ and assert that C_{O′}(t) decays to a value that vanishes as N→∞. That is an extra dynamical assumption, not a consequence of the conservation law. In an integrable chain with many conserved charges it is plausible — the standard GGE argument gives a 1/N tail for a generic local operator — but they only check it for σ^z_1 at N=6,8,10. The other two operators in Eq. (45) are left untested. I would not call this fatal, because the argument is clear and the numerics for σ^z_1 is consistent, but a referee should push them to either prove the decay or show a scaling collapse for at least one of the σ^x_1σ^z_2 / σ^y_1σ^z_2 autocorrelations.\n\nTwo other complaints, both minor. The δ→0 Trotter limit is asserted rather than demonstrated; I expect it's fine, but the paper should say more. And the construction leans heavily on Refs [26,49], both 'to appear' from the same group. The paper is reasonably self-contained because the appendices give the MPO elements, but those to-appear papers should be made available.\n\nOn the 'ESZM' label: they don't check Ψ² ∝ 1, which is in the standard SZM definition. For an exactly conserved, localized operator this may be a matter of taste, but it's worth flagging.\n\nBottom line: the algebraic construction is a real advance and I'd be happy to see it published after the coherence-time claim gets an added argument or more numerics. I'd cite it as the first ESZM with U(1)-breaking boundaries, and I'd send it to a serious referee rather than desk reject.\n\nBest,","headline":"First exact strong zero mode for U(1)-breaking integrable boundaries; algebraic core is solid, coherence-time claim needs one more piece of evidence.","tokens_in":18444,"tokens_out":5040,"would_cite":true,"duration_ms":43509,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["82B20","82B23","81R12"],"pacs":[],"model":"deepseek-v4-flash","headline":"Exactly conserved boundary operators exist in the integrable brick-wall circuit and the XXZ spin chain even when boundary fields break the bulk U(1) symmetry, provided the left field lies in the x-y plane.","keywords":["exact strong zero modes","open boundary conditions","XXZ spin chain","integrable quantum circuits","edge coherence times","matrix product operators","asymmetric simple exclusion process","U(1) symmetry breaking"],"falsifier":"Compute the infinite-temperature autocorrelation C_{σ^z_1}(t) for N=14,16,18 in the XXZ chain with h_1=(1,0.1,0), h_N=(0.25,0.5,1), and check whether the late-time plateau approaches |c_Ψ1|² as the system grows; alternatively, compute the autocorrelation of O' = σ^z_1 − c_1 Ψ directly and test whether its late-time value vanishes with N.","tokens_in":17492,"feed_emoji":"🧲","tokens_out":6968,"duration_ms":57916,"temperature":0.7,"pith_summary":"This paper proves that exact strong zero modes—operators that commute exactly with the time evolution and are exponentially localised near a boundary—exist in the integrable brick-wall quantum circuit and the spin-1/2 XXZ chain with the most general integrable open boundary conditions, provided the left boundary magnetic field lies in the x-y plane. This removes a long-standing restriction: such modes were previously known to exist only when the boundary respected a global Z2 or U(1) symmetry. The exact zero mode forces infinite-temperature autocorrelations of boundary operators such as σ^z_1 to plateau at a finite value set by their overlap with the mode, signalling infinite boundary coherence times. The authors also show that under the mapping to the asymmetric simple exclusion process the mode loses its spatial locality, so it cannot shape the stochastic dynamics of that process.","feed_headline":"Exact edge mode exists even when U(1) breaks","feed_subtitle":"For the XXZ chain and its Trotter circuit, a left field in the x-y plane is enough for infinite boundary coherence.","key_machinery":"The central object is the double-row transfer matrix T(u) of the six-vertex model with Sklyanin's open-boundary K-matrices, which generates the conserved charges of both the circuit and the Hamiltonian. Evaluating T at the special point u*=iπ/2 makes the right boundary operator proportional to σ^z, and fixing the left boundary parameter ξ^(L)=iπ/2 makes the resulting conserved operator exponentially localised at that edge. The operator is written as a matrix-product operator (MPO) with matrices A±; the transfer matrix à of the MPO norm turns out to be independent of the ± labels, and its largest eigenvalue d6<1, given explicitly, controls the exponential decay of the Hilbert–Schmidt norms an","core_discovery":"The central claim is that coexistence of an exactly conserved boundary operator with non-diagonal boundary terms is possible. For the XXZ Hamiltonian with boundary fields, an ESZM Ψ with [Ψ,H]=0 exists precisely when the left boundary field has no z-component, h^z_1=0; the right boundary field may be arbitrary. The same statement holds for the stroboscopic evolution of the integrable brick-wall circuit obtained by Trotterising the XXZ chain. The operator is constructed from the double-row transfer matrix at the special value u*=iπ/2, where the right boundary K-matrix collapses to σ^z, and with the left boundary parameter fixed to ξ^(L)=iπ/2 to guarantee exponential localisation. The paper gi","pith_inferences":["The explicit decay rate d6(η,δ) gives an analytic handle on the edge-mode localisation length; one testable extension is to compare it with the correlation length of the bulk XXZ chain and check whether the localisation length diverges at the isotropic point.","The result that h^z_1=0 is sufficient suggests a broader design principle: an ESZM survives any boundary perturbation that preserves a single discrete symmetry at the edge, even if it breaks all continuous symmetries of the bulk.","The ASEP non-locality result indicates that exact boundary coherence is coordinate-dependent; for dissipative or stochastic embeddings of integrable models, one should expect similar washing-out of the edge mode.","The plateau prediction relies on the decay of the orthogonal part of the autocorrelation, which the paper checks only for small N; a Bethe-ansatz or transfer-matrix computation of the full autocorrelation would confirm the plateau for larger systems."],"forward_implications":["For any left boundary field in the x-y plane and arbitrary right boundary field, the XXZ chain and its Trotter circuit carry an exactly conserved operator localised at the left edge, so σ^z_1, σ^x_1σ^z_2, and σ^y_1σ^z_2 autocorrelations saturate at values set by c_Ψ1 and c_Ψ2.","The brick-wall circuit requires only local gates, so the h^z_1=0 condition is straightforward to implement on a quantum simulator; the predicted plateau in edge autocorrelations is directly measurable.","Under the similarity transformation to the asymmetric simple exclusion process, the ESZM's Hilbert–Schmidt weight at any fixed boundary site vanishes as N→∞, so the process does not inherit boundary coherence from the spin chain.","Because the zero mode is exactly conserved for every finite N, there is no finite-size crossover: the plateau value is exact as soon as the boundary geometry is fixed."],"fun_headline_variants":["Exact edge mode survives non-diagonal boundaries in XXZ chain","Zero z-field on left boundary yields exact conserved edge mode","Infinite boundary coherence from exact zero mode despite broken U(1)","Non-diagonal boundary still permits exact local conserved operator"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The 'infinite boundary coherence' claim rests on the expectation that the autocorrelation of the part of a boundary operator orthogonal to the zero mode decays to a value vanishing with system size—verified numerically only for σ^z_1 at N=6,8,10.","fun_headline_variants_meta":{"raw":{"variants":["Exact edge mode survives non-diagonal boundaries in XXZ chain","Zero z-field on left boundary yields exact conserved edge mode","Infinite boundary coherence from exact zero mode despite broken U(1)","Non-diagonal boundary still permits exact local conserved operator"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000551,"raw_usage":{"total_tokens":2412,"prompt_tokens":637,"completion_tokens":1775,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":381,"completion_tokens_details":{"reasoning_tokens":1705}},"tokens_in":381,"tokens_out":1775,"duration_ms":12102,"temperature":1.0,"reasoning_tokens":1705,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-03T23:28:03.866573+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the infinite-temperature autocorrelation C_{σ^z_1}(t) for N=14,16,18 in the XXZ chain with h_1=(1,0.1,0), h_N=(0.25,0.5,1), and check whether the late-time plateau approaches |c_Ψ1|² as the system grows; alternatively, compute the autocorrelation of O' = σ^z_1 − c_1 Ψ directly and test whether its late-time value vanishes with N.","supporting_citations":[],"review_version":1}