{"id":"bbb2ca5d-5335-48ce-85ec-e7abef1888f2","arxiv_id":"2512.07742","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Exact strong zero modes exist for integrable spin-S XXZ chains with open boundaries; for integer S they localize in a weak Hilbert-Schmidt sense, not as sharp edge operators.","lead":"The authors construct exact operators that commute with the Hamiltonian of integrable higher-spin spin chains and act like edge zero modes. These operators are less sharply localized than in the spin-1/2 case, but they still force edge autocorrelations to keep a finite value forever, meaning very long edge coherence.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The normalizability and near-inversion of Ψ rest on an unproved spectral-gap assertion about U and V in §4.2.1; if a subleading eigenvalue has modulus 1, Eqs. (93),(87) and the edge-plateau argument (102) fail.","rationale":"The paper's central algebraic claim — that T^{(S,1/2)}'(iπ/2) provides an exact ESZM for all S — is well separated into an exact commutation step and an analytic/norm step. The commutation step is solid because transfer matrices commute with the Hamiltonian by construction. The norm step is where the argument depends on the spectral properties of U and V. The reader identified this as the weakest assumption, and I agree: no proof is supplied for the 'unique eigenvalue 1, all others strictly smaller in modulus' statement, despite it being used to conclude exponential convergence of ||Ψ||² and ||Ψ²−1||². The S=1 issue of additional ESZMs is real but secondary: the constructed Ψ already gives a nonzero lower bound |c1|² for the plateau, so the existence of at least some non-decaying component is not at stake; only the full plateau value is unexplained. The S=3/2 ESZM conjecture and Bethe-state eigenvalue conjecture are explicitly flagged as conjectural by the authors and do not affect the core S=1/2 and general-S construction. Thus the most load-bearing, least secure point is the spectral gap of U and V. A direct numerical check for several S and η would increase confidence, while a Perron–Frobenius argument would provide the missing proof. Because this is a proof gap rather than a demonstrated contradiction, the CONDITIONAL verdict should stand unchanged.","tokens_in":39011,"tokens_out":5391,"duration_ms":56954,"concrete_test":"Independently diagonalize U and V constructed from Eqs. (81),(88) for S=1/2,1,3/2,2 and several η>0 (e.g. 0.1, 0.5, 1, 2), verifying that the only eigenvalue on the unit circle is 1 and that the spectral gap γ=−log(max_{λ≠1}|λ|) is positive. To settle 'for any S', seek a similarity transformation making U (and V) nonnegative and apply Perron–Frobenius, using the explicit eigenvectors in (91) as a guide; if such a gauge exists, it yields a uniform proof. If any subleading eigenvalue with |λ|=1 appears, Eqs. (93),(87) and the plateau argument (102) must be revised.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The construction's two key asymptotic properties — ||Ψ||²=1+O(e^{-γL}) (Eq. 93) and ||Ψ²−1||²=O(e^{-γL}) (Eq. 87) — both follow from the unproved assertion in §4.2.1 that the auxiliary transfer matrix U (and likewise V) has a unique eigenvalue equal to 1 and all other eigenvalues strictly smaller in modulus for every S and real η. The paper states this was 'verified by direct calculation' but gives no derivation or explicit computation. This is not a cosmetic omission: if U has another eigenvalue on the unit circle, the exponentially small corrections become O(1), and the normalizability/inversion properties that define the ESZM fail; Eq. (102), which converts the ESZM overlap into a nonzero autocorrelation plateau, would then not follow. Because this spectral gap is the bridge from the algebraic commutant construction to the claimed infinite coherence time physics, it is load-bearing. The S=1 mismatch between |c1|²=0.264 and the observed plateaus is important, but it concerns completeness of the plateau-value explanation; the spectral assertion is needed even for the existence of a single conserved edge mode.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper constructs, for integrable open spin-S XXZ chains with a Z2-invariant left boundary and generic right boundary, an operator Ψ obtained from the derivative of the transfer matrix T^{(S,1/2)}(u) at u=iπ/2. The authors show that Ψ commutes exactly with the Hamiltonian, has a Hilbert-Schmidt norm equal to 1 up to exponentially small corrections, nearly squares to the identity in the same norm, and is weakly localized near the left edge. They argue that these properties imply finite late-time plateaus in edge autocorrelation functions, present numerical evidence for S=1/2 and S=1, and discuss connections to boundary bound states in the Bethe ansatz for S=1/2.","tokens_in":39421,"tokens_out":2386,"duration_ms":25805,"significance":"If the construction is correct, it provides exact strong zero modes for arbitrary spin S in an integrable setting, extending the earlier MPO/transfer-matrix construction for S=1/2. The exact commutation with the Hamiltonian is a strong structural result, and the operator-norm formulation of locality is a useful generalization. The paper also gives a clear physical interpretation of the degenerate ground states as a first-order transition line and presents reproducible numerics for edge autocorrelators. The weakening of locality for S≥1 and the explanation in terms of an odd number of ground states for integer S is an interesting conceptual contribution. However, the central claims rely on a spectral-gap assertion that is stated but not proved, and for S=1 the constructed operator does not by itself explain the numerically observed plateau heights.","major_comments":[{"comment":"The normalizability and near-inversion properties of Ψ rest on the assertion, made after Eq. (91), that the ancillary transfer matrix U (and similarly V in Eq. (95)) has a unique eigenvalue equal to 1 and all other eigenvalues strictly smaller in modulus for all S and η∈R. The text says this was 'verified by direct calculation', but no derivation or explicit computation is given. This is load-bearing: Eqs. (93) and (87), and hence the plateau argument leading to Eq. (102), all follow from the exponential decay of subleading eigenvalues. If a subleading eigenvalue had modulus 1, the exponentially small corrections would become O(1) and the ESZM would lose its normalizability and inversion properties. Please provide a proof, or at least a complete and reproducible algebraic verification, including a bound on the second-largest eigenvalue.","section":"§4.2.1, Eqs. (91)–(96)"},{"comment":"For S=1 the overlap of s^z_1 with the constructed Ψ is |c1|²=0.2637 at η=1, whereas the numerical plateau in Fig. 11 is substantially larger. The paper correctly concludes that 'there must be additional ESZM operators', but none are constructed. This is not a cosmetic gap: the abstract and conclusion claim that the constructed ESZMs imply the observed infinite coherence times, but for S=1 the single explicitly constructed operator cannot account for the plateau. The argument needs either an explicit additional conserved edge operator, or a more careful statement that the plateau demonstrates the existence of additional ESZMs whose construction is left open.","section":"§4.2.2, Fig. 11 and Eq. (103)"},{"comment":"The claimed S=3/2 ESZM Ψ_{3/2}=T^{(3/2,3/2)}(iπ/2) is a conjecture: only the first two terms of its large-η expansion are matched with the perturbative construction. Since the paper uses this to argue that a 'conventional' ESZM exists for S=3/2, this part is not established. This does not affect the main T^{(S,1/2)} construction, but the text should clearly separate conjecture from proof, especially in the Conclusions where it is used as evidence.","section":"§4.4, Eq. (116)"}],"minor_comments":[{"comment":"There are several typographical issues: 'spin-Schains' in the title line, 'cf.section' in §3.1, 'withre' in §4.2, and missing spaces around parentheses. These should be corrected.","section":"Throughout"},{"comment":"The conjecture that the ESZM eigenvalue is +1 without a right-boundary string and −1 with one is supported by numerical results for L≤12, N≤6 in Appendix D. This is reasonable evidence, but the statement should explicitly note the limited system sizes, as the text already does implicitly in Table 1.","section":"§5, Eqs. (126)–(127)"},{"comment":"The lower bound in Eq. (102) assumes that the only conserved edge-localized operators are the ESZMs. For S=1 the paper itself shows there must be more, so the inequality is not tight. This should be stated directly where the equation appears, not only later.","section":"§4.2.2, Eq. (102)"},{"comment":"The construction relies on the companion papers [39,45,74], two of which are preprints. If the journal allows, it would help readers if the key MPO identities from [45] were summarized in an appendix, since the present paper's normalizability proof depends on them.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"The central transfer-matrix construction is elegant and the exact commutation with the Hamiltonian is solid. The main risk is the unproved spectral-gap assertion for U and V; I would want a proof or an explicit finite-matrix computation before accepting the claim of exponentially small corrections. The S=1 plateau mismatch is also a substantive completeness issue, not a presentation detail, so major revision is appropriate."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The core of this paper is a transfer-matrix construction of exact strong zero modes for the integrable spin-S XXZ chain with open boundaries. That construction is genuinely new beyond S=1/2, and it recovers the known spin-1/2 results as a special case. The paper is also honest about what it does not prove: the S=3/2 mode is conjectural, the Bethe-state eigenvalue statement is numerically supported but not proven, and for S=1 the constructed operator cannot account for the observed plateau heights, forcing the authors to invoke additional, unconstructed ESZMs. None of that kills the central algebraic claim, but the gaps are real.\n\nThe strongest part is the MPO formulation and the identification of a weak-locality notion that is natural for integer spin, where an odd number of ground states makes the usual edge-locality impossible. The exact commutation [H,Ψ]=0 is structural and unlikely to be wrong. The overlap calculations are explicit, not fitted, and the S=1/2 limit checks out. I also like the contact with boundary strings: the eigenvalue of Ψ on Bethe states apparently toggles between +1 and -1 depending on which boundary string is present, which ties the abstract construction to the Bethe-ansatz picture.\n\nThe soft spot that matters most is Section 4.2.1. The normalizability and near-inversion of Ψ rest entirely on the claim that the auxiliary transfer matrix U has a unique eigenvalue equal to 1 and all others strictly smaller in modulus, for every S and real η. The paper says this was 'verified by direct calculation' but gives no derivation, no explicit computation, and no reference. That is load-bearing: if a subleading eigenvalue also has modulus 1, then the O(e^{-γL}) corrections in (93) and (87) become O(1), and the plateau argument (102) collapses. The rest of the paper cannot be evaluated until this assertion is either proved or at least backed by a systematic numerical study over a wide range of S and η.\n\nThe S=1 plateau mismatch is a separate, less severe issue: the paper's own lower bound (102) is valid, but the claim that the constructed Ψ explains the observed edge autocorrelators is incomplete. The authors say so themselves, but that means the main physical payoff for integer spin still depends on an existence statement about additional ESZMs that is not demonstrated. The paper is transparent about this, which I credit.\n\nBottom line: this is a serious paper with a substantial new construction, and the identified gaps are concrete and addressable rather than fundamental. I would send it to a referee who knows transfer matrices and SZM theory, and would ask that referee to focus on the spectral assertion. If that gap closes, the paper is a strong contribution. If it does not, the construction is still worthwhile but the claims about infinite coherence times need to be scaled back.","headline":"Solid arbitrary-spin ESZM construction with a real weak-locality story; the main gap is an unproved spectral assertion on the auxiliary transfer matrix.","tokens_in":39848,"tokens_out":1852,"would_cite":true,"duration_ms":21263,"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":"This paper derives exact edge operators that commute with the Hamiltonian of any integrable spin-S chain, guaranteeing infinitely long edge coherence times even where locality is weak.","keywords":["strong zero modes","integrable spin chains","XXZ chain","matrix product operators","transfer matrices","edge coherence","Bethe ansatz","boundary strings"],"falsifier":"Compute the spectrum of the four-dimensional transfer matrix U (or V) for a specific S and real η and find an eigenvalue of modulus greater than or equal to 1 other than the Perron eigenvalue; then the claimed exponential decay of ∥Ψ_j∥ and the limit ∥Ψ²−1∥²→0 fail. Alternatively, for the spin-1 chain with boundary conditions of Eq. (59), compute the edge autocorrelator plateau at larger L and show it decays to a value below |c_1|²+|c_2|², contradicting the claim that additional ESZMs exist.","tokens_in":38856,"feed_emoji":"🧲","tokens_out":4685,"duration_ms":38811,"temperature":0.7,"pith_summary":"The paper sets out to show that exact strong zero modes (ESZMs) — operators that commute exactly with the Hamiltonian and are localized near the edge — exist not just for the spin-1/2 XXZ chain but for every integrable spin-S XXZ chain with open boundaries and a boundary field. Because the integrable spin-S chain has 2S+1 degenerate ground states, and an odd number of them for integer S, the new ESZMs cannot have the same sharp locality or the property of squaring to the identity that the spin-1/2 mode has. Instead the authors prove these properties hold in a weaker, Hilbert-Schmidt sense, and argue that this is still enough to force edge autocorrelation functions to plateau at nonzero values — infinite edge coherence times. The construction uses the family of commuting transfer matrices of the model and gives explicit matrix-product forms for the modes. A sympathetic reader should care because this gives exact, parameter-free statements about long-lived edge physics in a class of strongly interacting many-body systems, and identifies why integer spin is structurally different.","feed_headline":"Exact edge modes found for every integrable spin-S chain","feed_subtitle":"Even where locality is weak, the new modes enforce infinite edge coherence times.","key_machinery":"The central object is the double-row transfer matrix T^{(S,1/2)}(u) built from quantum-group generators S^z, S^± and diagonal reflection matrices; it commutes with the Hamiltonian for all u. The special value u* = iπ/2 makes the L-operator satisfy L_j(u*) σ^z_a L_j(u*) = σ^z_a, which collapses the transfer matrix to zero and turns its derivative into a sum of local pieces Ψ_j. Writing Ψ as a matrix-product operator with a four-dimensional ancilla, the Hilbert-Schmidt norms are evaluated by a transfer matrix U = (1/(2S+1))Tr(A⊗A); the claim that U has a unique eigenvalue 1 with all others smaller in modulus is what makes the mode normalizable and almost idempotent.","core_discovery":"On the paper's own terms: for any spin S and real anisotropy η, the derivative of the auxiliary-spin-1/2 transfer matrix at the special value u* = iπ/2 yields a Hermitian operator Ψ that commutes exactly with the integrable open spin-S Hamiltonian. In the Hilbert-Schmidt norm, Ψ is normalized to 1 and squares to the identity up to corrections exponentially small in system size, and each local piece Ψ_j of its MPO decomposition decays exponentially in j; however, for S ≥ 1 the spectral norm of Ψ_j does not decay, so locality is strictly weaker than the spin-1/2 case. The authors explain this weakening as forced by the odd number of degenerate ground states for integer S, and show numerically","pith_inferences":["If the spectral-gap assertion holds, the construction likely extends by fusion to higher auxiliary spins, and the paper's conjectured S=3/2 ESZM from T^{(3/2,3/2)}(iπ/2) would give a conventional strongly localized mode; a direct check of its full MPO would settle that.","The 'weak locality but normalizable' pattern suggests a general principle: when the ground-state manifold has odd degeneracy, edge operators can still be exact conserved charges but must act nontrivially on an exponentially small (in Hilbert-Schmidt measure) fraction of the Hilbert space, so disorder-free prethermal-like plateaus survive.","The boundary-string characterization for S=1/2 suggests a dictionary between ESZM eigenvalues and root structure that, if extended to higher S, would let one compute exact edge-coherence plateaus directly from Bethe ansatz data without diagonalizing.","Perturbing away from integrability while keeping the U(1) symmetry, the plateau in edge autocorrelators appears robust at accessible sizes; a quantitative prediction is that the plateau height for large L is set by the sum of squares of overlaps with all ESZMs, which could be tested against exact numerics."],"forward_implications":["For every S, the open integrable spin-S XXZ chain with appropriate boundary fields has an exactly conserved edge operator, so local observables overlapping it show nonzero infinite-temperature autocorrelation plateaus.","The weaker locality for S≥1 is not a technical accident but forced by the odd number of degenerate ground states for integer S, so any attempt to construct a sharp SZM in those models must fail.","The SZM (non-exact) version is recovered when both boundaries are set to the integrable spin-flip-invariant value, with commutator exponentially small in Hilbert-Schmidt norm but not spectral norm.","In the spin-1/2 case, the ESZM's action on energy eigenstates is exactly characterized by the Bethe-ansatz boundary strings: eigenvalue -1 precisely on states containing the right-boundary string.","Numerical edge-autocorrelator plateaus for the spin-1 chain require additional ESZMs beyond the single constructed Ψ, indicating a family of conserved edge operators."],"fun_headline_variants":["Zero modes for spin-S chains: locality weaker, coherence infinite","Exact edge modes in all spin-S chains, even weak locality","Spin-S edge modes exact, but odd spins need weaker locality","Infinite coherence from exact zero modes in spin-S chains","For integrable spin-S: exact edge modes, weaker locality"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The entire normalizability and near-inversion of Ψ rests on an unproved spectral assertion: that the auxiliary transfer matrix U (and its fourth-power analogue V) has a unique eigenvalue of modulus 1 and a gap below it; the paper says this was 'verified by direct calculation' but gives no derivation.","fun_headline_variants_meta":{"raw":{"variants":["Zero modes for spin-S chains: locality weaker, coherence infinite","Exact edge modes in all spin-S chains, even weak locality","Spin-S edge modes exact, but odd spins need weaker locality","Infinite coherence from exact zero modes in spin-S chains","For integrable spin-S: exact edge modes, weaker locality"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000654,"raw_usage":{"total_tokens":2772,"prompt_tokens":625,"completion_tokens":2147,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":369,"completion_tokens_details":{"reasoning_tokens":2061}},"tokens_in":369,"tokens_out":2147,"duration_ms":13617,"temperature":1.0,"reasoning_tokens":2061,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-03T17:51:59.323237+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the spectrum of the four-dimensional transfer matrix U (or V) for a specific S and real η and find an eigenvalue of modulus greater than or equal to 1 other than the Perron eigenvalue; then the claimed exponential decay of ∥Ψ_j∥ and the limit ∥Ψ²−1∥²→0 fail. Alternatively, for the spin-1 chain with boundary conditions of Eq. (59), compute the edge autocorrelator plateau at larger L and show it decays to a value below |c_1|²+|c_2|², contradicting the claim that additional ESZMs exist.","supporting_citations":[],"review_version":1}