{"id":"bf078a4a-0331-4371-806a-340437ff3851","arxiv_id":"2602.15116","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"The stabilizer Rényi entropy of an infinite matrix product state decomposes into bulk, boundary, and exponentially decaying parts, and the associated 'magic correlation length' diverges at criticality with a different exponent than the standard correlation length.","lead":"The paper shows that 'magic' — the quantum resource behind universal quantum computing — has its own correlation length in matrix-product-state wavefunctions, one that diverges at quantum phase transitions. This gives a new, computation-relevant way to detect criticality in quantum many-body systems.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"General claim that ξ_SRE diverges at every continuous phase transition rests on an unproven gap-closing step in the replica transfer matrix; exact proof and numerics cover only special cases.","rationale":"The paper's clean contributions are the spectral decomposition (Eqs. (13)-(20)) and the exact χ=2 skeleton results (Eqs. (40)-(51)); these are internally consistent and checkable. The load-bearing step is the generalization from these special cases to 'continuous phase transitions' generally. In the abstract and Conclusions, the paper asserts that ξ_SRE diverges at continuous phase transitions and is a robust criticality diagnostic. For a finite-χ iMPS, this must mean that, as the physical transfer matrix gap closes with increasing bond dimension (or along the skeleton trajectory), the SRE replica transfer matrix gap also closes. This is a statement about the spectrum of a 2n-replica object with a nontrivial local filter Λ. There is no derivation; the only exact example is a χ=2 stabilizer-at-criticality skeleton, and the numerical evidence is confined to one Z2 model family with a narrow and partially unconverged window (χ_t≤64). The same gap-closing step is also needed for the spatial-response interpretation of Sec. IV, although there the derivation is more direct. The BCFT-inspired scaling form Eq. (53) cannot rescue the general claim because its coefficient is imported by analogy and the numerical test in Appendix E shows ξ_SRE is not converged. I see no internal contradiction in the paper; the problem is an unsupported generalization. If the out-of-sample test at, e.g., the XXZ critical point shows ξ_SRE diverging, the paper's central claim is much stronger; if not, the claim should be restricted to the models where it is verified. The reader's CONDITIONAL verdict is therefore appropriate, and no adjustment is needed.","tokens_in":29018,"tokens_out":22279,"duration_ms":218761,"concrete_test":"Test the gap-closing premise out of sample: compute the SRE correlation length for an iMPS ground state at a continuous phase transition not in the Z2 family—e.g., the critical XXZ chain (c=1)—using VUMPS at increasing bond dimensions χ=50,100,200 and Pauli-basis truncation χ_t=56,64,128. Plot |µ_2/µ_1| against the physical |λ_2|/λ_1; if the SRE ratio remains bounded away from 1 while the physical gap closes with increasing χ, the universal claim fails and the paper should be revised to restrict conclusions to the Z2 models/skeletons studied. If the ratio approaches 1 in this independent model, the central premise survives the test.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The headline claim—that the SRE correlation length ξ_SRE (Eq. (20)) diverges at continuous phase transitions and is a universal criticality diagnostic—requires that the subleading gap of the SRE replica transfer matrix E (Eq. (9)) closes whenever the physical transfer matrix gap closes (|λ_2|→1). This is not derived anywhere. The text proves it only for the χ=2 skeleton (Eqs. (50)-(51)) and verifies it numerically along the cluster-Ising Z2 critical lines with χ_t≤64 (Fig. 5). E is a 2n-replica matrix with a nontrivial local Pauli filter Λ; the spectral gap of such a filtered object is not guaranteed to track the physical gap, and one can imagine filters that project out the soft mode. Since this silent assumption is exactly what converts the exact skeleton calculation into the broad universal statement in the abstract and Conclusions, the central claim is not yet established. The proposed scaling form Eq. (53) with coefficient 1/8 is also imported by analogy from BCFT Eq. (27), not derived, and the numerical support is pre-asymptotic: ξ_SRE saturates at the largest χ_t and Appendix E states it is not fully converged. A single continuous-transition universality class where ξ_SRE stays finite while ξ diverges would falsify the headline, even though the χ=2 skeleton results would remain correct.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops a spectral transfer-matrix framework for the stabilizer Rényi entropy (SRE) of infinite matrix product states. For an N-site subsystem, Eqs. (17)–(19) decompose the SRE into an extensive term set by the dominant eigenvalue μ1 of the 2n-replica transfer matrix E, an O(1) boundary term, and subleading corrections that decay with a newly defined SRE correlation length ξ_SRE = −1/log|μ2/μ1|. The paper shows that ξ_SRE controls the exponential decay of the SRE response to two local unitary perturbations (Sec. IV). For the χ=2 cluster–Ising MPS skeleton, the eigenvalues of E are obtained in closed form (Eq. (40)), giving ξ and ξ_SRE that diverge at g=0 with different exponents (Eq. (51)). Numerics for the full cluster–Ising model map m^(2), L∞^(2), and ξ_SRE across the phase diagram (Fig. 4) and test a proposed universal scaling W∞^(2) = (1/8) log ξ_SRE + b along the Z2 critical lines (Eq. (53), Fig. 5). Appendices provide exact diagonalization benchmarks and a χ=4 skeleton example.","tokens_in":29399,"tokens_out":11750,"duration_ms":108642,"significance":"The spectral decomposition in Sec. III is clean and exact for any iMPS, and the skeleton results (Eqs. (40), (50), (51)) provide a valuable analytic benchmark. The perturbation-response formula in Sec. IV gives ξ_SRE a concrete operational meaning that goes beyond a formal definition. If the universal divergence of ξ_SRE at continuous transitions is generic, the paper would establish a new nonstabilizerness length scale and a route to detect criticality using magic even when the SRE density is smooth. However, the generality of that claim and the coefficient 1/8 in Eq. (53) are not yet established: the exact evidence covers a two-parameter skeleton, and the numerical evidence is limited to one model with a narrow converged window (χ_t ≤ 64).","major_comments":[{"comment":"The abstract and Conclusions assert that ξ_SRE diverges at continuous phase transitions in general, but the proof covers only the χ=2 skeleton (Eq. (50)) and the χ=4 skeleton (Appendix C), plus numerical cluster-Ising Z2 lines with χ_t≤64 (Fig. 5). The subleading gap of the filtered Pauli-replica transfer matrix E (Eq. (9)) is not guaranteed to close whenever the physical transfer-matrix gap closes; the local Pauli filter Λ could, in principle, suppress the soft mode. This unproven gap-closing step is exactly what converts the skeleton calculation into the headline universality claim. The authors should either prove gap closure for a class of MPS (e.g., symmetry-constrained or free-fermion states) or explicitly restrict the claim to the studied cases. A concrete test would be a different universality class, such as the XXZ or J1–J2 chain.","section":"III A, VI B, Conclusions"},{"comment":"The universal scaling form W∞^(n) = (2Δ_{2n}/(n−1)) log ξ_SRE + b is introduced by analogy with the BCFT result Eq. (27), with an ad hoc factor-of-two reduction for a single boundary; it is not derived from the spectral decomposition. Numerical support is partial: only g_c=0 and g_c=2 are consistent with the 1/8 slope, while intermediate points overshoot (Fig. 5(c)). Appendix E states that ξ_SRE is not fully converged at the largest χ_t and that the cleanest linear behavior is found using log ξ rather than log ξ_SRE. Since Eq. (53) underlies the 'universal scaling' claim in the abstract, this is load-bearing. The authors should either derive the form from the spectral framework or clearly label it as a conjecture supported only at the Ising and cluster endpoints.","section":"Eq. (53), Sec. VI B"},{"comment":"The decomposition relies on f(N)≪c1 in the large-N limit, i.e., on |μ2/μ1|<1. At a continuous transition in the χ→∞ limit, μ2→μ1, so the exponential correction does not decay and the limits N→∞ and χ→∞ do not commute. The paper should state the finite-χ interpretation of ξ_SRE in Eq. (53) and explain how the double-scaling limit is taken. Without this, the definition of ξ_SRE as a 'diverging correlation length' at criticality is ambiguous, and the numerical extraction of the 1/8 slope in Fig. 5 rests on an implicit choice of ordering of limits.","section":"Eqs. (19)–(20), Sec. III A"}],"minor_comments":[{"comment":"The text near Fig. 3 says 'Figure 2(c) shows the effect of separation distance r...' — the correct cross-reference is Fig. 3(c).","section":"Sec. V B"},{"comment":"The notation M^(n)(ρ) for the mixed-state SRE is used before it is defined. Please give the explicit mixed-state definition (analogous to Eq. (6)) or cite the original reference more precisely.","section":"Eq. (12)"},{"comment":"The superscripts S^(2)(ρ_AB) and I^(2)(A:B) should be S^(n)(ρ_AB) and I^(n)(A:B) for the general n-th order Rényi and mutual SRE, unless the authors intend to specialize to n=2.","section":"Appendix B, Eqs. (B6)–(B7)"},{"comment":"The divergence of ξ_SRE along the critical lines is visually inferred from a color plot. A logarithmic color scale or contour lines would make the divergence much easier to assess.","section":"Fig. 4(c)"},{"comment":"The statement that c3 diverges while c1→0.25 as g→−1, and that these divergences 'are acceptable' because 'the coefficients c_i are weighted with the corresponding eigenvalue,' would benefit from a brief explanation of the regularization mechanism.","section":"Sec. V A, c_i discussion"}],"recommendation":"major_revision","confidential_remarks":"The exact skeleton calculations are convincing and the transfer-matrix framework is likely to be reused. The main gap is between the evidence and the broad abstract/Conclusions claims of universal divergence and universal 1/8 scaling. These issues are fixable within the scope of the manuscript: the authors can (i) prove or clearly restrict the gap-closing claim, (ii) derive or explicitly label Eq. (53) as a conjecture, and (iii) incorporate the convergence caveats from Appendix E into the main text. I would not reject; the paper is a solid contribution once the claims are matched to the evidence."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"First thing to know: the paper is worth reading for the exact χ=2 skeleton analysis and the three-term decomposition of the SRE of a finite subsystem. Eq. (19) is clean, the definition of ξ_SRE as −1/log|μ2/μ1| is natural, and the closed-form eigenvalues in Eq. (40) with the ξ~g^{-1}, ξ_SRE~g^{-2} exponents in Eq. (51) are checkable and likely correct. The χ=4 skeleton in Appendix C shows the framework generalizes, and the authors are honest about where their numerics are not converged (Appendix E). This is real work.\n\nThe soft spot is exactly what the stress-test note says: the leap from these examples to the abstract's claim that ξ_SRE diverges at continuous phase transitions in general is not proven. The argument requires the subleading gap of the SRE replica transfer matrix E to close whenever the physical transfer matrix gap closes, and that is asserted, not derived. It holds for the skeleton and is verified numerically along the cluster-Ising Z2 lines, but nothing in the paper rules out a filter Λ that suppresses the soft mode. So the universal claim is a conjecture, and the authors should say so more clearly in the abstract.\n\nThe numerical evidence for universality is also pre-asymptotic. ξ_SRE saturates at the largest χ_t, the accessible window in Fig. 5 is narrow, and the scaling form Eq. (53) with coefficient 1/8 is imported from BCFT by analogy, not derived from the replica theory. The intermediate-g_c slopes overshoot 1/8; the authors argue they drift toward it, and the ED data in Appendix D supports a crossover picture, but that is not the same as establishing the universal coefficient. Add that no code or data is provided; for a numerical paper that is a practical obstacle for referees who want to check the convergence claims.\n\nNone of this sinks the exact results. The decomposition, the skeleton eigenvalues, and the two-point-perturbation response (Sec. IV) will stand. The paper is a good candidate for a serious referee: the core derivations are solid, the claims are appropriately qualified in the body (mostly), and the limitations are acknowledged. I would send it to review with a request for code/data, error estimates, and either a proof or a downgrade of the general gap-closing claim. A motivated reader can fix the presentation in a revision.\n\nWho should read it: anyone working on nonstabilizerness in tensor networks or transfer-matrix methods. I'd cite it for the skeleton results and the ξ_SRE definition, and I'd bring it to a reading group focused on magic in many-body systems.","headline":"Exact skeleton results and the SRE correlation length definition are the real contributions; the universal-criticality claim is a conjecture supported by suggestive, not conclusive, numerics.","tokens_in":29880,"tokens_out":4210,"would_cite":true,"duration_ms":39680,"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":"The stabilizer Rényi entropy of an infinite matrix product state carries its own correlation length, distinct from the standard one, that diverges at continuous phase transitions.","keywords":["stabilizer Rényi entropy","nonstabilizerness","matrix product states","transfer matrix","correlation length","cluster-Ising model","quantum criticality","mutual magic"],"falsifier":"Compute the ratio |μ2/μ1| of the SRE replica transfer matrix for an iMPS approximation of a known continuous phase transition, e.g., in the XXZ or J1–J2 chain, and check whether it approaches 1, i.e., whether ξ_SRE diverges. If for some transition the ordinary correlation length diverges while |μ2/μ1| stays bounded below 1, the universal-criticality claim fails. A more direct check is to look for the predicted exponential decay e^(−r/ξ_SRE) in the two-point SRE response in finite exact-diagonalization chains; if the decay instead follows the standard correlation length ξ, the distinct-length c","tokens_in":28886,"feed_emoji":"🪄","tokens_out":4198,"duration_ms":40254,"temperature":0.7,"pith_summary":"The paper sets out to show that nonstabilizerness—'magic'—is not just a resource count but carries genuine many-body correlation structure: for a subsystem embedded in an infinite matrix product state, the stabilizer Rényi entropy decomposes into an extensive bulk term, a boundary mutual-magic term, and an exponentially decaying correction. The decay rate defines an SRE correlation length, extracted from the subleading spectrum of the replica transfer matrix, which is generally different from the ordinary correlation length and diverges at continuous phase transitions. On the exactly solvable χ = 2 MPS skeleton of the cluster-Ising model, the paper derives closed forms for all these quantities and shows the SRE correlation length diverges as g^(−2) while the standard one goes as g^(−1). Along the Z2 critical lines of the cluster-Ising model, the mutual SRE approaches a universal logarithmic scaling whose coefficient is consistent with the boundary conformal field theory value 1/8. A sympathetic reader would care because this gives a concrete, computable diagnostic through which magic can detect criticality even when the SRE density itself behaves smoothly.","feed_headline":"SRE correlation length diverges at continuous phase transitions","feed_subtitle":"A magic-specific length scale, distinct from the usual correlation length, blows up at criticality.","key_machinery":"The central object is the replica transfer matrix E (Eq. 9), built from 2n copies of the MPS tensor and Pauli replica operators; its spectral decomposition E = Σ μ_i |R_i^m)(L_i^m| generalizes the ordinary MPS transfer matrix. The ratio of subleading to leading eigenvalues defines ξ_SRE, while the overlap c1 between the dominant eigenvectors of E and the replicated ordinary transfer matrix defines the boundary/mutual SRE. This spectral split converts the SRE of a finite subsystem into a three-term expression (extensive, boundary, exponential correction), and converts two-point responses of magic into a sum of a disconnected term plus an exponential with length ξ_SRE.","core_discovery":"The central claim is that the eigenspectrum of the SRE replica transfer matrix E—not just its dominant eigenvalue—carries universal information. For an N-site subsystem, the paper derives M^(n)(ρ) = N log μ1/(1−n) + [log(c1 + f(N))]/(1−n), with f(N) ≈ c2 e^(−N/ξ_SRE) for large N; the length ξ_SRE = −1/log|μ2/μ1| is the SRE correlation length. This length governs both the approach of subsystem SRE to its thermodynamic limit and the exponential decay of SRE correlations induced by two spatially separated local unitary perturbations. The paper argues and verifies that ξ_SRE diverges at continuous phase transitions—with a different exponent from the standard correlation length—so nonstabilizerne","pith_inferences":["If the gap of the SRE replica transfer matrix closes generically at criticality, ξ_SRE could become a standard numerical diagnostic for magic-specific length scales, complementing entanglement entropy.","The exact skeleton shows ξ_SRE diverges faster than ξ, suggesting magic correlations may be longer-ranged than ordinary ones near criticality; if this persists in other models, it would mean nonstabilizerness is a more sensitive probe of long-range order.","A testable extension is to compute ξ_SRE for other critical chains, e.g., the XXZ chain or J1–J2 chain, and check whether the ratio of critical exponents tracks properties of the replicated CFT.","The sign of the mutual SRE is negative in the SPT phase of the skeleton, hinting that it encodes entanglement dominance; one could ask whether this sign correlates with symmetry-protected topological order in general."],"forward_implications":["The SRE of a finite subsystem is not featureless: it carries an exponentially decaying correction whose length scale can be read off from the second eigenvalue of the replica transfer matrix.","The SRE correlation length diverges at continuous phase transitions, so it can label critical points even where the SRE density or mutual SRE looks smooth.","Magic correlations respond to local unitaries with a characteristic decay e^(−r/ξ_SRE), giving an operational way to measure nonstabilizerness length scales.","For the χ = 2 cluster-Ising skeleton, exact formulas give ξ ≈ 1/(2g) and ξ_SRE ≈ 1/(14g²); the SRE reaches its maximum at g* = ±(3−2√2), where the preparing unitaries are closest to magic gates.","Along the Z2 critical line, the mutual SRE grows as (1/8) log ξ_SRE after a pre-asymptotic crossover, matching the predicted boundary CFT coefficient."],"fun_headline_variants":["Magic length scale diverges at phase transitions","New quantum correlation length from magic blows up at criticality","Nonstabilizerness reveals criticality via its own length scale","SRE correlation length: a fresh criticality probe","Magic's correlation length diverges at continuous transitions"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The broad claim that ξ_SRE diverges at every continuous phase transition rests on the assumption that the gap between the two leading eigenvalues of the SRE replica transfer matrix closes at the same critical points as the ordinary transfer matrix gap; this is proven exactly only for the χ = 2 skeleton and verified numerically along the cluster-Ising critical lines within a limited window.","fun_headline_variants_meta":{"raw":{"variants":["Magic length scale diverges at phase transitions","New quantum correlation length from magic blows up at criticality","Nonstabilizerness reveals criticality via its own length scale","SRE correlation length: a fresh criticality probe","Magic's correlation length diverges at continuous transitions"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000182,"raw_usage":{"total_tokens":1144,"prompt_tokens":735,"completion_tokens":409,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":479,"completion_tokens_details":{"reasoning_tokens":332}},"tokens_in":479,"tokens_out":409,"duration_ms":4066,"temperature":1.0,"reasoning_tokens":332,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-02T22:58:45.175001+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the ratio |μ2/μ1| of the SRE replica transfer matrix for an iMPS approximation of a known continuous phase transition, e.g., in the XXZ or J1–J2 chain, and check whether it approaches 1, i.e., whether ξ_SRE diverges. If for some transition the ordinary correlation length diverges while |μ2/μ1| stays bounded below 1, the universal-criticality claim fails. A more direct check is to look for the predicted exponential decay e^(−r/ξ_SRE) in the two-point SRE response in finite exact-diagonalization chains; if the decay instead follows the standard correlation length ξ, the distinct-length c","supporting_citations":[],"review_version":1}