{"id":"f0c58f77-bc8e-4ae8-8c75-d9bd4c5846a5","arxiv_id":"2512.19523","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"The 2D random-bond Ising model at the Nishimori point has two conformal boundary fixed points (free: unstable, fixed: stable) with boundary entropies, b.c.c. dimensions, and multifractal boundary-spin exponents extracted.","lead":"A tensor-network study maps out the boundary critical behavior of the two-dimensional random-bond Ising model at the Nishimori multicritical point, identifying free and fixed conformal boundary conditions and extracting boundary entropies and operator dimensions. The results add a boundary-universality layer to a well-studied bulk critical point and could eventually inform decoding thresholds in quantum error correction, though that connection is only claimed in the abstract.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The numerical BCFT data all rest on the asserted exactness of the F' last layer in Eq. (7); a direct check against exact partition functions is needed before the boundary entropy and dimension values can be treated as universal.","rationale":"The reader's conditional verdict is justified. The single most load-bearing assumption is the exactness of the F' layer and the associated twist-operator formula. The body's numerical protocol is careful: TEBD truncation error <1e-10, L_y=10L_x, 1e6 samples, central charge c_vN~0.42 matching previous work. These are real strengths. But they do not establish that |Psi_theta(J)> is exactly the boundary partition function. The Appendix's argument is heuristic: it identifies which bonds are missed by the brick-wall pattern and writes an F' that cancels them, but it does not prove uniqueness or check Eq. (17) independently. Given that all universal numbers are read off from this state, this is the weakest link in the logical chain. The abstract overclaims (three fixed points, QEC bridge) and the deferred RG derivation are real but secondary; they do not invalidate the numerics, and they are already noted. My proposed test is inexpensive and decisive. If Eq. (17) passes, I see no remaining objection to the central numerical claim, pending the editorial fixes.","tokens_in":23751,"tokens_out":10081,"duration_ms":106788,"concrete_test":"Directly verify Eq. (17) for a small system (e.g., L_x=8, L_y=16) at p_c=0.1092212. For at least 100 disorder realizations (J1,J2), compute the tensor-network overlap <Psi_theta'(J2)|Psi_theta(J1)> with TEBD truncation error below machine precision (or exact MPS/contraction) and compare with brute-force enumeration of the RBIM cylinder partition function on the right-hand side of Eq. (17) for theta,theta' in {0,pi/2}. If the relative difference is not zero within Monte Carlo error (say <~1e-6), the F' layer does not restore the brick-wall seam, and the BCFT identification of all extracted numbers fails. If it passes, the entropy formula can then be checked separately by comparing S_A[Psi_theta] with the replica expression -partial_n(Z_n/Z_1^n)|_{n=1} computed by explicit replicated transfer matrices.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central quantitative claims—S_bdy(pi/2)=-0.574(1), 1/nu=0.75(1), Delta_f+/-=0.0446(2), Delta_+-=1.055(2), and the multifractal Delta_n in Eq. (13)—are all extracted from the tensor-network state |Psi_theta(J)> defined by Eq. (7). The pipeline is valid only if two statements hold exactly: (i) the overlap <Psi_theta'(J2)|Psi_theta(J1)> equals the RBIM cylinder partition function with boundary orientations theta, theta', Eq. (17); (ii) the half-chain von Neumann entropy of this state obeys the BCFT twist-operator formula Eq. (8). The Appendix asserts that the special last layer F' is 'uniquely determined' by these requirements, but it gives no derivation and no independent numerical verification. The construction is plausible, but good log-L fits in Figs. 2-4 cannot detect a systematic mismatch in the seam: a defect line would also produce log-L behavior with modified coefficients. Thus the extracted boundary entropies, b.c.c. dimensions, and boundary multifractal exponents are not protected against the one assumption that everything else inherits. This is more load-bearing than the abstract's third fixed point or the deferred RG derivation, because those do not enter the numerical extraction.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies boundary critical behavior at the two-dimensional Nishimori multicritical point of the random-bond Ising model. Using a tensor-network transfer-matrix construction with a one-parameter family of boundary spin orientations, the authors extract boundary entropies from half-chain von Neumann entropies, identify free and fixed conformal boundary fixed points from data collapse, obtain boundary-condition-changing operator dimensions from wavefunction overlaps, and fit moments of boundary spin-spin correlations to a multifractal spectrum. The paper also quotes a one-loop 6−ϵ boundary RG result and uses it in a Padé interpolation. The central body claims are: (i) the free and fixed boundaries are conformal fixed points, with S_bdy(π/2)=−0.574(1) and 1/ν=0.75(1) at the free fixed point; (ii) the b.c.c. dimensions are Δ_f±≈0.0446(2) and Δ_+−≈1.055(2); (iii) the boundary spin-field moments satisfy Δ_1=Δ_2≈0.263(1), Δ_3=Δ_4≈0.369(1), Δ_5=Δ_6≈0.436(1), Δ_7=Δ_8≈0.486(1).","tokens_in":24211,"tokens_out":6185,"duration_ms":65834,"significance":"If the construction is valid, this is the first systematic BCFT characterization of Nishimori boundary criticality, and the numerical data would be a valuable benchmark: S_bdy(π/2)=−0.574(1), 1/ν=0.75(1), Δ_f±=0.0446(2), Δ_+−=1.055(2), and the multifractal ladder in Eq. (13). The simulations are careful and reasonably extensive (L up to 64, 10^6–10^7 samples, small quoted errors), and the benchmark c_vN≈0.42 is consistent with Ref. [42]. The paper also gives an explicit falsifiable structure: the equality of odd and even moments follows from a gauge-symmetry argument and is confirmed numerically. However, the central pipeline currently rests on an asserted—not derived or checked—final-layer transfer matrix F', and the advertised controlled RG analysis is deferred to a companion paper. These gaps must be closed before the extracted numbers can be treated as universal BCFT data.","major_comments":[{"comment":"All quantitative claims (S_bdy, b.c.c. dimensions, and the multifractal exponents) are extracted from states |Ψ_θ(J)> defined with the special final layer F'. The Appendix asserts that F' is \"uniquely determined\" by the requirements that the overlap equals the cylinder partition function (Eq. (17)) and that the Rényi entropy has the twist-operator form, but it does not give a derivation or any independent verification. A microscopic defect line along the seam would also produce logarithmic-in-L behavior with modified coefficients, so the excellent fits in Figs. 2–4 cannot by themselves validate the construction. I request either a full derivation showing that all seam bonds are exactly reproduced, or a direct small-L check of Eq. (17) (and of the analogous F used for |Ψ(J)> in Eq. (11)) against exact RBIM cylinder partition functions for several disorder realizations. This is load-bearin","section":"Eq. (7) and Appendix (\"The final-layer transfer matrix\")"},{"comment":"The abstract and introduction advertise a \"controlled renormalization group analysis,\" but the manuscript contains no derivation. Eq. (15) is asserted, and the sentence \"The full derivation of the boundary RG will be presented elsewhere [62]\" explicitly defers it. Likewise, the claim \"we find η_M̂ = η_Q̂\" is stated without any calculation of the surface self-energy or anomalous dimensions. Since Eq. (16) uses Eq. (15) as input for the Padé interpolant, this is not a peripheral remark. The authors should either include the one-loop boundary calculation (at least in an appendix) or clearly reframe Eq. (15) as a conjecture from work in preparation, and adjust the abstract accordingly.","section":"Sec. \"RG calculation and discussion\", Eq. (15)"},{"comment":"The abstract claims three conformal boundary fixed points (free, fixed, and random) and \"a bridge between boundary universality class and boundary decoding threshold.\" The body, by contrast, states in the Introduction that \"we find two conformal boundary fixed points\" and the concluding discussion only mentions possible future generalizations to disordered boundary ensembles and interfaces. No random fixed point is constructed or analyzed, and no quantum error-correcting code threshold is computed. This is not a minor wording issue: the abstract promises results that are absent from the paper. Please either remove these claims or present the missing results.","section":"Abstract vs. body"}],"minor_comments":[{"comment":"The irrelevant exponent 1/ν_irr≈−0.9 is quoted without a statistical error, and the collapse is shown for L=8,16,24,32 only. Since this exponent is used to conclude that the fixed boundary is stable, please provide an error estimate and, if possible, a stability check with larger L.","section":"Fig. 2(d)"},{"comment":"Please define ϵ=6−d explicitly where the Padé approximant is introduced. The text switches between d and ϵ, and the reader must infer the relationship from the preceding paragraph.","section":"Eq. (16)"},{"comment":"The central charge is fitted from only L=8,16,32,64. The values c_vN=0.416(2) and 0.419(2) are consistent with Ref. [42], but the fit range is short; a sentence noting that the extrapolation was checked with the L_y=10L_x aspect ratio would be useful.","section":"Eq. (8) and Fig. 2(a)"},{"comment":"The phrase \"directly extracts universal boundary data\" in the Conclusion overstates the current status, given the unverified F' construction. I suggest softening until the Appendix provides the missing derivation or numerical check.","section":"General"}],"recommendation":"major_revision","confidential_remarks":"The numerical work appears genuine and the data are internally consistent, but the paper's advertised analytical component and the abstract's three-fixed-point/QEC claims need to be reconciled with the body. The most serious technical issue is the asserted final-layer transfer matrix F': because every BCFT number depends on it, a direct exact check is essential. I do not see grounds for rejection, provided the authors can supply the missing derivation/check or appropriately limit their claims."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: the tensor-network boundary data are the strongest part and deserve to be seen, but the abstract sells more than the manuscript currently delivers. The third fixed point and the quantum-error-correction bridge appear in the abstract and then don't show up in the body.\n\nWhat is genuinely new: this is the first systematic extraction of boundary entropies, b.c.c. operator dimensions, and boundary multifractal exponents at the 2D Nishimori multicritical point. The numerics look careful: L up to 64, 1e6-1e7 disorder samples, TEBD truncation error below 1e-10, and the fitted c_vN about 0.42 matches the literature. The odd/even moment pairing Delta_2k-1 = Delta_2k follows from the Nishimori gauge symmetry and is confirmed numerically; that's a nice internal consistency check. The data collapse for the free boundary (1/nu ~ 0.75) and for the fixed boundary (irrelevant direction) is plausible and coherent. Credit is earned here: this gives a nonperturbative anchor for a genuinely hard problem.\n\nWhere it gets soft, in decreasing order of importance. First, the whole pipeline sits on the special last-layer F' in Eq. (7). The appendix says F' is uniquely determined by requiring the overlap to equal the cylinder partition function and the von Neumann entropy to follow the twist-operator formula, but there is no derivation and no independent numerical check. A direct comparison against an exact partition function at small L would settle whether the seam is really restored. Good log-L fits cannot detect a conformal defect line at the seam, so this is the piece I would want verified before calling these numbers universal BCFT data. Second, the abstract promises three conformal boundary fixed points, but the body only analyzes the free and fixed points. The random boundary condition is never defined or studied. Either trim the abstract or add the third point. Third, the analytic result, Eq. (15), is asserted and attributed to a companion paper; the derivation is not in the text. The Pade in Eq. (16) is fairly labeled as an illustration, so that's better. Fourth, the QEC/decoding-threshold connection is a sketch, not a result.\n\nBottom line: this is a solid numerical contribution, worth a serious referee, not a desk reject. I would ask the authors to add the F' verification or at least discuss its uncertainty, and to align the abstract with the body. People working on Nishimori criticality or boundary CFT in disordered systems will get real value from this.","headline":"Boundary BCFT numerics look credible and are the real contribution; the abstract's three-fixed-point and QEC claims outrun the body, and the F' seam assumption needs a direct check.","tokens_in":24569,"tokens_out":3605,"would_cite":true,"duration_ms":35941,"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 establishes that boundary criticality at the Nishimori point is governed by two conformal boundary fixed points—free and fixed—and that the free boundary exhibits multifractal scaling of spin fields.","keywords":["boundary criticality","Nishimori point","random-bond Ising model","boundary conformal field theory","boundary entropy","multifractality","tensor network","boundary condition changing operator"],"falsifier":"In a direct Monte Carlo simulation of the RBIM on a finite cylinder, measure the n-th moments of the boundary spin-spin correlation function for n=1,...,8 and check whether they follow ⟨σ0σ_l⟩_n ∼ L^{-2Δ_n} with the reported Δ_n; a mismatch beyond error bars would refute the multifractal claim.","tokens_in":23686,"feed_emoji":"🎲","tokens_out":10103,"duration_ms":87762,"temperature":0.7,"pith_summary":"The paper asks what happens to boundaries at the Nishimori multicritical point of the two-dimensional random-bond Ising model—a strongly disordered critical point whose bulk is well characterised but whose boundary universality class was previously unknown. Using tensor-network methods that map a cylinder of the classical model with rotated boundary spins to a quantum state, it identifies two conformal boundary fixed points: a free boundary (spin orientation angle θ=0) and a fixed boundary (θ=π/2). The free boundary is shown to be an unstable fixed point (1/ν=0.75(1)), while the fixed boundary is stable; the authors extract boundary entropy S_bdy(π/2)=−0.574(1), boundary-condition-changing dimensions Δ_f±≈0.0446 and Δ_+-≈1.055, and a multifractal spectrum of boundary spin-field exponents at the free boundary, with paired values Δ_1=Δ_2≈0.263, Δ_3=Δ_4≈0.369, Δ_5=Δ_6≈0.436, Δ_7=Δ_8≈0.486. A controlled 6−ε boundary RG calculation complements the numerics, finding equal surface anomalous dimensions for the magnetization and spin-glass fields and a one-loop formula for the multifractal dimensions. If these results are right, they supply the first systematic boundary-conformal-field-theory data for a disordered multicritical point and tie boundary universality to the decoding threshold of quantum error-correcting codes.","feed_headline":"Two conformal boundary fixed points found at Nishimori criticality","feed_subtitle":"At the random-bond Ising multicritical point, the free boundary is unstable and multifractal while the fixed boundary is stable.","key_machinery":"The central mechanism is the tensor-network representation of the RBIM partition function on a cylinder, with a special final-layer transfer matrix F′ that compensates for the missing bonds in the brick-wall pattern. This F′ makes the overlap ⟨Ψ_{θ'}(J2)|Ψ_θ(J1)⟩ exactly proportional to the RBIM partition function on a cylinder with boundary spin orientations θ' and θ, and makes the half-chain von Neumann entropy of |Ψ_θ(J)⟩ equal to a twist-operator expectation in the associated BCFT. The continuous family of boundary conditions is parameterized by the angle θ in the boundary tensor B^θ_σ = cos(π/4−θ/2)δ_{σ,+1} + sin(π/4−θ/2)δ_{σ,−1}, which rotates the boundary spin orientation from free (θ","core_discovery":"The paper's central claim is that the boundary of the 2D random-bond Ising model at its Nishimori multicritical point has two conformal boundary fixed points—free (θ=0) and fixed (θ=π/2)—and that boundary spin fields at the free fixed point exhibit multifractal scaling. A tensor-network construction with a seam-correcting final layer F′ makes wavefunction overlaps equal to RBIM cylinder partition functions with rotated boundary spins, and half-chain von Neumann entropy obey the BCFT twist formula. From this the authors extract boundary entropy S_bdy(π/2)=−0.574(1), stability data (1/ν=0.75(1) at the free fixed point, irrelevant perturbation at the fixed point), and paired multifractal expone","pith_inferences":["The stability pattern (free unstable, fixed stable) suggests that in any physical simulation of the RBIM on a lattice with open boundaries, the boundary will generically be in the fixed-boundary universality class unless a boundary field is fine-tuned to zero; a direct Monte Carlo measurement of the surface magnetization profile on finite cylinders would be a clean, independent test.","The paired multifractal spectrum at the free boundary is a signature of the underlying supersymmetry of the Nishimori point; if the supersymmetric CFT description is developed, it could predict the full Δ_n spectrum, and the n≤8 data here would be a benchmark.","The one-loop 6−ε formula, combined with the d=2 numerical anchor, could be extrapolated to predict higher-moment boundary exponents (n>8); running the same tensor-network pipeline for n up to, say, 16 would test whether the multifractal spectrum follows the analytic form for all n."],"forward_implications":["The free boundary is unstable (1/ν=0.75(1)) and the fixed boundary stable, so a generic microscopic boundary condition flows to the fixed boundary—any boundary field not tuned to θ=0 will pull the boundary to the fixed-point condition.","The extracted boundary data—S_bdy(π/2), Δ_f±, Δ_+−, and the multifractal spectrum—are universal BCFT data for a nonunitary disordered critical point, usable as anchors for resummations of the 6−ε expansion.","The numerical confirmation that Δ_{2k−1}=Δ_{2k}, required by the Nishimori gauge symmetry, shows the gauge constraint survives at the boundary and can be used to simplify boundary operator content.","The paper's stated connection to quantum error-correcting codes implies that boundary universality class (specifically boundary entropy and b.c.c. dimensions) controls the boundary decoding threshold on the Nishimori line."],"fun_headline_variants":["Multifractal boundary at Nishimori point","Boundary fixed points at Nishimori multicriticality","Random Ising boundary reveals two universality classes","Nishimori boundary: free is multifractal, fixed is stable","Boundary criticality in random 2D Ising model"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The load-bearing premise is that the special seam-correcting layer F′ exactly restores the missing bonds of the brick-wall pattern, so the tensor-network state's overlaps and entropies are genuinely the RBIM cylinder partition function and its twist-operator expectation; if F′ is only approximate, the extracted numbers are not universal BCFT data.","fun_headline_variants_meta":{"raw":{"variants":["Multifractal boundary at Nishimori point","Boundary fixed points at Nishimori multicriticality","Random Ising boundary reveals two universality classes","Nishimori boundary: free is multifractal, fixed is stable","Boundary criticality in random 2D Ising model"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000328,"raw_usage":{"total_tokens":1652,"prompt_tokens":707,"completion_tokens":945,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":451,"completion_tokens_details":{"reasoning_tokens":862}},"tokens_in":451,"tokens_out":945,"duration_ms":8198,"temperature":1.0,"reasoning_tokens":862,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-03T14:39:00.520992+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"In a direct Monte Carlo simulation of the RBIM on a finite cylinder, measure the n-th moments of the boundary spin-spin correlation function for n=1,...,8 and check whether they follow ⟨σ0σ_l⟩_n ∼ L^{-2Δ_n} with the reported Δ_n; a mismatch beyond error bars would refute the multifractal claim.","supporting_citations":[],"review_version":1}