{"id":"7162132d-af72-4987-ba42-96631472d8b2","arxiv_id":"1908.07700","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"When an entanglement cut ends on a gapped boundary of a 2+1D topological phase, the topological entanglement entropy is controlled by the half-linking matrix, which replaces the modular S matrix used without boundaries.","lead":"This paper calculates how much hidden quantum information is stored across a cut that ends on the physical boundary of a topologically ordered material, and identifies the half-linking number as the quantity that controls it. It gives researchers a way to compute boundary corrections to topological entanglement entropy in Chern-Simons theories.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Section 3.3 shows the half-linking-matrix normalization, and hence the claimed topological entanglement entropy, shifts by Eq. (3.48) under the non-universal edge parameter r; the advertised universality is therefore not fully established.","rationale":"The reader's weakest assumption identifies the junction boundary condition and the r = 1 normalization as the fragile point; the paper itself, in Sections 3.3 and 4, admits that the normalization of the half-linking matrix can be altered by edge-theory choices and that this shifts the topological entanglement entropy. This is precisely the load-bearing concern about the central claim, because Eq. (2.14) presents -ln γ_{x0} as a topological invariant analogous to the S-matrix, yet Eq. (3.48) shows it is ambiguous up to -N ln \\tilde r. The explicit Z_N Abelian Chern-Simons calculations are internally consistent for a fixed r and match earlier lattice results, which is genuine support, but the general universality claim is not supported. The paper is honest about the limitation, so a conditional verdict is appropriate; no change to the reader's verdict is needed.","tokens_in":23406,"tokens_out":4852,"duration_ms":46131,"concrete_test":"Perform a lattice computation of the Z_N toric code on a cylinder with electric-electric gapped boundaries (the e+e case of Section 3.0.2), using the explicit boundary Hamiltonian of Ref [3]; compute the strip entanglement entropy for N = 2, 3, 5, subtract the area-law term, and compare the topological contribution with Eq. (2.14) at r = 1 (i.e., -ln γ_{x0}) and with the r-scaled value from Eq. (3.48) using the effective r of the lattice edge. If the lattice result matches -ln γ_{x0} independent of the microscopic boundary coupling, the universal r = 1 normalization is selected and the concern is defused; if it matches the r-shifted value or varies with boundary coupling, the advertised topological entanglement entropy is non-universal.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central result, Eq. (2.14), states S_same bc = 2N(cl/12ε - ln γ_{x0}), presenting -ln γ_{x0} as the topological entanglement entropy in direct analogy to the modular S matrix. For this to be a topological invariant, γ must be fixed by the bulk order alone. Section 3.3 explicitly undermines that: Eq. (3.48) gives γ_{xc} → γ_{xc}/√\\tilde r and ΔS_EE = -N ln \\tilde r under the edge parameter r appearing in the effective action (A.8). Appendix A states that the parameter matrix V_IJ, hence r, is not determined by the bulk Chern-Simons action. The Z_N examples choose r = 1 to preserve electric-magnetic symmetry, but the Z_pq construction in Section 3.3 requires r_T = q/p to impose the conformal boundary conditions (3.53), and the authors note this yields a half-linking matrix with different normalization than the unitary one in [14]. Thus the claimed universal quantity depends on a microscopic junction or edge parameter; the paper's own computation shows the normalization is not fixed by the bulk topological order. The central claim is therefore only true after an additional, non-topological choice is made, exactly as flagged in the paper's conclusion.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies entanglement entropy in 2+1 dimensional Abelian Chern-Simons theories with gapped boundaries, in the case where the entanglement cut touches or cuts through a physical boundary. The authors argue from bulk-boundary correspondence that the Rényi entropies involve twisted characters, and that the topological contribution to the entanglement entropy is controlled by the half-linking matrix γ_xc, in analogy to the role played by the modular S matrix on closed surfaces. They illustrate the claim with explicit Z_N toric-code examples (electric/magnetic, electric/electric, magnetic/magnetic boundary conditions), a generic 2×2 K-matrix analysis, and an extended Z_pq example. They also construct Ishibashi/Cardy-type states from γ_xc and discuss a condensed-confined duality. The central formula is Eq. (2.14), S_same bc = 2N(cl/12ε − ln γ_x0), and the paper emphasizes that the normalization of γ can be altered by the edge parameter r, shifting the apparent topological entropy by a non-universal amount.","tokens_in":23654,"tokens_out":4538,"duration_ms":43737,"significance":"If the claimed universality held, the paper would establish a boundary analogue of the Kitaev-Preskill/Levin-Wen topological entanglement entropy, replacing the modular S matrix by the half-linking matrix when the cut meets a gapped boundary. The explicit Abelian computations are a real strength: boundary conditions are imposed, modes are quantized, Ishibashi states are constructed, and the modular transformations are performed with Poisson resummation; the e+e and m+m results match the earlier lattice computation [3]. The closed-string picture and the condensed-confined duality are useful and go beyond the existing literature. However, the central universality claim is not established because the normalization of γ, and hence the value of −ln γ_x0, depends on a choice of edge parameter that is not fixed by the bulk topological order; the paper's own Section 3.3 demonstrates this dependence explicitly. The contribution is therefore significant but requires either a proof that the symmetric choice r=1 is forced, or a reformulation of the result as a conditional statement about a chosen junction regularization.","major_comments":[{"comment":"The central claim expressed in Eq. (2.14), that the topological entanglement entropy for a cut touching a gapped boundary is −ln γ_x0 per cut, is not invariant under the edge parameter r introduced in Appendix A. Equation (3.48) shows γ_xc → γ_xc/√\\tilde r and ΔS_EE = −N ln \\tilde r, and Appendix A states that V_IJ, hence r, is not fixed by the bulk Chern-Simons action. The paper's own Z_pq construction in Section 3.3 requires r_T = q/p to impose conformal boundary conditions and produces a half-linking matrix with a different normalization from the unitary one in [14]. Until the symmetric choice r=1 is shown to be forced by topological invariance, or the result is restated as conditional on a concrete edge definition, the advertised replacement of the S matrix by the half-linking matrix is not established.","section":"§3.3, Eq. (3.48)"},{"comment":"The computation assumes that the endpoint of the entanglement cut obeys the same conformal boundary condition as a physical gapped boundary, Eq. (3.1), with the normalization r=1. The introduction explicitly acknowledges that the precise treatment at the junction affects the topological entanglement entropy. The recovery of the lattice results [3] for the e+e and m+m cases is strong evidence for those specific boundary conditions, but it does not select the junction condition for general boundaries such as the Z_pq case. The paper should either derive this junction condition from the bulk topological data or state clearly that the resulting −ln γ is a property of a chosen junction regularization rather than a universal topological invariant.","section":"§3.1, Eq. (3.1)"}],"minor_comments":[{"comment":"The word 'incontractible' appears twice in the paragraph beginning 'In e+e/m+m case GSD=N'; it should be 'non-contractible'.","section":"§3.1"},{"comment":"The quantity \\tilde r is defined only as 'the least positive number which makes r\\tilde r a perfect square'; this definition is ambiguous for general rational r and should be stated explicitly, for example in terms of the squarefree part of r.","section":"§3.3"},{"comment":"The convention for q changes between Eq. (2.3), where q = exp(iτ) with τ = i2πϵ/l, and the discussion after Eq. (3.9), where q is set to exp(−8πϵ/2l); the factor of 2 from the closed-string doubling should be explained more prominently to avoid confusion.","section":"§3.0.1"},{"comment":"The phrase 'the Majorana mode supposedly trapped at the junction' is informal; since the trapping is a stated physical result, either cite the proof or phrase it as a claim to be established.","section":"§3.0.1"}],"recommendation":"major_revision","confidential_remarks":"The paper is a companion to [5] and leans on the half-linking matrix and defect Verlinde formula of [14]; the circularity concern is mitigated because the Abelian examples rederive the relevant matrices from edge-mode quantization and Poisson resummation. The main issue is not the technical execution but the gap between the abstract's universality claim and the demonstrated r-dependence; this should be resolvable by a careful restatement or by a proof that the symmetric normalization is the unique one consistent with topological invariance."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: the paper earns its place as the boundary analog of the modular-S computation of topological entanglement entropy, but the advertised universality in Eq. (2.14) is real only after a choice of boundary condition that the bulk topological order does not fix. That caveat is in the paper itself, mostly in Sec. 3.3, so this is an honest limitation rather than a hidden flaw.\n\nWhat is new: the twisted-character argument for cuts ending on gapped boundaries, the half-linking matrix as the object that replaces S, and the generalized Cardy states in Eq. (2.9) for non-diagonal RCFTs. The Abelian Chern-Simons examples are worked through in real detail: boundary conditions imposed by hand, modes quantized, modular transforms done by Poisson resummation. The e+m result with the Majorana zero mode and the e+e/m+m results recovering the lattice computation of Ref. [3] are concrete, checkable payoffs. The condensed-confinement duality in Sec. 3.2 is a nice extra observation, with a D(S3) example in the appendix.\n\nSoft spots: the general formula (2.12)-(2.14) is argued by analogy and bulk-boundary correspondence, not derived to the same standard as the Abelian examples. The gamma matrix is imported from the same group's companion paper [14]. Neither of those is disqualifying: the Abelian calculations re-derive gamma from edge-mode quantization, and the [3] agreement is an independent benchmark. The larger issue is the one the stress-test flags. Sec. 3.3 shows that an edge parameter r, not fixed by the bulk Chern-Simons action, rescales gamma and shifts the entropy by -N ln r_tilde (Eq. (3.48)). At the symmetric point r=1 the matrix is unitary and -ln gamma_x0 has the advertised meaning; away from it, the result changes. The paper says this in the conclusion, but the Introduction and Eq. (2.14) present the S-matrix analogy without the qualification. That should be fixed.\n\nThe citation pattern is self-heavy, but it is legitimate here: Part I and Ref. [14] are the actual predecessors, and the lattice result they match is external. I do not see a circularity problem.\n\nWho it is for: people computing entanglement entropy in topological orders with gapped boundaries, and people working on generalized boundary states in CFT. It deserves a serious referee. I would send it out, with the request that the authors qualify the central claim to say the universal statement holds at the symmetric point, and either prove or clearly delimit the general formula.","headline":"Solid companion paper that makes the explicit Abelian case for half-linking-matrix control of boundary TEE; the advertised universality of Eq. (2.14) is qualified by the paper's own r-dependence in Sec. 3.3.","tokens_in":24194,"tokens_out":3006,"would_cite":true,"duration_ms":102482,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper argues that when an entanglement cut ends on a gapped boundary, the topological entanglement entropy is fixed by the half-linking matrix $\\gamma_{xc}$, which takes over the role the modular $S$ matrix plays for cuts without…","keywords":["topological entanglement entropy","gapped boundaries","half-linking number","Ishibashi states","Abelian Chern-Simons theory","anyon condensation","twisted characters","boundary conformal field theory"],"falsifier":"In a lattice realization of the $Z_N$ toric code with electric boundaries at both ends, compute the von Neumann entropy of a strip whose cut touches both boundaries: the claim predicts a topological part of $(1/2)\\ln N$ per cut, independent of the microscopic edge term. Alternatively, in the effective edge theory, changing the edge parameter $r$ should shift the entropy by exactly $-N\\ln\\tilde r$ as in Eq. (3.48); either observation going otherwise would falsify the $\\gamma_{xc}$ control.","tokens_in":23150,"feed_emoji":"✂️","tokens_out":15081,"duration_ms":704835,"temperature":0.7,"pith_summary":"This paper asks what controls the topological entanglement entropy of a 2+1-dimensional topological phase when the entanglement cut is drawn so that it starts and ends on gapped boundaries, the physical edges where certain anyons condense and cease to be conserved. The authors argue that in that geometry the topological part of the entropy is governed not by the bulk modular $S$ matrix but by the half-linking number $\\gamma_{xc}$, the same quantity that diagonalizes defect fusion rules. For a strip whose cut touches two boundaries carrying the same condensate, each cut contributes $-\\ln\\gamma_{x0}$ to the entropy on top of the area term; when the two boundaries carry different condensates, extra terms such as a $\\log\\sqrt{2}$ Majorana contribution appear. If correct, this turns boundary data into a computable, geometry-dependent probe of the bulk topological order and extends the Ishibashi-state technology to open systems with physical edges.","feed_headline":"Half-linking numbers govern entanglement when cuts reach a boundary","feed_subtitle":"Boundary-touching cuts replace the S-matrix rule with the half-linking matrix γ.","key_machinery":"The load-bearing object is the half-linking matrix $\\gamma_{xc}$: the transformation between the Wilson-loop basis labelled by confined sectors $x$ and the Wilson-line basis labelled by shared condensed sectors $c$ on a cylinder with two gapped boundaries; in Abelian Chern-Simons theories it equals $S_{xc}/\\sqrt{S_{0c}}$, where $S$ is the modular $S$ matrix. This matrix fixes the modular transformation of the twisted characters that appear in the Renyi entropy, and it enters the construction of Cardy-like boundary states $|B_x\\rangle=\\sum_c (\\gamma_{xc}/\\sqrt{\\gamma_{0c}})|c\\rangle_L\\otimes|c\\rangle_R$, which provide the closed-string picture. The Ishibashi state at each fake cut is the other half of the machinery: it glues the two sides of the cut and produces the reduced density matrix whose trace powers are the twisted characters.","core_discovery":"The central claim is that, for a cylinder of 2+1-dimensional topological order with gapped boundaries, bi-partitioned by a strip whose vertical entanglement cuts terminate on the boundaries, the reduced density matrix is assembled from Ishibashi states whose trace powers yield twisted characters, and the modular transformation of these characters is governed by the half-linking matrix, $\\chi_x(\\tilde q)=\\sum_c \\gamma_{xc}\\chi_c(q)$. The topological entanglement entropy is then $-\\ln\\gamma_{x0}$ per cut when both boundaries share the same condensate, in direct analogy to the role of the modular $S$ matrix in the absence of boundaries. The paper verifies this by explicit open-string quantization of edge modes in the $Z_N$ toric code for electric/magnetic, electric/electric, and magnetic/magnetic boundary conditions (including the extra $\\sqrt{2}$ from a Majorana mode in the electric-plus-magnetic case), by constructing closed-string boundary states from $\\gamma_{xc}$ that need not be diagonal rational-conformal-field-theory boundary states, and for generic $2\\times 2$ $K$-matrix Abelian Chern-Simons theories, where a Poisson-resummation argument identifies the transformation matrix $\\sigma_{xc}$ with $\\gamma_{xc}$ and proves a condensed-confinement duality. It also shows that the normalization of $\\gamma$ can be altered by the edge parameter $r$; the unitary normalization is singled out at the symmetric point $r=1$, and other choices shift the topological entropy by $-N\\ln\\tilde r$.","pith_inferences":["If boundary-touching topological entanglement is genuinely controlled by $\\gamma_{xc}$, then the entropy of an open geometry is a family of boundary-sensitive invariants labelled by pairs of condensates; measuring it on cylinders with different boundary pairs could be used to extract the Lagrangian algebra of an unknown topological phase.","The $r$-dependence of the topological term suggests that, in any concrete material, boundary-touching entanglement measures a combination of bulk order and edge-junction data; comparing a measured value with the unitary prediction would test whether the junction sits at the symmetric point.","The same twisted-character and half-linking machinery should apply, via the folding trick, to entanglement cuts that pass through gapped interfaces rather than terminate on boundaries; computing interface entanglement this way gives a concrete prediction that lattice simulations can check."],"forward_implications":["For cuts ending on two boundaries with the same condensate, the topological entropy is $-\\ln\\gamma_{x0}$ per cut; in the $Z_N$ toric code with electric-electric or magnetic-magnetic boundaries this gives $(1/2)\\ln N$ per cut, hence $\\ln N$ for the strip.","When the cut ends on two different condensates (electric and magnetic in $Z_N$), the ground state is unique and a Majorana zero mode at each junction contributes an extra $-\\ln\\sqrt{2}$, so the topological term is $-\\ln 2$ for the strip.","The half-linking matrix yields closed-string boundary states for non-diagonal RCFTs, so the Ishibashi/Cardy machinery applies to boundary conditions not captured by diagonal rational conformal field theories.","For generic $2\\times 2$ $K$-matrix Abelian theories, the same formula holds with $\\gamma_{xc}$ identified with the unitary matrix $\\sigma_{xc}$ from the lattice resummation; the condensed-confinement duality guarantees the construction is symmetric under swapping condensate and confinate.","In the doubled-theory case $B=C\\boxtimes\\bar C$ with a diagonal Lagrangian algebra, $\\gamma_{xc}$ reduces to $S^C_{xc}$, so the boundary-touching formula reproduces the familiar closed-surface result after unfolding."],"supporting_citations":[{"why":"Defines the half-linking matrix and the defect Verlinde formula that diagonalizes defect fusion rules; the central object claimed to control the entropy.","marker":"[14]"},{"why":"Part I of the authors' own work: establishes the Ishibashi-state computation of topological entanglement and the anyon-condensation description of gapped boundaries used throughout.","marker":"[5]"},{"why":"Introduces the edge-theory/Ishibashi-state strategy of treating the entanglement cut as a fake boundary, the method this paper adapts to boundary-touching cuts.","marker":"[17]"},{"why":"Lattice-model computation of topological entanglement with boundaries that the present open-string results reproduce in the electric-electric and magnetic-magnetic cases.","marker":"[3]"},{"why":"Supplies the twisted characters and their modular properties in rational conformal field theory, the open-string frame used in Section 2.","marker":"[13]"},{"why":"Provides the defect theory of Abelian topological states and the Majorana zero modes at electric-magnetic junctions invoked for the $\\sqrt{2}$ contribution.","marker":"[18]"},{"why":"Defines topological boundary conditions in Abelian Chern-Simons theory through Lagrangian subgroups and the mutual-null condition, used to characterize condensates.","marker":"[21]"},{"why":"Shows how to gap edge modes and extend the K matrix by trivial channels, needed for the $Z_{pq}$ example with enlarged boundary theory.","marker":"[22]"}],"fun_headline_variants":["Half-linking matrix replaces S-matrix for boundary-touching cuts","Twisted characters and Ishibashi states fix boundary entropy","Boundary cuts: half-linking number controls topological entanglement","Ishibashi states yield twisted characters for gapped-boundary entanglement","Gamma-matrix, not S-matrix, rules entropy for boundary-terminating cuts"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that an entanglement cut ending on a gapped boundary behaves exactly like a physical gapped boundary, with a particular symmetric choice of edge normalization; if the cut's endpoint is not pinned by the bulk topological order, the claimed topological entropy is not universal.","fun_headline_variants_meta":{"raw":{"variants":["Half-linking matrix replaces S-matrix for boundary-touching cuts","Twisted characters and Ishibashi states fix boundary entropy","Boundary cuts: half-linking number controls topological entanglement","Ishibashi states yield twisted characters for gapped-boundary entanglement","Gamma-matrix, not S-matrix, rules entropy for boundary-terminating cuts"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000796,"raw_usage":{"total_tokens":3532,"prompt_tokens":1005,"completion_tokens":2527,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":621,"completion_tokens_details":{"reasoning_tokens":2435}},"tokens_in":621,"tokens_out":2527,"duration_ms":21036,"temperature":1.0,"reasoning_tokens":2435,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T12:00:16.467249+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"In a lattice realization of the $Z_N$ toric code with electric boundaries at both ends, compute the von Neumann entropy of a strip whose cut touches both boundaries: the claim predicts a topological part of $(1/2)\\ln N$ per cut, independent of the microscopic edge term. Alternatively, in the effective edge theory, changing the edge parameter $r$ should shift the entropy by exactly $-N\\ln\\tilde r$ as in Eq. (3.48); either observation going otherwise would falsify the $\\gamma_{xc}$ control.","supporting_citations":[{"cited_title":"A Defect Verlinde Formula","cited_arxiv_id":"1901.08285","evidence_quote":"Defines the half-linking matrix and the defect Verlinde formula that diagonalizes defect fusion rules; the central object claimed to control the entropy."},{"cited_title":"Ishibashi states, topological orders with boundaries and topological entanglement entropy. part i,","cited_arxiv_id":null,"evidence_quote":"Part I of the authors' own work: establishes the Ishibashi-state computation of topological entanglement and the anyon-condensation description of gapped boundaries used throughout."},{"cited_title":"Entanglement Entropy of Topological Orders with Boundaries","cited_arxiv_id":"1804.05725","evidence_quote":"Lattice-model computation of topological entanglement with boundaries that the present open-string results reproduce in the electric-electric and magnetic-magnetic cases."}],"review_version":1}