{"id":"7cd67ebc-9588-425c-93a5-61111a4a307d","arxiv_id":"2412.07563","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"A tensor-network anomaly indicator detects strong, weak, and intrinsic subsystem symmetry-protected topological phases, and persists for average symmetries in mixed states.","lead":"The paper develops a tensor network based anomaly indicator that detects boundary anomalies of linear subsystem symmetries in 2D subsystem symmetry-protected topological (SSPT) phases, and extends it to mixed states with average symmetries. A smart generalist would read it because it offers a numerical method to identify and distinguish topological phases stabilized by subsystem symmetries, including a new intrinsic Z2 phase and disorder-tolerant mixed-state phases.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Eq. (16) is proven only under the adjacent-factorization ansatz W^Right(g[y])=V_y(g[y])V_{y+1}(g[y]) (Eq. B2); the claimed extension to 'more complex cases' is unproved, so the indicator is not established as a general detector of subsystem-symmetry boundary anomalies.","rationale":"Read in good faith, the paper does what it claims for adjacent, factorizable boundary anomalies. The cluster-state derivation in Appendix A is self-consistent and gives Q = -1, matching Eq. (16); the tunable-tensor numerics are consistent at gamma = +/-1; and the intrinsic Z2 example has the expected 2^{Ly-1} entanglement degeneracy. The weakest point is not the fixed-point algebra but the generality of the indicator. Every step that proves Q = phi, namely Eq. (B16), the trace collapse in Eq. (15), and the claim that defect insertion in the overlapping region suffices, uses W^Right(g[y]) = V_y(g[y])V_{y+1}(g[y]). Without this factorization, W^Right W^Left need not cancel and the phase from non-overlapping components never enters the numerator. The paper explicitly restricts to adjacent subsystems in the abstract and Sec. II A, but also asserts, without evidence, that larger regions can be treated 'using a similar approach.' That extension is not a corollary of the presented proof: the number of bonds covered by the twisted-sector defect is fixed by the ansatz, so longer-range cocycles are invisible to the published indicator. This does not invalidate the cluster-state or intrinsic-Z2 results, but it means the headline method is a partial detector rather than a general one. The reader's conditional verdict already captures this; the suggested tensor construction would make the limitation concrete.","tokens_in":27072,"tokens_out":28653,"duration_ms":280255,"concrete_test":"Construct a Ly=4, translation-invariant PEPS whose right-boundary representation is W^Right(g[y]) = X_{Lx,y} Z_{Lx,y+2} (a non-adjacent, three-row-window projective representation of G_h = prod_y Z_2) and which satisfies the symmetry condition Eq. (8). Compute Q(g[1], f[2]) exactly as defined in Eq. (10) using the transfer-matrix formula (C3). The boundary algebra gives {W^Right(g[1]), W^Right(f[3])} = 0, i.e. phi(g[1], f[3]) = -1, while no anomaly is expected between rows 1 and 2. If the computed Q(g[1], f[2]) converges to +1 rather than exposing the -1 anomaly of the state, the indicator is blind to non-adjacent anomalies and Eq. (16) cannot be extended beyond ansatz (B2). If such a tensor cannot be built with finite local bond dimension, the ansatz follows from locality and the concern is vacuous.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The identity Q(g[y], f[y+1]) = phi(g[y], f[y+1]) in Eq. (16) is the central result. Its proof passes through Eq. (B16), which converts the mixed anomaly into a commutation relation between V_{y+1}(g[y]) and V_{y+1}(f[y+1]) on a single virtual bond. This conversion uses the ansatz W^Right(g[y]) = V_y(g[y])V_{y+1}(g[y]) (Eq. B2) and the cancellation W^Right W^Left = I needed for the trace collapse in Eq. (15). If the boundary representative has support on a wider window, e.g. W^Right(g[y]) = V_y(g[y])V_{y+2}(g[y]), then the twisted-sector defect at bond y+1 does not overlap the full support of W^Right(g[y]); the V_{y+2} factor commutes through the trace and contributes no phase, so Q need not equal the true boundary anomaly phi(g[y], f[y+2]). The paper asserts in Sec. II A that 'expanding the region of investigation' covers such cases, but provides no construction, proof, or numerical example. The analytic examples (cluster state, intrinsic Z2) and the Ly=2 numerics all lie inside the adjacent-factorizable class, so the concern is one of scope rather than an error in those fixed-point computations.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a tensor-network anomaly indicator Q(g[y], f[y+1]) for two-dimensional subsystem symmetry-protected topological (SSPT) phases. A twisted-sector state is formed by inserting a symmetry defect V_{y+1}(f[y+1]) on one virtual bond of a cylinder, and Q is defined as the expectation value of the horizontal subsystem symmetry S_h(g[y]) in that state. Under the assumption that boundary subsystem symmetry operators factorize into adjacent local operators, W^Right(g[y]) = V_y(g[y]) V_{y+1}(g[y]), the paper proves Q = φ(g[y], f[y+1]) (Eq. 16), i.e., the indicator equals the mixed anomaly of adjacent row symmetries. The identity is verified analytically for the 2D cluster state and numerically with transfer-matrix calculations on Ly=2 cylinders for a tunable tensor T(γ), which distinguishes strong (γ=-1) and weak (γ=1) Z2^τ x Z2^σ SSPT phases. The authors also introduce a fixed-point tensor T_Z2(φ) with Z2 subsystem symmetry, identify an intrinsic Z2 SSPT phase without a weak counterpart, and detect it through Q=-1 and a degenerate entanglement spectrum. The method is further extended to mixed-state density matrices with average and exact subsystem symmetries in uniform and alternating disorder ensembles, with numerical results for an interpolated density matrix ρ(t).","tokens_in":27446,"tokens_out":15098,"duration_ms":147880,"significance":"The central identity is a useful and nontrivial transfer-matrix characterization for the adjacent-subsystem class of SSPT boundary anomalies. The strengths of the paper are the analytic proof of Q=φ under the stated ansatz, the explicit cluster-state example, and the set of reproducible transfer-matrix numerics that separate strong, weak, and intrinsic phases. The mixed-state extension is a reasonable bridge between pure-state SSPT anomalies and average-symmetry anomalies, and the identification of an intrinsic Z2 SSPT phase is potentially interesting. The main caveat is that the established result is restricted to boundary operators of the adjacent-factorized form (Eq. B2); the broader generality statements in the title, Sec. II A, and the conclusion go beyond what is proven, and the stability claim for the intrinsic phase is asserted rather than demonstrated.","major_comments":[{"comment":"The central identity Q(g[y], f[y+1]) = φ(g[y], f[y+1]) (Eq. 16) is proven only for boundary subsystem symmetry operators satisfying W^Right(g[y]) = V_y(g[y]) V_{y+1}(g[y]) (Eq. B2). The proof passes through Eq. (B16), which reduces the mixed anomaly to a commutation relation of the single-bond operator V_{y+1}(g[y]) with V_{y+1}(f[y+1]); if a boundary representative instead has support on a wider window, for example W^Right(g[y]) = V_y(g[y]) V_{y+2}(g[y]), then the V_{y+2} factor commutes through the trace in Eq. (15) and Q need not equal the actual boundary anomaly φ(g[y], f[y+2]). The statement in Sec. II A that \"by expanding the region of the investigation... the analysis is also applicable to more complex cases\" is not supported by any proof, construction, or numerical example, and the Ly=2 numerics in Figs. 3, 5, and 9 cannot test non-adjacent boundary anomalies. The scope restriction should therefore be stated as a hard limitation, or an explicit proof and example for wider operator support should be supplied.","section":"Sec. II A, Eq. (B2); Appendix B, Eq. (B16); Eq. (16)"},{"comment":"The mixed-state indicators Qexa and Qave are derived in Appendix E for a one-dimensional MPO with global symmetry; the application to 2D subsystem symmetries in Sec. V is not given the same level of proof. I ask the authors to state explicitly how the adjacent-factorization ansatz of Eq. (B2) is inherited by the boundary operators of the Choi-state representation of the PEPDO, and to show the transfer-matrix steps that lead to Qexa(k,~g)=φ and Qave(g,~g)=φ for the subsystem operators in Eqs. (83) and (86). In addition, Eq. (E10) as written has the right conjugation by V(k)⊗I, whereas Eq. (E9) and the trace evaluation in Eq. (E11) appear to require conjugation by V(g)⊗I; please clarify whether this is a typo or a step relying on the group-extension structure. Without this clarification, the numerical values in Fig. 9 do not by themselves establish the claimed mixed-state anomalies.","section":"Sec. V, Appendix E, Eqs. (47), (49), (E10), (E11)"},{"comment":"Section IV B states that the intrinsic Z2 SSPT phase is stable under the subsystem-symmetric perturbation U^v_y(η)=∏_x exp(η X_{x,y}), with φ(g[y],g[y+1]) remaining -1 and the entanglement spectrum remaining fully degenerate \"throughout the nontrivial SSPT phase\". No numerical data or analytic argument for this stability is provided. Because calling T_Z2(φ) at φ=(2m+1)π a phase rather than a fixed point requires a finite region of stability, please include the Q(g[1],g[2]) and entanglement-spectrum results as functions of η, or state that only the fixed-point behavior is established.","section":"Sec. IV B, Eq. (40)"}],"minor_comments":[{"comment":"The conclusion overstates the proven scope by presenting the method as \"a numerical method to detect quantum anomalies of subsystem symmetries\" without repeating the adjacent-subsystem restriction; I recommend qualifying this sentence.","section":"Sec. VI"},{"comment":"The thermodynamic-limit formula for Q involves the ratio (λ0(g[y],e)/λ0(e,e))^{Lx-1}; please specify how the phase of λ0 is treated when the transfer matrix is non-Hermitian, and state under what conditions Q is guaranteed to be real.","section":"Appendix C, Eq. (C4)"},{"comment":"Several tensor-network equations are presented diagrammatically without explicit algebraic definitions of every tensor; providing explicit local tensor components or a short pseudocode would help readers verify the commutation relations that the boundary-anomaly graphs in Eqs. (66) and (80) rely on.","section":"Secs. V A and V B, Eqs. (60), (63), (74), (76)"},{"comment":"The captions do not state the bond dimension (or truncation) used in the entanglement-spectrum calculations, nor the specific Lx values for the horizontal-cylinder geometry; these parameters are needed to assess the claims of full degeneracy and gap closing.","section":"Figs. 4 and 10"}],"recommendation":"major_revision","confidential_remarks":"The derivation of the central identity is sound within the explicitly stated adjacent-subsystem class, and I do not see a basis for rejection. The main revision request is to align the paper's generality claims with the actual proof, to add the missing stability data for the intrinsic Z2 phase, and to fix the apparent index inconsistency in Eq. (E10). The numerical section is modest but adequate for the claims once the scope is stated precisely."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague—\n\nThis is a genuinely useful method paper. The authors take the MPS twisted-sector anomaly indicator from [30,31,72] and adapt it to 2D linear subsystem symmetries. The new pieces: an indicator Q(g[y], f[y+1]) for adjacent subsystem anomalies, a construction of an intrinsic Z2 SSPT state with degenerate entanglement spectrum, and an extension to mixed states with average and exact subsystem symmetries. The cluster state is worked out fully analytically and Q=-1 matches the mixed anomaly; the tunable tensor numerics at Ly=2 show clean convergence to the predicted values at gamma=±1 and phi=nπ. That part is solid.\n\nThe central identity Q=phi is proven under an explicit assumption, stated in the main text and in App. B: the boundary operator W^Right(g[y]) factorizes as V_y(g[y])V_{y+1}(g[y]), so mixed anomalies live only between adjacent rows. The abstract says the same. Within that scope the proof in App. B is plausible and I do not see a gap. The stress-test note is right, though, that the one-line claim 'expanding the region of investigation... more complex cases' is not supported. The indicator is constructed by inserting a defect at a single virtual bond, so it can only detect the commutator of the two V operators on that bond; an anomaly between g[y] and f[y+2] would be missed by construction. The paper does not provide an argument for why all physical boundary anomalies should be adjacent-local. That is a genuine scope limitation and should be stated more conservatively, or proved.\n\nMinor soft spots: numerics are all Ly=2 transfer-matrix spectra, no code or data deposited, and the mixed-state part leans on the same ansatz plus a somewhat more elaborate construction. The appendices are compressed; a referee should re-check the tensor equation signs in App. A and B16, but nothing there looks wrong to me.\n\nOverall: for tensor-network and SSPT people this is a worthwhile method paper, and it is honest about its main restriction. I would send it to peer review, with the request that the authors either prove the extension to non-adjacent anomalies or explicitly frame the indicator as an adjacent-anomaly detector.","headline":"A solid, genuinely new anomaly-indicator method for adjacent-subsystem SSPT phases; central claim holds within its stated scope, but the 'more complex cases' generalization is unproved.","tokens_in":27977,"tokens_out":3165,"would_cite":true,"duration_ms":32407,"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 establishes that the normalized subsystem-symmetry charge of a twisted sector state equals the mixed boundary anomaly of adjacent subsystem symmetries, giving a transfer-matrix detector for strong, weak, and intrinsic SSPT…","keywords":["subsystem symmetry-protected topological phases","boundary anomaly","anomaly indicator","twisted sector state","tensor network","entanglement spectrum","average symmetry","mixed-state anomaly"],"falsifier":"Take a symmetric tensor state whose boundary operator $W^{\\mathrm{Right}}(g[y])$ is supported on rows $y$, $y+1$, and $y+2$, and compute $Q(g[y], f[y+2])$ from the transfer-matrix formula; the paper's assumptions force $\\phi(g[y], f[y+2]) = 1$ and hence $Q = 1$, so observing any value other than $1$ would show the indicator misses non-adjacent mixed anomalies.","tokens_in":26841,"feed_emoji":"⚛️","tokens_out":7879,"duration_ms":76216,"temperature":0.7,"pith_summary":"The paper sets out to show that boundary anomalies of linear subsystem symmetries in two-dimensional states can be read off directly from the bulk wave function. It defines an anomaly indicator $Q(g[y], f[y+1])$ as the normalized subsystem-symmetry charge of a twisted sector state and derives, for symmetric states, that it equals the mixed-anomaly phase $\\phi(g[y], f[y+1])$ between adjacent row symmetries. That equality turns a hard classification question into a transfer-matrix calculation, so one numerical routine can tell strong from weak $Z_2^\\tau \\times Z_2^\\sigma$ SSPT phases, detect an intrinsic $Z_2$ SSPT phase, and, after extension to density matrices, find the same anomalies under uniform or alternating disorder. A sympathetic reader would care because previous invariants such as spurious topological entanglement entropy could flag nontrivial phases but could not distinguish which SSPT phase a state is in.","feed_headline":"One twisted-sector charge flags strong, weak, and intrinsic SSPT phases","feed_subtitle":"The same indicator sees boundary anomalies survive uniform and alternating disorder.","key_machinery":"The central object is the anomaly indicator $Q(g[y], f[y+1])$, a ratio of symmetry charges evaluated from the spectra of transfer matrices with an $f[y+1]$ symmetry defect. The load-bearing identity is that reordering the local virtual-space operators $V_{y+1}(g[y])$ and $V_{y+1}(f[y+1])$ on the shared virtual bond produces the phase $\\phi(g[y], f[y+1])$, and this phase factors out of the transfer-matrix trace to give Eq. (16). The locality decomposition $W^{\\mathrm{Right}}(g[y]) = V_y(g[y]) V_{y+1}(g[y])$ restricts the analysis to mixed anomalies between adjacent subsystems and is what makes the transfer-matrix proof work.","core_discovery":"The central discovery is the relation $Q(g[y], f[y+1]) = \\phi(g[y], f[y+1])$ for symmetric subsystem states (Eq. 16): inserting a zero-dimensional symmetry defect $V_{y+1}(f[y+1])$ on one virtual bond and measuring the charge of a horizontal subsystem symmetry $S_h(g[y])$ in the resulting twisted sector state reproduces the factor-system phase $\\phi$ that labels the mixed anomaly between rows $y$ and $y+1$. The paper verifies this analytically for the 2D cluster state, where $Q = -1$ matches $\\phi(g[y]^\\tau, g[y+1]^\\sigma) = -1$, and numerically for a tunable tensor $\\Gamma(\\gamma)$ that interpolates between strong ($\\gamma = -1$) and weak ($\\gamma = 1$) $Z_2^\\tau \\times Z_2^\\sigma$ SSPT phases. It then constructs an intrinsic $Z_2$ SSPT phase with no weak counterpart, detected by $Q(g[1], g[2]) = -1$ and a fully degenerate entanglement spectrum, and extends the indicator to density matrices via $Q_{\\mathrm{ave}}$ and $Q_{\\mathrm{exa}}$, showing that the anomalies survive both uniform and alternating disorder.","pith_inferences":["The locality assumption suggests a natural generalization the authors do not construct: insert defects spanning $k$ adjacent rows and define $Q(g[y], f[y+k])$ to capture longer-range mixed anomalies between non-adjacent subsystems.","Because the indicator is built from the topological response rather than from any specific Hamiltonian, it should transfer to foliated fracton and higher-order topological phases in 3D, where planar subsystem symmetries replace row symmetries.","The mixed-state version could serve as a practical probe of strong-to-weak spontaneous symmetry breaking in disordered or decohered systems, since $Q_{\\mathrm{ave}}$ jumps between $\\pm 1$ exactly where the entanglement gap closes; this use goes beyond the paper's stated claims."],"forward_implications":["The same transfer-matrix routine separates strong from weak SSPT phases: in the weak $Z_2^\\tau \\times Z_2^\\sigma$ phase, $Q(g[y]^\\tau, g[y+1]^\\sigma) = 1$ while the same-row indicator $Q(g[y]^\\tau, g[y]^\\sigma) = -1$ remains nontrivial.","An intrinsic $Z_2$ SSPT phase exists whose only nontrivial invariant is the adjacent-row mixed anomaly; its entanglement spectrum is fully degenerate and the indicator stays at $-1$ under symmetric perturbations.","When a subsystem symmetry is spontaneously broken, the anomaly indicator decays to zero in the thermodynamic limit, so the method simultaneously detects symmetry breaking as the loss of anomaly signal.","In mixed states with average subsystem symmetries, the exact and average symmetry charges of the twisted density matrix reproduce the mixed anomaly, and the anomaly persists under both uniform and alternating disorders, connecting pure- and mixed-state boundary anomalies."],"supporting_citations":[{"why":"Supplies the 1D MPS anomaly-indicator construction (twisted-sector symmetry charge) that this paper extends to 2D subsystem symmetries.","marker":"[30]"},{"why":"Gives the topological-field-theory and matrix-product-state framework used for the twisted-sector partition functions behind $Q$.","marker":"[31]"},{"why":"Supplies the SSPT classification, the weak/strong equivalence relation, and the factor-system language $\\phi(g[y], f[y+1])$ that the indicator is designed to reproduce.","marker":"[51]"},{"why":"Defines subsystem symmetry-protected order and the boundary projective representations on which the locality assumption rests.","marker":"[50]"},{"why":"Establishes anomaly inflow for subsystem symmetries, justifying the map from bulk SSPT wave functions to boundary anomalies.","marker":"[53]"},{"why":"Provides the mixed-state/MPO anomaly-indicator formalism generalized here to average subsystem symmetries and $Q_{\\mathrm{ave}}$, $Q_{\\mathrm{exa}}$.","marker":"[72]"},{"why":"Supplies the result that average symmetry alone cannot protect a nontrivial topological phase, used to interpret $Q_{\\mathrm{ave}}=1$ as trivial and $-1$ as nontrivial ASSPT.","marker":"[73]"},{"why":"Gives the PEPS entanglement-spectrum formula used to connect degenerate boundary spectra to the anomalous edge theory.","marker":"[17]"}],"fun_headline_variants":["Twisted-sector charge flags strong, weak, and intrinsic SSPT","One charge probe detects all three SSPT orders","Single twisted charge reads out the anomaly phase","Anomaly indicator survives disorder, finds new phase","Charge-phase equality pins down SSPT phases"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole scheme assumes that a boundary subsystem-symmetry operator acts only on two neighboring rows and factorizes as a product of two local operators; if a real phase has boundary anomalies spread across three or more rows, the indicator would fail to see them.","fun_headline_variants_meta":{"raw":{"variants":["Twisted-sector charge flags strong, weak, and intrinsic SSPT","One charge probe detects all three SSPT orders","Single twisted charge reads out the anomaly phase","Anomaly indicator survives disorder, finds new phase","Charge-phase equality pins down SSPT phases"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000982,"raw_usage":{"total_tokens":4179,"prompt_tokens":965,"completion_tokens":3214,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":581,"completion_tokens_details":{"reasoning_tokens":3140}},"tokens_in":581,"tokens_out":3214,"duration_ms":21788,"temperature":1.0,"reasoning_tokens":3140,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T18:41:46.826659+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a symmetric tensor state whose boundary operator $W^{\\mathrm{Right}}(g[y])$ is supported on rows $y$, $y+1$, and $y+2$, and compute $Q(g[y], f[y+2])$ from the transfer-matrix formula; the paper's assumptions force $\\phi(g[y], f[y+2]) = 1$ and hence $Q = 1$, so observing any value other than $1$ would show the indicator misses non-adjacent mixed anomalies.","supporting_citations":[{"cited_title":"Van Acoleyen, N","cited_arxiv_id":null,"evidence_quote":"Supplies the 1D MPS anomaly-indicator construction (twisted-sector symmetry charge) that this paper extends to 2D subsystem symmetries."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the topological-field-theory and matrix-product-state framework used for the twisted-sector partition functions behind $Q$."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the SSPT classification, the weak/strong equivalence relation, and the factor-system language $\\phi(g[y], f[y+1])$ that the indicator is designed to reproduce."},{"cited_title":"Pollmann, S","cited_arxiv_id":null,"evidence_quote":"Defines subsystem symmetry-protected order and the boundary projective representations on which the locality assumption rests."},{"cited_title":"Huang, L","cited_arxiv_id":null,"evidence_quote":"Establishes anomaly inflow for subsystem symmetries, justifying the map from bulk SSPT wave functions to boundary anomalies."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the mixed-state/MPO anomaly-indicator formalism generalized here to average subsystem symmetries and $Q_{\\mathrm{ave}}$, $Q_{\\mathrm{exa}}$."}],"review_version":1}