{"id":"5c3ca9c0-9f16-4f00-a3b4-91e68e26fdc7","arxiv_id":"2412.20546","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Anomaly-free fusion category symmetries have a canonical trivial phase, and the three Rep†(D8) symmetry-protected topological phases are explicitly realized by Q-system lattice models connected by an S3 duality.","lead":"This paper builds simple lattice models for phases of matter protected by non-invertible 'fusion category' symmetries, including a trivial product-state phase that earlier work thought impossible. It gives explicit Hamiltonians for all three Rep†(D8) SPT phases and a duality that permutes them.","discovery_kind":"first_principles","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The classification relies on an extra minimality/onsite postulate: if non-minimal dualizable MPOs are admitted, the trivial phase and Q-system classification shift. The pre-reduction cluster MPO D must be tested for dualizability.","rationale":"The reader's weakest assumption is essentially the same definitional issue: the pair (C,f) and the onsite bond-dimension condition are what produce the canonical trivial phase and the Q-system classification. My concern sharpens that issue: the paper's justification of Eq. (2) via dualizability depends on already choosing a minimal MPO presentation, and the cluster-state MPO D in Appendix A is a concrete case where a non-minimal representation appears before reduction. The paper does not determine whether D satisfies the dualizability axiom, so it leaves open whether the onsite condition is a derived physical constraint or an external convention. This is not an internal inconsistency; the D8 fixed-point calculations are explicit and checkable, and the Q-system machinery is coherent once the pair (C,f) is fixed. But the abstract states the Q-system classification as a general result, while the proof establishes it only for Q-system fixed-point models. That gap is addressable, so a conditional verdict remains appropriate. I do not recommend REJECT because the central Rep†(D8) construction is explicit and the missing dualizability check is a well-defined computational task. I partially agree with the reader because my concern centers on the minimality/dualizability relation and the completeness of the classification rather than solely on the philosophical choice of what counts as a symmetry.","tokens_in":52414,"tokens_out":9469,"duration_ms":103846,"concrete_test":"Take the MPO D from Appendix A before the CZ conjugation, i.e. the four-dimensional-bond representation of Rep†(D8) used in Ref. [35]. Construct the dual tensor D* and directly check the four dualizability conditions of Eq. (1) together with the pulling-through condition on an irreducible charge sector such as σ. If D passes, the onsite/minimal condition is an extra postulate and the Q-system classification should be re-run with v_σ = 4 to see whether the product state and the three phase labels are still distinguished. If D fails, the cluster-state MPO is not a legitimate symmetry operator under the paper's own criterion, and the definitional dispute is resolved in the paper's favor.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is that SPT phases with an anomaly-free fusion category symmetry are classified by Q-systems A in the charge category with fgt A a matrix algebra. The load-bearing step is not the algebraic bijection in Theorem VI.4 but the identification of physical symmetry actions with the 'onsite' minimal MPO representations defined by Eq. (2). The paper argues this condition follows from dualizability: the fusion equations for the virtual bond dimensions v_s have, for Rep†(D8), the unique integer solution v_s = d_s. But this only holds after one has already chosen the minimal representation. The cluster-state symmetry operator D discussed in Appendix A has v_σ = 4, not 2, and the appendix does not test whether D satisfies dualizability Eq. (1) before reducing it by a CZ conjugation. If D is dualizable and satisfies pulling-through, then the same algebraic symmetry data admit two different UV realizations, one with a canonical product state and one without; the Q-system/trivial-phase classification and the S3-duality statement are then properties of the chosen presentation, not of the fusion category symmetry itself. If, instead, D fails dualizability, the paper's criterion is a genuine selection rule; but this is not shown. Either way, the abstract's unconditional classification exceeds what is proved: Proposition V.3 applies to Q-system fixed-point models, not to arbitrary (C,f)-symmetric gapped Hamiltonians, and no completeness argument is supplied.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a microscopic framework for (1+1)d SPT phases with anomaly-free fusion category symmetry, defined as a pair (C,f) of a unitary fusion category and a fiber functor. The authors argue that the fiber functor determines an onsite MPO symmetry action, a charge category, and a canonical trivial phase. They classify SPT phases by Q-systems A in the charge category whose image under the forgetful functor is a matrix algebra in Hilb. Explicitly, they construct commuting-projector Hamiltonians realizing three Rep†(D8) SPT phases—one a product state and two with edge modes—and exhibit an S3-duality acting on these phases. They also compare their constructions with the cluster-state models of [35], claiming equivalence after reducing the cluster MPO to onsite form.","tokens_in":52707,"tokens_out":8854,"duration_ms":95959,"significance":"If the classification claim can be made fully rigorous, the paper gives a concrete algebraic handle on non-invertible SPTs: phases correspond to Morita classes of Q-systems, and the trivial phase is canonically fixed by the fiber functor. The Rep†(D8) models are exactly solvable commuting projectors; Propositions V.2–V.4 provide a clean operator-algebraic proof of the fixed-point ground-state characterization, including edge-mode stability. The explicit S3-duality and the reduction of the cluster-state example are valuable constructive results. The main gaps are the completeness of the fixed-point classification and the status of the minimal/onsite MPO condition, both of which are load-bearing for the abstract's unconditional statement.","major_comments":[{"comment":"The text claims that dualizability plus the fusion rule forces v_s = d_s (Eq. (2)), but the argument assumes the MPO has already been brought to its minimal closure. The cluster-state MPO D in Appendix A has v_σ = 4, and the appendix reduces it to a two-dimensional bond dimension by conjugating with CZ without first verifying that D satisfies the dualizability condition Eq. (1). If D is dualizable, the same categorical data admit two different UV realizations, one with a canonical product state and one without, making the Q-system classification and the existence of a canonical trivial phase properties of the chosen presentation; if D is not dualizable, that needs to be shown, since otherwise the claimed derivation of Eq. (2) does not go through. Either way, the abstract's unconditional classification exceeds what is proved.","section":"II.A, Eq. (2); Appendix A"},{"comment":"Proposition V.3 characterizes the ground state of the Q-system fixed-point model, not the ground state of an arbitrary (C,f)-symmetric gapped Hamiltonian. The classification statement in the abstract ('an SPT phase corresponds to a Q-system...') requires a completeness argument showing that every gapped phase with a unique symmetric ground state can be deformed, within the symmetric phase, to a Q-system model. No such argument is given, and the operator-algebraic results in Section V only establish the one-way statement for the fixed-point family. This is a load-bearing gap between the theorem proved and the classification claimed.","section":"V.A, Proposition V.3; abstract"},{"comment":"The S3-duality transformation imposes the Gauss law energetically with J >> 1. Footnote 56 acknowledges that this should be a kinematic constraint. For finite J, the constrained subspace is not exactly invariant under the Hamiltonian, and the claim that the transformed Hamiltonian 'shares the same energy spectrum' and remains gapped is not justified: second-order processes through the high-energy sector can renormalize the low-energy Hamiltonian and, in principle, change the phase. To substantiate the lattice duality, either impose the constraint exactly at the Hilbert-space level or provide a uniform gap estimate showing that the effective low-energy theory is exactly the original H for all J above some finite threshold.","section":"II.D, footnote 56"}],"minor_comments":[{"comment":"The notation n_s(g) and n_r(g) is used before being defined; the number-parity maps should be introduced explicitly before Eq. (1) and Table I.","section":"II.A"},{"comment":"The maximally entangled state in Eq. (6) is written as a vector in W*⊗W, but in Section II.D it is used as a projector on a link between (i,R) and (i+1,L); the orientation and identification of the dual basis should be clarified.","section":"II.B, Eq. (6); II.D"},{"comment":"Phase-equivalence is defined via equivalence of C-modules Hilb_f and Hilb_h, but the preceding theorem and corollaries are phrased in terms of monoidal equivalences π of C; the relation between these two notions should be stated explicitly.","section":"VI, Definition VI.5"},{"comment":"The deformation path L∘γ is claimed to connect fixed-point models without a phase transition, but only the gap of each individual model is computed; a uniform (or at least pathwise positive) lower bound on the gap along the path is not given.","section":"G.4"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is substantial and the fixed-point constructions are careful. The main concern is that the classification statements in the abstract and Section VI are more general than the theorems proved; I would ask the authors to narrow the claims or supply the missing completeness and minimality arguments. The note added acknowledges concurrent work [88], so I have not treated overlap as an issue."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: the D8 construction is the real thing—explicit, checkable, and it resolves the dispute over whether non-invertible SPTs can have a trivial phase. The general classification claim, however, is not as general as the abstract makes it sound.\n\nWhat is new and good: the commuting-projector Hamiltonians realizing all three Rep†(D8) SPT phases, with a product state as the trivial phase, are a concrete advance. The comparison with the cluster-state models [35] is careful and shows those models become the same phases (including trivial) after a symmetry-preserving local unitary. The Q-system criterion—an SPT phase corresponds to a Q-system whose image under the forgetful functor is a matrix algebra—is clean and is proved for fixed-point Q-system models in Section V. The categorical theorems in Section VI are solid as far as I can tell, and the S3-duality construction gives a nice lattice realization of the fiber-functor torsor.\n\nThe soft spot is the definition of 'onsite' and the move from fixed-point models to a general classification. The paper claims dualizability forces virtual bond dimension equal to quantum dimension, but the argument only shows that a closed set of dualizable MPOs satisfying the fusion equation has the unique integer solution vs=ds. It does not prove that every dualizable representation of the symmetry is closed in this sense. In Appendix A, the cluster MPO D has bond dimension 4 before reduction; the authors never check whether D itself satisfies the dualizability condition (Eq. 1) before they conjugate it by CZ. If D is dualizable, then the same categorical symmetry data admit two different UV realizations, one with a canonical product state and one without—and the Q-system classification would be a property of the presentation, not of the symmetry. If D is not dualizable, that is easy to say and would settle the point. Either way, Proposition V.3 is a statement about Q-system fixed-point models, not about arbitrary symmetric gapped Hamiltonians, and no completeness argument is supplied. The abstract's unconditional classification overreaches.\n\nThe S3-duality section also imposes the Gauss law energetically, which the authors themselves note is a kinematic constraint; it is a reasonable low-energy statement, but not an exact lattice equivalence.\n\nThis paper deserves serious peer review. The referee should ask the authors to clarify the status of the onsite criterion and either prove the completeness claim or scale it back. The D8 construction will be cited regardless.","headline":"Explicit D8 construction settles the trivial-phase question, but the general Q-system classification claim needs a tighter proof.","tokens_in":53248,"tokens_out":5657,"would_cite":true,"duration_ms":58229,"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":"An anomaly-free fusion category symmetry is a pair $(\\mathcal{C}, f)$, and every (1+1)d SPT phase is a Q-system in the charge category that becomes a matrix algebra when the symmetry is forgotten.","keywords":["non-invertible symmetry","fusion category symmetry","symmetry-protected topological phases","fiber functor","Q-system","matrix product operator","Rep†(D8)","lattice model"],"falsifier":"Numerically compute the ground-state degeneracy on a half-infinite open chain at the two non-trivial exactly solvable points $(\\lambda_0,\\lambda_1,\\lambda_2)=(0,1,0)$ and $(0,0,1)$ of the Section II model: the paper predicts a two-fold degenerate edge mode at both, and none at $(1,0,0)$. A tensor-network calculation showing any other degeneracy pattern, or showing that the $S_3$-duality maps the phases differently from the coset action on $S_3/K$, would settle the claim.","tokens_in":52162,"feed_emoji":"⚛️","tokens_out":10777,"duration_ms":90899,"temperature":0.7,"pith_summary":"This paper argues that a non-invertible fusion category symmetry in one spatial dimension is not fully specified by a fusion category $\\mathcal{C}$ alone; the microscopic definition must include a fiber functor $f:\\mathcal{C}\\to\\mathrm{Hilb}$, which fixes how the symmetry operators act on a tensor-product Hilbert space. With that data, the paper proposes that every (1+1)d symmetry-protected topological (SPT) phase with such a symmetry is described by a Q-system in the charge category whose image under the forgetful functor is a matrix algebra. It proves a correspondence between fiber functors and such Q-systems and gives an explicit commuting-projector lattice model realizing all three $\\mathrm{Rep}^\\dagger(D_8)$ SPT phases, one of which is a product state. The model also exhibits an $S_3$-duality that permutes the three phases, matching the monoidal automorphisms of the symmetry category. If correct, this provides a systematic lattice-level classification and construction for non-invertible SPTs, including a canonical trivial phase.","feed_headline":"Non-invertible SPT phases reduce to Q-systems in the charge category","feed_subtitle":"A lattice model realizes all three Rep†(D8) SPTs, including a trivial product state, and an S3-duality permutes them.","key_machinery":"The central object is the onsite matrix-product-operator (MPO) representation of the symmetry, whose virtual bond dimension is forced to equal the quantum dimension of the corresponding object; this is the lattice avatar of the fiber functor. On top of it, the classification rests on the Q-system, a unitary separable algebra object in the charge category, together with the condition that its forgetful image is a matrix algebra. The Q-system supplies the commuting-projector Hamiltonian and the fixed-point tensor network, while the matrix-algebra condition guarantees a unique symmetric ground state and a well-defined edge mode. The charge category itself is the representation category of the Hopf algebra obtained from $(\\mathcal{C}, f)$ by Tannaka duality, for example $\\mathrm{Hilb}_{D_8}$ for $\\mathrm{Rep}^\\dagger(D_8)$.","core_discovery":"On the paper's own terms, the discovery is that the right UV description of an anomaly-free fusion category symmetry is the pair $(\\mathcal{C}, f)$, and that the SPT phases enriched by this symmetry correspond to Morita classes of Q-systems $A$ in the charge category $\\mathcal{C}^\\vee_{\\mathrm{Hilb}_f}$ for which the forgetful image $\\mathrm{fgt}\\,A$ is a matrix algebra in $\\mathrm{Hilb}$. A Q-system is a unitary separable algebra object; when the symmetry is forgotten it must look like $\\mathrm{End}(W)$ for some Hilbert space $W$, which is exactly the condition for a unique ground state in the thermodynamic limit and a stable edge mode. The paper realizes this concretely for $\\mathrm{Rep}^\\dagger(D_8)$, where the charge category is $\\mathrm{Hilb}_{D_8}$: the three Q-systems $A_0=\\mathbb{C}e$, $A_1=\\mathbb{C}_\\omega\\langle r^2,s\\rangle$, and $A_2=\\mathbb{C}_\\omega\\langle r^2,sr\\rangle$ exhaust the Morita classes whose forgetful images are matrix algebras, giving two non-trivial SPTs and the trivial product state. It then shows that the $S_3$ monoidal automorphisms of $\\mathrm{Rep}^\\dagger(D_8)$ act on these phases by a lattice duality implemented with projective charges, and that the previously proposed cluster-state models become these Q-system models after their MPOs are reduced to the onsite form.","pith_inferences":["Beyond the paper: if the pair $(\\mathcal{C}, f)$ is accepted as the definition of symmetry, then 'anomaly-free' becomes a property of the chosen local realization, not of the abstract fusion category alone; two fiber functors can produce different charge categories for the same $\\mathcal{C}$, so the same Hamiltonian can be symmetric in one microscopic description and not in another.","Beyond the paper: the matrix-algebra criterion gives a direct numerical signature: for a Q-system model, the dimension of the center $\\mathcal{Z}(\\mathrm{fgt}\\,A)$ should match the bulk ground-state degeneracy observed in the thermodynamic limit, so tensor-network calculations on the Section II Hamiltonians can test the classification without any categorical input.","Beyond the paper: applying the same Q-system construction to fermionic fusion category symmetries, with the fiber functor landing in super-vector spaces, would predict that fermionic SPTs correspond to Q-systems that are simple in $\\mathrm{sVec}$ while preserving fermion parity; an explicit lattice realization would extend the framework to fermionic chains."],"forward_implications":["For any anomaly-free fusion category symmetry $(\\mathcal{C}, f)$, the trivial phase is an integral part of the symmetry data rather than an extra choice: the unit object of the charge category gives a product state that serves as the reference SPT phase.","The classification of (1+1)d SPTs with such symmetries reduces to Morita classes of Q-systems in the charge category whose forgetful image is a matrix algebra in $\\mathrm{Hilb}$.","The three $\\mathrm{Rep}^\\dagger(D_8)$ SPT phases are permuted by an $S_3$ duality realized on the lattice through projective charges, matching the action of the monoidal automorphisms on the three fiber functors.","The cluster-state models previously proposed for $\\mathrm{Rep}^\\dagger(D_8)$ reduce, after a symmetry-preserving local unitary, to the Q-system models of this paper; the cluster state itself becomes the trivial product state once its MPO is put in onsite form.","The same Q-system criterion identifies the SPT phases for other categories, for example $\\mathrm{Rep}^\\dagger(S_3\\times Z_3)$, which has one trivial and one non-trivial symmetric phase."],"supporting_citations":[{"why":"establishes the TQFT classification of 1d C-SPTs by fiber functors, which the paper turns into a UV principle.","marker":"[15]"},{"why":"the earlier cluster-state realization of Rep†(D8) phases that the paper revisits, reducing its MPO to onsite form to expose the trivial phase.","marker":"[35]"},{"why":"supplies the general construction of fusion-category-symmetric lattice models from charge categories.","marker":"[36]"},{"why":"introduces the completeness/onsite condition for MPO symmetries that the paper adopts as the definition of onsite action.","marker":"[37]"},{"why":"provides the Q-system fixed-point model and commuting-projector Hamiltonian underlying the SPT classification.","marker":"[45]"},{"why":"defines the Tambara-Yamagami category and shows Rep†(D8) has three fiber functors, the source of the three phases.","marker":"[51]"},{"why":"gives Tannaka duality and the classification of Q-systems in HilbG by subgroups and cocycles, used to enumerate the matrix algebras.","marker":"[59]"}],"fun_headline_variants":["Q-systems classify non-invertible SPTs in 1+1D","Onsite MPOs realize all D8 SPT phases, S3-dual","Non-invertible SPTs reduce to Q-systems in charge category","Trivial phase included: Q-system model for Rep†(D8) SPTs","Fusion category SPTs are Q-systems up to Morita"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is the definition that an anomaly-free fusion category symmetry is a pair $(\\mathcal{C}, f)$ with a chosen fiber functor, together with the requirement that 'onsite' means each symmetry MPO has virtual bond dimension equal to its quantum dimension; if one instead treats the bare fusion category as the symmetry, the canonical trivial phase and the Q-system classification do not follow.","fun_headline_variants_meta":{"raw":{"variants":["Q-systems classify non-invertible SPTs in 1+1D","Onsite MPOs realize all D8 SPT phases, S3-dual","Non-invertible SPTs reduce to Q-systems in charge category","Trivial phase included: Q-system model for Rep†(D8) SPTs","Fusion category SPTs are Q-systems up to Morita"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000709,"raw_usage":{"total_tokens":3247,"prompt_tokens":1053,"completion_tokens":2194,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":669,"completion_tokens_details":{"reasoning_tokens":2088}},"tokens_in":669,"tokens_out":2194,"duration_ms":15783,"temperature":1.0,"reasoning_tokens":2088,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T23:18:40.953943+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Numerically compute the ground-state degeneracy on a half-infinite open chain at the two non-trivial exactly solvable points $(\\lambda_0,\\lambda_1,\\lambda_2)=(0,1,0)$ and $(0,0,1)$ of the Section II model: the paper predicts a two-fold degenerate edge mode at both, and none at $(1,0,0)$. A tensor-network calculation showing any other degeneracy pattern, or showing that the $S_3$-duality maps the phases differently from the coset action on $S_3/K$, would settle the claim.","supporting_citations":[],"review_version":1}