{"id":"ac7e6837-01e4-4478-88c8-6882d225f8a0","arxiv_id":"1908.07019","paper_version":3,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"A moduli stack of two-dimensional mod p Galois representations of a p-adic field is constructed, and its irreducible components are shown to correspond exactly to the Serre weights of the representations.","lead":"This paper constructs a geometric moduli stack whose points are two-dimensional mod p representations of the Galois group of a p-adic field. It then proves that the irreducible components of this stack are naturally labelled by the Serre weights that appear in the weight part of Serre's conjecture.","discovery_kind":"unification","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The tame-descent-data dictionary in Section 2 is asserted rather than proved, and Theorem 3.9.2(3) depends on it; if the I(K'/K)-equivariance in Lemma 3.5.16 fails, the Fp-point bijection and component labelling collapse.","rationale":"The reader's weakest assumption is exactly the tame-descent-data dictionary in Section 2, and I agree that this is the most load-bearing point. The reason for moving from ACCEPT to CONDITIONAL is that the paper itself flags the absence of a treatment in the literature, and the step is indispensable: Theorem 3.9.2(3) identifies the Fp-points of Z and Z_tau, and the component labelling in Theorem 1.1.1 is built on that identification. The concern is not an internal inconsistency, and the surrounding construction appears coherent, with external inputs from [PR09], [EG19b], [Kis09], and [GK14] used in a clearly specified way. But Proposition 2.3.6 and Lemma 3.5.16 contain the two places where tame descent data must actually be matched with Galois representations, and both are asserted by analogy or by 'one sees from its construction' rather than proved. A conditional acceptance requiring the tame-descent-data equivalence to be written out, or verified by the concrete comparison test above, is proportionate. If the check passes, ACCEPT is justified.","tokens_in":66308,"tokens_out":11869,"duration_ms":138644,"concrete_test":"Independently establish Proposition 2.3.6 for a concrete tame type rather than by analogy. Take p>2, K=Q_p, a tame principal-series type tau=eta⊕eta' on K'=Q_p(pi^{1/(p-1)}), and a reducible r=chi_1⊕chi_2. Use the rank-one extension computations of Section 4 to write the quotient R^{≤1}_{r|G_K∞} explicitly as a power-series ring modulo the relations defining height-one Breuil-Kisin modules with descent data. Then compute R^{0,1}_r from the known potentially Barsotti-Tate deformation ring of [GK14], using the normalizations of Section 1.7, and check that the restriction map is an isomorphism by comparing defining equations. Also compute T(M) for one non-split extension M of type tau and verify, via [Kis09, Prop. 1.1.13], that the resulting G_K-representation has inertial type tau.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central theorem's Fp-point description rests on Theorem 3.9.2(3), which asserts that Fp-points of Z_tau are exactly the mod p representations admitting a potentially Barsotti-Tate lift of type tau. This depends on two unproved parts of the Section 2 dictionary. First, Proposition 2.3.6 is proved by saying that the tame-descent-data case goes 'in exactly the same way' as [Kim11, Cor. 2.2.1], even though the authors state that no treatment of tame descent data for Breuil-Kisin modules exists in the literature. Second, Lemma 3.5.16 reduces the compatibility of inertial type with descent data to the assertion that the isomorphism (3.5.17) is 'in fact equivariant' for the action of I(K'/K), with no calculation supplied. If the equivariance fails, a Breuil-Kisin module of type tau could correspond to a Galois representation whose inertial type is a twist of tau, or the restriction functor from G_K to G_K∞ could fail to be an isomorphism on the relevant deformation rings. In either case the points of Z_tau would not be the representations with a potentially Barsotti-Tate lift of type tau, and the labelling of irreducible components by Serre weights in Theorem 1.1.1 would not follow. This is not a disagreement with external consensus; it is a load-bearing omitted proof, explicitly acknowledged in Section 2, and it should be supplied or independently checked.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper constructs moduli stacks of two-dimensional mod p Galois representations of a p-adic local field, using Breuil–Kisin modules with tame descent data and the scheme-theoretic image construction of [EG19b]. The central result, Theorem 1.1.1, asserts that the resulting stack Z is an algebraic stack of finite type over F_p, equidimensional of dimension [K:Q_p], and that its irreducible components are labelled by Serre weights in such a way that the F_p-points of the component Z(σ) are precisely those representations r:G_K→GL_2(F_p) having σ as a Serre weight. The paper also studies the auxiliary stack C of Breuil–Kisin modules, proves equidimensionality and generic reducedness results, relates the singularities to local models of Shimura varieties, and proves a geometric Breuil–Mézard statement for tamely potentially Barsotti–Tate deformation rings.","tokens_in":66496,"tokens_out":9404,"duration_ms":102851,"significance":"If the main theorem is correct, this is a substantial advance: it gives a geometric object whose irreducible components encode the weight part of Serre's conjecture for GL(2), something the authors plausibly argue is not accessible from the explicit descriptions of the weight sets W(r). The paper contains a large amount of original technical work, including the explicit extension calculations of Section 4, the versal-ring arguments of Section 3.10, and the local-model comparisons of Section 3.7–3.8. The writing is generally careful and the authors are explicit about what is known and what is being assumed. However, the foundational dictionary in Section 2 between Breuil–Kisin modules with tame descent data and the relevant Galois representations is asserted rather than proved, and that dictionary is load-bearing for the identification of F_p-points of Z_τ and hence for the Serre-weight labelling. This is a correctness-risk concern internal to the manuscript, not a disagreement with external consensus.","major_comments":[{"comment":"Proposition 2.3.6 is the key bridge between Galois deformation rings and Breuil–Kisin modules with tame descent data, but its proof is the single sentence that it can be proved 'in exactly the same way' as [Kim11, Cor. 2.2.1], even though the authors state at the start of Section 2 that no treatment of tame descent data for Breuil–Kisin modules exists in the literature. This isomorphism Spec R^{[0,1]}_r → Spec R^{≤1}_{r|G_K∞} is used in Lemma 3.10.9 and ultimately in Theorem 3.9.2(3), so the omission is load-bearing. I request a complete proof, or at least a precise statement of the functor on deformation rings together with a verification that the descent data, the height-one condition, and the strong determinant condition are preserved by Kisin's construction.","section":"Section 2, Proposition 2.3.6"},{"comment":"The type-compatibility part of Lemma 3.5.16 depends entirely on the assertion that the isomorphism (3.5.17) is 'in fact equivariant' for the action of I(K'/K), with no calculation supplied. This step is essential because it identifies the type τ of a Breuil–Kisin module, defined through the action on M_i/uM_i, with the inertial type of the associated Galois representation, defined through D_pcris. If the equivariance holds only after a twist, or if reducing modulo u and its divided powers changes the character decomposition, then the F_p-points of Z_τ would not be the representations with a potentially Barsotti–Tate lift of type τ, and the labelling in Theorem 5.2.2 would not follow. Please supply the computation, or a reference that treats this exact comparison for tame descent data.","section":"Section 3.5, Lemma 3.5.16"},{"comment":"The proof identifies F'-points of Z_τ with Breuil–Kisin modules over F' satisfying the strong determinant condition and having type τ, and then asserts that 'by Lemma 3.5.16 and Corollary 3.8.3' these correspond to representations having a potentially Barsotti–Tate lift of type τ. Lemma 3.5.16 is a statement about Spf(O_E')-points, not about F'-points, and Corollary 3.8.3 only gives flatness and reducedness of the special fibre. An additional deformation-theoretic argument is needed to pass from a finite-field-valued Breuil–Kisin module to a characteristic-zero lift of the same inertial type. Since this bijection is exactly what makes the irreducible components of Z_τ carry the Serre-weight labels, the missing step is load-bearing and should be written out.","section":"Section 3.9, proof of Theorem 3.9.2(3)"}],"minor_comments":[{"comment":"In the first paragraph there is a duplicated phrase: 'the special fibres of these moduli spaces would be would be moduli spaces of mod p representations'; please delete the second 'would be'.","section":"Section 1.1"},{"comment":"In the displayed definition of D_η for cuspidal types, the action of the (f+1)st copy of G_m on the variable α is not specified, although the quotient is by G_m^{f+1}; please clarify the action and the resulting invariant description.","section":"Section 3.11, cuspidal case"},{"comment":"The passage comparing the strong determinant condition over K' and L' says that the latter implies the former 'up to an [l':k']th root of unity', and then that comparing the X_{0,σ_i} terms forces the root to be 1. This is quite compressed; a short explicit equation would make the argument easier to verify.","section":"Section 3.6, proof of Proposition 3.6.1(2)"},{"comment":"The notation lim←− R_{τ,a} is used both for a projective limit of rings and for a formal scheme Spf(lim←− R_{τ,a}); please add a sentence fixing the precise meaning in each occurrence.","section":"Section 3.10, Corollary 3.10.18"}],"recommendation":"major_revision","confidential_remarks":"The missing tame-descent-data dictionary is the only serious obstruction I see to the main theorem. The paper is otherwise substantial, carefully written, and plausible. If the authors can supply the missing proofs or precise references for Proposition 2.3.6 and Lemma 3.5.16, and clarify the F'-point step in Theorem 3.9.2(3), the paper should be acceptable; in its current form I cannot certify the main theorem from the written proof. The authors' own explicit admission that the tame-descent-data theory is not in the literature makes this a genuine gap rather than a mere citation issue."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"This is a real step forward. It gives the first stack-theoretic geometrisation of the weight part of Serre's conjecture for GL(2) over p-adic fields, and the main theorem delivers exactly what it promises: an algebraic stack Z, equidimensional of dimension [K:Q_p], whose irreducible components are labelled by Serre weights, with the F_p-points of the component Z(σ) being precisely the representations having σ as a Serre weight. The construction of the stack, its resolution via Breuil–Kisin modules, and the component labelling are new, and the paper does a lot of honest work: long proofs, careful dimension computations, and a clear relation to local models of Shimura varieties. The use of external results, especially [GK14] and [EG19b], is documented and the main theorem is not a restatement of known material.\n\nThe soft spot is exactly where the reader's report puts it, and I think the stress-test note is on target: the tame descent data dictionary in Section 2 is asserted rather than proved. Proposition 2.3.6 says the key isomorphism can be proved 'in exactly the same way' as [Kim11, Cor. 2.2.1], and Lemma 3.5.16 asserts that the isomorphism (3.5.17) is 'in fact equivariant' without a calculation. This matters because Theorem 3.9.2(3), the bijection between F_p-points of Z_τ and representations admitting a potentially Barsotti–Tate lift of type τ, rests on it. If the equivariance fails, the component labelling in Theorem 1.1.1 would not follow. So this is a genuine load-bearing gap, not a cosmetic one.\n\nThat said, the authors flag it openly, the arguments are believable to an expert, and the rest of the paper is carefully built. I would not call it fatal, but it deserves scrutiny. A referee should ask for a fuller proof of the Section 2 dictionary, or a reference to a companion note where it is written out. The paper also discloses the exclusions (p>2, non-très-ramifiée) honestly, and the relation to [EG19a] is clearly stated.\n\nThis is a paper for specialists in p-adic Hodge theory and modular forms. It deserves a serious referee and likely acceptance after a requested expansion of Section 2. I would send it out.","headline":"A serious, well-built paper that geometrises Serre weights via a moduli stack; the main risk is a load-bearing but unproved tame-descent-data dictionary in Section 2.","tokens_in":67137,"tokens_out":1607,"would_cite":true,"duration_ms":20772,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["11F80","14D23"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper constructs a moduli stack of two-dimensional mod p Galois representations of a p-adic local field and proves its irreducible components are naturally labelled by Serre weights.","keywords":["moduli stacks","Galois representations","Serre weights","Breuil–Kisin modules","weight part of Serre's conjecture","p-adic Hodge theory","Langlands program","formal algebraic stacks"],"falsifier":"Enumerate, for a concrete case such as $K=\\mathbb{Q}_p$ with $p=3$, the $\\mathbb{F}_p$-points of the component $\\mathcal{Z}(\\sigma)$ associated to a Steinberg Serre weight, using the explicit extension families of Section 4, and compute the Serre-weight set $W(r)$ of each resulting representation by the known explicit rules. A single point $r$ whose weight set does not contain $\\sigma$ would disprove Theorem 1.1.1; conversely, a single representation with $\\sigma\\in W(r)$ that is not an $\\mathbb{F}_p$-point of $\\mathcal{Z}(\\sigma)$ would show the component labelling is wrong.","tokens_in":65998,"feed_emoji":"🧩","tokens_out":12364,"duration_ms":113466,"temperature":0.7,"pith_summary":"This paper tries to establish a geometric home for the weight part of Serre's conjecture for $\\mathrm{GL}_2$ over a $p$-adic field. It constructs a moduli stack $\\mathcal{Z}$ whose $\\mathbb{F}_p$-points are, up to the explicitly understood très ramifiée exceptions, the two-dimensional mod $p$ representations of the absolute Galois group $G_K$. The main theorem (Theorem 1.1.1) says that $\\mathcal{Z}$ is an algebraic stack of finite type over $\\mathbb{F}_p$, equidimensional of dimension $[K:\\mathbb{Q}_p]$, and that its irreducible components are labelled by Serre weights: a representation $r$ lies on the component $\\mathcal{Z}(\\sigma)$ labelled by $\\sigma$ exactly when $\\sigma$ is one of the Serre weights of $r$. A sympathetic reader should care because this recasts a set of representation-theoretic weight sets, normally described by complicated explicit recipes, as the irreducible-component geometry of a single object.","feed_headline":"Serre weights label the components of one Galois-rep stack","feed_subtitle":"For a p-adic field K, the stack is [K:Q_p]-dimensional and its points are exactly the 2D mod p representations.","key_machinery":"The load-bearing object is the stack $\\mathcal{Z}$, obtained as the scheme-theoretic image of $\\mathcal{C}\\to\\mathcal{R}$. Here a rank-two Breuil–Kisin module with tame descent data is a projective module over $S_A=(W(k')\\otimes_{\\mathbb{Z}_p}A)[\\![u]\\!]$ equipped with a Frobenius whose cokernel is killed by $E(u)$, together with commuting descent data from a tame extension $K'/K$; its generic fibre is an étale $\\phi$-module, which by the standard correspondence corresponds to a representation of $G_{K_\\infty}$, and the height-one condition forces a canonical extension to $G_K$ that is potentially Barsotti–Tate. The morphism $\\mathcal{C}\\to\\mathcal{R}$ is proper, so the scheme-theoretic image is an algebraic stack, and the strong determinant condition (a determinant equality selecting Hodge–Tate weights $\\{0,1\\}$) identifies the Barsotti–Tate substack. The singularities of $\\mathcal{C}$ are controlled by comparison with local models at Iwahori level for the Weil restriction of $\\mathrm{GL}_2$, which yields Cohen–Macaulayness, flatness and reducedness; the components of $\\mathcal{Z}$ are then produced by closures of explicit families of extensions of characters, indexed by Serre weights through the inertial local Langlands correspondence.","core_discovery":"The central claim is that the geometry of $\\mathcal{Z}$ is a complete invariant for the weight part of Serre's conjecture in dimension two. On the paper's own terms, $\\mathcal{Z}$ is defined as the scheme-theoretic image of the proper morphism $\\mathcal{C}\\to\\mathcal{R}$, where $\\mathcal{C}$ is the moduli stack of rank-two Breuil–Kisin modules of height at most one with tame descent data and $\\mathcal{R}$ is the moduli stack of étale $\\phi$-modules; the strong determinant condition cuts out the potentially Barsotti–Tate locus. Theorem 1.1.1 then asserts that $\\mathcal{Z}$ is finite type over $\\mathbb{F}_p$, equidimensional of dimension $[K:\\mathbb{Q}_p]$, and that the map $\\sigma\\mapsto\\mathcal{Z}(\\sigma)$ is a bijection between Serre weights and irreducible components, with $r\\in\\mathcal{Z}(\\sigma)(\\mathbb{F}_p)$ if and only if $\\sigma\\in W(r)$. The authors also show the generic point of each component is an explicit extension of characters, that the auxiliary stack $\\mathcal{C}$ resolves $\\mathcal{Z}$ enough to prove generic reducedness, and that the special fibres of tamely potentially Barsotti–Tate deformation rings are generically reduced.","pith_inferences":["Inference beyond the paper: applying the same scheme-theoretic-image construction to height $h>1$ or rank $d>2$ Breuil–Kisin modules should produce stacks whose irreducible components encode generalized weight sets; the pattern established here predicts the labels will again come from inertial types and their Jordan–Hölder contents.","Inference beyond the paper: because irreducible representations are limits of reducible families, patching arguments may be recast as intersection theory on $\\mathcal{Z}$, replacing explicit Ext computations by component geometry.","Inference beyond the paper: for small $p$ and $K=\\mathbb{Q}_p$, the component labelling could be verified independently by computing the closures of the explicit extension families, offering a test of the dictionary that does not pass through deformation rings."],"forward_implications":["The weight part of Serre's conjecture for $\\mathrm{GL}_2$ is equivalent to a geometric statement about the $\\mathbb{F}_p$-points of one stack: the fibre of $\\mathcal{Z}(\\sigma)$ is exactly the set of representations having $\\sigma$ as a Serre weight.","The special fibres of tamely potentially Barsotti–Tate deformation rings are generically reduced (Proposition 5.1.1), which the authors note is hard to obtain purely from formal deformation theory and is expected to be useful for mod $p$ Hilbert modular forms.","Irreducible mod $p$ representations arise as specializations of reducible extension families inside $\\mathcal{Z}$, so the geometric picture of the moduli problem is not separated into irreducible and reducible points in the usual way.","Since $\\mathcal{Z}$ is equidimensional of dimension $[K:\\mathbb{Q}_p]$, each component $\\mathcal{Z}(\\sigma)$ has known dimension, and its generic point has an explicit description as an extension of two inertia-restricted characters."],"supporting_citations":[{"why":"Establishes algebraicity of the stack of Breuil–Kisin modules and gives the quotient presentation used to define $\\mathcal{C}$.","marker":"[PR09]"},{"why":"Supplies the theory of Breuil–Kisin modules of height at most one and their relation to finite flat group schemes and local models, used throughout.","marker":"[Kis09]"},{"why":"Provides the notion of scheme-theoretic image for proper morphisms to a stack and the criteria that make $\\mathcal{Z}$ algebraic.","marker":"[EG19b]"},{"why":"Underlies the equivalence between étale $\\phi$-modules and representations of $G_{K_\\infty}$.","marker":"[Fon90]"},{"why":"Gives the description of Serre-weight sets in terms of potentially Barsotti–Tate deformation rings that is interpreted geometrically in Section 5.","marker":"[GK14]"},{"why":"Supplies the geometric Breuil–Mézard formulation for deformation rings, converted here to formal completions of the stacks.","marker":"[EG14]"},{"why":"Provides the modern formulation of the weight part of Serre's conjecture whose weight sets the paper geometrises.","marker":"[BDJ10]"},{"why":"Connects the moduli stacks $\\mathcal{C}_\\tau$ to local models of Shimura varieties, giving control of singularities and dimensions.","marker":"[CL18]"}],"fun_headline_variants":["Serre weights label components of a 2D Galois-rep stack","2D mod p Galois reps: stack geometry encodes Serre weights","Galois-rep moduli stack: Serre weights as irreducible components","Stack of 2D Galois reps: each component is one Serre weight"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The most fragile premise is the Section 2 dictionary that Breuil–Kisin modules with tame descent data, together with the strong determinant condition selecting Hodge–Tate weights $\\{0,1\\}$, classify the appropriate potentially Barsotti–Tate Galois representations; the paper states that no treatment of tame descent data for Breuil–Kisin modules exists in the literature and that the arguments are almost identical to known cases, so they are given briefly.","fun_headline_variants_meta":{"raw":{"variants":["Serre weights label components of a 2D Galois-rep stack","2D mod p Galois reps: stack geometry encodes Serre weights","Galois-rep moduli stack: Serre weights as irreducible components","Stack of 2D Galois reps: each component is one Serre weight"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000491,"raw_usage":{"total_tokens":2357,"prompt_tokens":828,"completion_tokens":1529,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":444,"completion_tokens_details":{"reasoning_tokens":1447}},"tokens_in":444,"tokens_out":1529,"duration_ms":11159,"temperature":1.0,"reasoning_tokens":1447,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T12:29:24.305123+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Enumerate, for a concrete case such as $K=\\mathbb{Q}_p$ with $p=3$, the $\\mathbb{F}_p$-points of the component $\\mathcal{Z}(\\sigma)$ associated to a Steinberg Serre weight, using the explicit extension families of Section 4, and compute the Serre-weight set $W(r)$ of each resulting representation by the known explicit rules. A single point $r$ whose weight set does not contain $\\sigma$ would disprove Theorem 1.1.1; conversely, a single representation with $\\sigma\\in W(r)$ that is not an $\\mathbb{F}_p$-point of $\\mathcal{Z}(\\sigma)$ would show the component labelling is wrong.","supporting_citations":[],"review_version":1}