{"id":"877860b6-4b26-4283-8a69-714828d3ea1f","arxiv_id":"2411.11800","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The Artin shape is introduced and used to compute complete motivic decompositions of unitary involution varieties and Weil transfers of Severi-Brauer varieties.","lead":"The authors define the Artin shape of an algebraic variety's motive, a tool for recording how the motive splits into simple Artin pieces. They use it to compute complete motivic decompositions for unitary involution varieties and for Weil transfers of Severi-Brauer varieties, finding patterns that overturn some earlier guesses.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Unproved projective-bundle structure and base-change descent in Theorem 3.3 are load-bearing; without them the new decompositions (3.4) and (2.2) do not follow.","rationale":"The reader flagged dependence on [5] and the descent step. I agree on the descent step but locate the prior unverified input: the asserted projective bundle structure of Y over L, which is used to get the L-decomposition. Even if [5] is correct, Theorem 3.3's final step needs a Galois-invariance check; base change of Chow motives is not conservative. This does not overturn the conditional verdict, since the results may be true and the gap may be fixable, but it sharpens the condition the authors must meet. The small-case checks mentioned by the reader are consistent but do not exercise the descent at the level of idempotents.","tokens_in":5727,"tokens_out":21515,"duration_ms":221502,"concrete_test":"For p=3,n=1, let D be a cubic division L-algebra with a unitary F-involution and let Y be its rank-one involution variety. Compute the L-form Y_L from the definition of right isotropic ideals and determine whether it is a P^1-bundle over the Severi-Brauer variety X; then verify that the idempotents in the resulting decomposition M(Y)_L ≅ M(X){0}⊕M(X){1} are Gal(L/F)-invariant. If the bundle structure fails, or if the idempotents cannot be chosen Galois-stable, equation (3.4) over F and hence Theorem 3.5 are unsupported.","verdict_should_be":"UNCHANGED","load_bearing_attack":"In the proof of Theorem 3.3, the sentence \"Over L, the variety Y becomes a rank p^n−2 projective bundle over the Severi-Brauer variety X and so the motive of Y decomposes over L as in (3.4). It follows that the same decomposition holds already over F\" carries the entire descent to the main result. The geometric assertion is not derived from the definitions and is not covered by the references cited in Proposition 3.1, which only compute shape counts. Moreover, the final descent is not automatic: the base-change functor from Chow motives over F to Chow motives over L is not conservative (the Artin motive A and the Tate motive F become isomorphic over L), so an isomorphism over L of M(Y) with ⊕ U(Y){i} does not by itself yield an isomorphism over F. A Galois semilinear descent argument or an explicit Galois-stable idempotent decomposition is required. Since Theorems 3.3 and 3.5 are the paper's principal new outputs, this unsubstantiated geometric-descent step is the most load-bearing point.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper introduces a notion of 'Artin shape' for Chow motives with mod p coefficients, describing the direct-sum decomposition of a motive over the function field of the variety of Borel subgroups. It applies this notion, together with the classification theorem [5, Theorem 0.1], to obtain complete motivic decompositions for two families of projective homogeneous varieties. Theorem 1.3 gives the complete decomposition of the Weil transfer R(X) of a Severi-Brauer variety over a quadratic extension in the 'balanced' case. Lemma 2.1 and Theorem 3.5 treat the 'unitary involution' case, and Theorem 3.3 gives a complete decomposition of the unitary involution variety Y as a sum of shifts of its upper motive. The paper is concise and relies on the unpublished classification theorem from [5].","tokens_in":5903,"tokens_out":15753,"duration_ms":142361,"significance":"If the main results are correct, they provide new and somewhat surprising complete motivic decompositions, explicitly contradicting prior expectations for projective homogeneous varieties. The paper's combinatorial 'Artin shape' formalism is a convenient organizing tool, and the ratio argument in §2 is elegant. The authors are careful to state that their main structural input, Theorem 0.1, comes from an unpublished preprint, and they do not oversell the novelty. The small cases and the counting in Proposition 3.1 appear consistent, and the dependence on [5] is transparent rather than circular. However, the proof of the central Theorem 3.3 contains a load-bearing descent step that is not justified, so the main decomposition theorem is not yet established.","major_comments":[{"comment":"The assertion that the decomposition over L of M(Y) as in (3.4) implies the same decomposition over F is not justified. The base-change functor from Chow motives over F to Chow motives over L is not conservative: the Artin motive A and the Tate motive F become isomorphic over L. An isomorphism over L between M(Y)_L and the direct sum of shifts of U(Y)_L does not by itself yield an isomorphism over F. A Galois semilinear descent argument, or an explicit proof that the relevant idempotents are fixed by Gal(L/F), is required. This step is load-bearing for (3.4) and for Theorem 3.5.","section":"Theorem 3.3, proof (last two sentences)"},{"comment":"The geometric statement that over L the variety Y becomes a rank p^n−2 projective bundle over the Severi-Brauer variety X is used to obtain the decomposition (3.4) over L, but it is not proved or referenced. The statement is plausible from the flag description in Proposition 3.1, where a flag V_1 ⊂ V_{p^n−1} maps to V_{p^n−1}; however, the proof should identify the relevant vector bundle and the base explicitly, since the decomposition over L depends on this structure.","section":"Theorem 3.3, proof"},{"comment":"All main results (Theorems 1.3, 3.3, and 3.5) are deduced from Theorem 0.1 of the unpublished preprint [5]. The authors state Theorem 0.1 but do not prove it and give no stable public reference. For the paper to be verifiable, either a proof of Theorem 0.1 should be included or a published/archived reference should replace the unpublished preprint, and the dependence should be explicitly flagged in the introduction.","section":"Sections 0–3, Theorem 0.1 and all main theorems"}],"minor_comments":[{"comment":"The sentence 'the shape of M(R(X)) is still given by (1.2)' appears to refer to (1.1), since (1.2) is the complete decomposition formula, not the shape.","section":"Section 2, first paragraph"},{"comment":"The counts in Corollary 3.2 are inconsistent with Proposition 3.1 by a factor of 2. For i=1, Proposition 3.1 gives (b_1+a_1)/2 = (p^n−1)(p^n+1)/2 Tate summands and (b_1−a_1)/2 = (p^n−1)^2/2 Artin summands, whereas Corollary 3.2 displays twice these numbers. The ratio is unchanged, and the divisibility argument in Theorem 3.3 still works, but the displayed counts should be corrected.","section":"Corollary 3.2"},{"comment":"The induction in Lemma 2.1 is described in words only. It would help to state the induction invariant explicitly and to justify why the position of the first A in the shape of U forces the corresponding coefficient pattern in the complete decomposition of M(R(X)).","section":"Lemma 2.1, proof"},{"comment":"The phrase 'rank p^n−2 projective bundle' is ambiguous. Please write 'projective bundle with fiber P^{p^n−2}' or 'projective bundle associated with a vector bundle of rank p^n−1' for clarity.","section":"Theorem 3.3, proof"}],"recommendation":"major_revision","confidential_remarks":"The paper's main structural input is an unpublished theorem from the authors' own preprint [5]. The editor may wish to verify that [5] is in a form that can be relied upon by the published literature, or require that the proof be included. The descent issue in Theorem 3.3 is the principal technical gap; it is local but essential, and a revision should address it in detail."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this paper is worth reading but not ready as written. The 'Artin shape' is a genuinely useful way to package the Galois action on the motive over the Borel function field, and the paper gives new complete motivic decompositions for unitary involution varieties and Weil transfers. But Theorem 3.3 contains a load-bearing proof gap, and the paper leans on an unpublished classification by De Clercq, Karpenko, and Quéguiner-Mathieu.\n\nWhat is good: Theorem 1.3 and the alternating shape in Theorem 3.5 are new, and the ratio argument in §2 is elegant. The authors are careful about the distinction between the two cases (D ≃ D'_L versus D admitting a unitary involution), and the small examples check out. The Artin shape concept should be reusable beyond these examples.\n\nWhere I worry: in the proof of Theorem 3.3, the sentence 'Over L, the variety Y becomes a rank p^n−2 projective bundle over the Severi-Brauer variety X' is asserted without proof or reference. That is not a formal consequence of the definitions as written, and the usual dimension picture for unitary involution varieties does not obviously support it. This claim is the pivot on which (3.4) rests. The next sentence, 'It follows that the same decomposition holds already over F,' is also too quick. Base change is not conservative (A becomes F over L), so an isomorphism over L does not automatically descend. I think this particular descent can be repaired by combining the earlier conclusion that M(Y) is a sum of shifts of U(Y) with the uniqueness of the complete decomposition over L, but the paper needs to say that explicitly.\n\nThe dependence on [5] is heavy. The main input is an unpublished preprint involving one of the current authors; that is not circular, but it makes independent verification hard until [5] is available. One minor issue: Cor 3.2 gives numbers exactly twice those from Prop 3.1 for i=1. The ratio is what is used, so it does not break the argument, but it should be cleaned up.\n\nBottom line: for a specialist, the results are probably true and the Artin shape concept is worth taking seriously. But as it stands, Theorem 3.3 is not fully proved. I would send it to a good referee and ask for a real proof of the projective bundle claim and a spelled-out descent. If those come back, this is a solid paper.","headline":"Useful bookkeeping and two plausible new decompositions, but Theorem 3.3 has an unproved projective-bundle claim and a descent step that is too quick as written.","tokens_in":6447,"tokens_out":28228,"would_cite":true,"duration_ms":278945,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["20G15","14C25"],"pacs":[],"model":"deepseek-v4-flash","headline":"Artin shapes determine complete motivic decompositions for unitary involution varieties and their Weil transfers.","keywords":["Artin shape","motivic decomposition","Chow motives","projective homogeneous varieties","upper motive","unitary involution variety","Weil transfer","p'-inner groups"],"falsifier":"One concrete test is to compute the Chow group of $Y$ with $\\mathbb{Z}/p\\mathbb{Z}$ coefficients over the function field of the variety of Borel subgroups and count the Tate shifts $F$: Theorem 3.3 requires the total count to be $(p^n+1)(p^n-1)$ and the shape of $U(Y)$ to alternate $F A F A \\cdots F$, so a direct computation giving a different count or a non-alternating shape would disprove the decomposition.","tokens_in":5502,"feed_emoji":"🧩","tokens_out":9239,"duration_ms":77206,"temperature":0.7,"pith_summary":"This paper introduces the Artin shape of a motive—the list of shifts of indecomposable Artin motives obtained after scalar extension to the function field of the variety of Borel subgroups—and uses it to find complete motivic decompositions of projective homogeneous varieties. For a unitary involution variety $Y$ attached to a degree $p^n$ central division algebra over a quadratic extension, the paper proves $M(Y) \\simeq \\bigoplus_{i=0}^{p^n-2} U(Y)\\{i\\}$, while for the Weil transfer $R(X)$ of the corresponding Severi-Brauer variety it proves $M(R(X)) \\simeq \\bigoplus_{i=0}^{p^n-1} U(R(X))\\{i\\}$. These decompositions are obtained by first determining the shape of the upper motive $U(Y)$ as an alternating pattern $F A F A \\cdots F$, then using the classification theorem of [5] to rule out any summand of the form $U(Y)\\otimes A$. The results show that complete motivic decompositions can be read off from a small shape datum and that, contrary to earlier expectations, some decompositions contain only shifts of the upper motive.","feed_headline":"Artin shapes settle decompositions for unitary varieties","feed_subtitle":"Counting Tate and Artin summands in the upper motive yields full decompositions that earlier guesses missed.","key_machinery":"The Artin shape of a motive $M$ is the list of shifts $F\\{i\\}$ and $A\\{i\\}$ that appear when $M$ is pulled back to the function field of the variety of Borel subgroups; it is unique by the Krull-Schmidt-type result quoted from [3] and [11]. The upper motive $U(X)$ is the unique indecomposable summand whose shape contains the unshifted Tate motive $F$. The machinery consists of three moves: using the classification theorem [5] to know that every summand is a shift of $U(Y)$ or $U(Y)\\otimes A$; determining the shape of $U(Y)$ by counting Tate versus Artin summands through the ratio $(p^n+1)/(p^n-1)$; and descending the resulting decomposition from the quadratic extension $L$ back to the base field $F$.","core_discovery":"The central claim is that the Artin shape—together with the classification of motivic summands for $p'$-inner groups stated as Theorem 0.1 of [5]—determines the complete motivic decomposition in the examples studied. Theorem 3.3 states that for the unitary involution variety $Y$, the upper motive has shape $F A F A \\cdots F$ and the complete decomposition is $M(Y) \\simeq \\bigoplus_{i=0}^{p^n-2} U(Y)\\{i\\}$. Theorem 3.5 states that the Weil transfer $R(X)$ of the Severi-Brauer variety of $D$ has upper motive with the same alternating shape and complete decomposition $M(R(X)) \\simeq \\bigoplus_{i=0}^{p^n-1} U(R(X))\\{i\\}$. Theorem 1.3 covers the other case, where $D$ descends to the base field, and gives a decomposition with alternating shifts of $U(R(X))$ and $U(R(X))\\otimes A$. Together these theorems exhibit motivic decompositions that differ from prior expectations, in particular by omitting tensor products with the nontrivial Artin motive $A$.","pith_inferences":["One can test whether the same ratio-counting argument applies to higher-rank unitary involution varieties $Y_i$: Proposition 3.1 already supplies the counts $(b_i+a_i)/2$ and $(b_i-a_i)/2$, so a similar alternating decomposition, if it exists, would be a direct extension of Theorem 3.3.","The shape method suggests that for any $p'$-inner group whose upper motive has a known alternating shape, the complete decomposition is determined by divisibility of the number of Tate summands by $p^n$; this turns a classification theorem into a purely combinatorial counting problem.","Because the descent step from $L$ to $F$ is verified only in the specific examples, checking whether the same descent works for non-quadratic Galois extensions or for non-balanced algebras could extend the method beyond the cases treated here."],"forward_implications":["If Theorem 3.3 is correct, the complete motivic decomposition of every unitary involution variety $Y$ in this setting is a string of consecutive shifts $U(Y)\\{0\\}, U(Y)\\{1\\}, \\ldots, U(Y)\\{p^n-2\\}$, with no tensor product by $A$ appearing.","If Theorem 3.5 is correct, the Weil transfer $R(X)$ has similar consecutive shifts up to $p^n-1$, so $M(R(X)) \\simeq M(Y) \\oplus U(Y)\\{p^n-1\\}$.","In the first case of Section 1, where $D$ descends to $F$, the decomposition is instead alternating between $U(R(X))$ and $U(R(X))\\otimes A$, showing that the presence of $A$-twisted summands depends on whether $D$ admits a unitary involution.","The ratio argument shows that if the number of Tate summands in the upper motive is not divisible by $p^n$, the $A$-twisted summands cannot occur at all, giving a conceptual reason for the two different decompositions."],"supporting_citations":[{"why":"Supplies the theorem that over the Borel-variety function field the motive splits into shifts of indecomposable Artin motives, defining the Artin shape.","marker":"[2, Theorem 7.5]"},{"why":"Classification theorem: every summand of a complete motivic decomposition for a p'-inner group is a shift of U(Y)⊗A; this is the engine of all decompositions here.","marker":"[5, Theorem 0.1]"},{"why":"Gives the isomorphism criterion U(X) ≃ U(X') via prime-to-p points, used to identify the upper motive of the Weil transfer.","marker":"[11, Corollary 2.15]"},{"why":"Shows Severi-Brauer motives are indecomposable, so the upper motive equals the whole motive in the first case.","marker":"[11, Corollary 2.22]"},{"why":"Defines balanced algebras, the criterion for the Weil-transfer group to be p'-inner.","marker":"[12, §4]"},{"why":"Relates a unitary involution on D to Brauer-triviality of the norm algebra, separating the two cases.","marker":"[13, Theorem 3.1(2)]"},{"why":"Provides the count a_i of Tate shifts in unitary Grassmannian shapes used in Proposition 3.1.","marker":"[10, Lemma 7.1]"},{"why":"Gives the analogous orthogonal Grassmannian count needed for Proposition 3.1.","marker":"[9, Formula 2.6]"}],"fun_headline_variants":["Artin shapes settle complete motivic decompositions","Artin shapes crack unitary variety decompositions","Artin shapes overturn prior motivic expectations"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is the classification theorem quoted as Theorem 0.1 of [5], which asserts that every summand in the complete motivic decomposition of a projective homogeneous variety under a $p'$-inner group is a shift of a tensor product $U(Y)\\otimes A$; if that unpublished theorem is false or fails to apply, and if the descent step from the quadratic extension $L$ to $F$ in the proof of Theorem 3.3 is not valid, the decompositions do not follow.","fun_headline_variants_meta":{"raw":{"variants":["Artin shapes settle complete motivic decompositions","Artin shapes crack unitary variety decompositions","Artin shapes overturn prior motivic expectations"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000524,"raw_usage":{"total_tokens":2478,"prompt_tokens":834,"completion_tokens":1644,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":450,"completion_tokens_details":{"reasoning_tokens":1598}},"tokens_in":450,"tokens_out":1644,"duration_ms":12577,"temperature":1.0,"reasoning_tokens":1598,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T18:08:24.787979+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"One concrete test is to compute the Chow group of $Y$ with $\\mathbb{Z}/p\\mathbb{Z}$ coefficients over the function field of the variety of Borel subgroups and count the Tate shifts $F$: Theorem 3.3 requires the total count to be $(p^n+1)(p^n-1)$ and the shape of $U(Y)$ to alternate $F A F A \\cdots F$, so a direct computation giving a different count or a non-alternating shape would disprove the decomposition.","supporting_citations":[],"review_version":1}