{"id":"76e9eafe-43c3-46f6-a698-d9e402b0efdd","arxiv_id":"1908.02942","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Every Albert division algebra over any field has an isotope that contains a cyclic cubic subfield, and its structure group then contains a subgroup of type ^3D4.","lead":"Mathematicians have long wondered whether every Albert division algebra, an exceptional 3 by 3 type of number system, contains a cyclic cubic number field inside it. This paper proves a weaker version: after a controlled change of multiplication called an isotope, every such algebra does contain one.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The characteristic-free normalization N_B(u)=N_K(μ)=1 is asserted without proof; if it fails, the constructed algebra A′ is inadmissible and the proof collapses.","rationale":"The reader's weakest assumption points to the normalization step, and my read agrees. I considered the other candidate weak points: the line 'σv being distinguished and 2.(c)-(iv) imply that f3(A) is hyperbolic' is a typo for f3(A′) and is easily repaired; the use of Wedderburn is sound because g3(A′)≠0 forces the underlying degree-3 algebra in the first Tits construction to be a division algebra; and the applications follow if the theorem is proved. The genuinely load-bearing step is the normalization, because A′=J(B,σ_v,1,μ) is admissible only when N_K(μ)=1, and the paper supplies no proof that every admissible pair can be normalized in arbitrary characteristic. The cited [6,(39.2)(2)] may provide such a result, and the author's parenthetical 'obviously works' may be true, but as written the argument depends on an unverified extension of a book result. The proposed check settles whether the normalization exists; until then, conditional acceptance is appropriate.","tokens_in":4639,"tokens_out":34103,"duration_ms":333992,"concrete_test":"Locate [6, Proposition (39.2)(2)] and write out the exact transformation it gives. Then apply it to an admissible pair with K a quadratic field, B a degree-3 central division algebra over K, and μ∈K^× such that N_K(μ) is not in N_K(Nrd(B^×)); determine whether a normalized presentation with N_B(u)=N_K(μ)=1 exists, possibly after passing to an isotope. If no such presentation exists, the proof's first step fails for this A and the theorem is unproved in arbitrary characteristic.","verdict_should_be":"UNCHANGED","load_bearing_attack":"In Section 3, the proof begins by invoking [6, (39.2)(2)] with the comment that the proof 'obviously works in any characteristic' to normalize a second Tits construction A=J(B,σ,u,μ) so that N_B(u)=N_K(μ)=1. This step is load-bearing: the subsequent algebra A′=J(B,σ_v,1,μ) is admissible only if N_K(μ)=1, and the whole argument depends on A′ being an Albert algebra. The cited result is not stated and the characteristic-free extension is not proved. For a degree-3 central division algebra B over a quadratic field K, the reduced norm image N_B(B^×) is generally a proper subgroup of K^×, so the transformation implicitly required—changing u and μ while preserving B and σ—is not obviously always available; it may require an isotope rather than an isomorphism, which the text does not specify. If this normalization is unavailable, A′ does not exist and the proof cannot proceed.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper addresses Albert's problem for Albert division algebras: whether every such algebra contains a cyclic cubic subfield. The main theorem states that, over a field of arbitrary characteristic, every Albert division algebra has an isotope containing a cyclic cubic extension of the base field. The proof represents A as a second Tits construction J(B,σ,u,μ), normalizes the admissible pair to NB(u)=NK(μ)=1, finds v∈B× making σv distinguished, and forms A′=J(B,σ_v,1,μ). It then uses the mod 3 invariant to show A′_L splits for a cyclic cubic L, forcing L into an isotope of A. A corollary asserts that if char(k)≠2,3, the structure group scheme of A contains a subgroup of type ^3D4.","tokens_in":4853,"tokens_out":16085,"duration_ms":175539,"significance":"If the proof is correct, the theorem is a substantial advance: it resolves the cyclicity question up to isotopy in all characteristics, and it yields a characteristic-free construction of a ^3D4 subgroup of the structure group of any Albert division algebra. The argument is concise and makes effective use of the mod 2 and mod 3 invariants, the second Tits construction, and Petersson's structure theorems. It also generalizes Petersson's earlier results and simplifies previous proofs. The paper is creditworthy for reducing a long-standing problem to a short argument based on published theorems and for clearly stating its dependence on external results. The main caveat is the unproved normalization step, which is load-bearing and prevents the proof from being fully convincing as written.","major_comments":[{"comment":"The proof begins by asserting that, by [6, (39.2)(2)], whose proof 'obviously works in any characteristic', one may assume NB(u)=NK(μ)=1. This normalization is load-bearing: A′=J(B,σ_v,1,μ) is an Albert algebra only when NK(μ)=1, and the equality g3(A′)=g3(A) requires the same μ. The paper neither states the normalization result nor proves its characteristic-free extension. The reduced norm of a degree-3 division algebra need not be surjective onto K^×, so the existence of an isomorphism-preserving transformation that makes both norms 1 while keeping μ is not evident. If the normalization is unavailable, the constructed A′ may not be admissible, and the proof collapses. Please provide the precise statement from [6] and a complete proof of the arbitrary-characteristic version, or a direct reference where this is proved.","section":"§3, Proof of the theorem"},{"comment":"The sentence 'σv being distinguished and 2.(c)-(iv) imply that f3(A) is hyperbolic' appears to contain a typo: it should read 'f3(A′) is hyperbolic'. Literally, the sentence is false for an Albert division algebra A, whose f3 invariant is non-hyperbolic, and the subsequent application of [9, 4.10] requires the hyperbolicity of f3(A′), not f3(A). Please correct this.","section":"§3, Proof of the theorem"}],"minor_comments":[{"comment":"There are several typographical errors: 'dating back t o 1965' has a spurious space, and 'the a forementioned question' should be 'the aforementioned question'.","section":"Abstract and Introduction"},{"comment":"The phrase 'whose proof obviously works in any characteristic' is too informal for a journal article; please replace it with a precise statement and argument.","section":"§3, Proof of the theorem"},{"comment":"Reference [4] lists 'Israel Journal of Mathematics TBD (2019)' without a final volume or page range; if the paper is in press, please update the reference to its final form.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"The normalization gap in §3 is the central issue. The stress-test concern is legitimate: the proof as written does not justify the characteristic-free normalization of the admissible pair. I recommend asking the author to supply a complete proof of the normalization statement, or to point to the exact statement in [6] and verify that it extends to arbitrary characteristic. If the author cannot, the theorem may still be true but would require a different argument. The typo involving f3(A) versus f3(A′) should be corrected immediately."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"This short paper proves that for any Albert division algebra A over a field of arbitrary characteristic, there is an isotope of A containing a cyclic cubic subfield. That is a genuine new result—previous work only had the cube-root-of-unity case and char 3, and the original unisotoped Albert problem remains open. The proof is a tidy combination of the second Tits construction, the mod-2 and mod-3 invariants, and a theorem of Petersson–Racine on cubic subfields. The corollary that the structure group of A contains a ^3D4 subgroup (away from char 2 and 3) is also a nice dividend.\n\nI think the core argument holds up, but there is one spot that needs scrutiny. The proof opens by using [6, (39.2)(2)] to normalize N_B(u)=N_K(μ)=1, with the comment that the proof 'obviously works in any characteristic.' That is load-bearing, since the constructed algebra A'=J(B,σ_v,1,μ) is admissible only if N_K(μ)=1. For a degree-3 division algebra B over a quadratic extension K, the reduced norm image can be a proper subgroup of K^×, so it is not obvious to me that every admissible pair is equivalent to one with norm-1 entries. This might be a standard fact in the Book of Involutions, but the author should spell it out or give a precise reference; if it fails, the proof collapses.\n\nThere is also a minor typo in the same paragraph: 'f3(A) is hyperbolic' should read 'f3(A') is hyperbolic'. Easy fix.\n\nThe paper relies heavily on external theorems, several by the author himself, but the central chain is carried by Petersson, Petersson–Racine, and Wedderburn; the self-citations are legitimate.\n\nWho's this for? Albert algebra specialists and people working on exceptional groups of type E6/E8. It would be a useful contribution to a seminar, though not a blockbuster. I'd send it to peer review: it deserves referee time, and the referee should ask for the normalized construction to be justified. If that checks out, it should be accepted.","headline":"New isotope-cyclicity theorem for Albert algebras; short proof, but the characteristic-free normalization step needs a stricter referee.","tokens_in":5341,"tokens_out":9510,"would_cite":true,"duration_ms":96843,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["17C40","20G15"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proves that every Albert division algebra over a field of arbitrary characteristic has an isotope containing a cyclic cubic extension of the base field.","keywords":["Albert algebras","Jordan algebras","cyclic cubic subfields","isotopes","Tits constructions","structure group schemes","Galois cohomology invariants","exceptional algebraic groups"],"falsifier":"Exhibit an Albert division algebra over a field that has no cyclic cubic extensions; the paper's corollary says none can exist, so any such construction refutes the theorem. Alternatively, in characteristic 2 or 3, produce a second Tits construction input whose admissible pair cannot be scaled to make both norms equal 1, which would break the proof's opening normalization.","tokens_in":4419,"feed_emoji":"🧮","tokens_out":11167,"duration_ms":102308,"temperature":0.7,"pith_summary":"The paper attacks the cyclicity problem for Albert algebras, the 27-dimensional exceptional Jordan algebras that carry a cubic norm form: does every division Albert algebra contain a cyclic cubic extension of the base field? It proves the affirmative answer up to isotopy: for every Albert division algebra $A$ over a field $k$ of arbitrary characteristic, some isotope of $A$ — the same underlying vector space with the Jordan product redefined by an invertible element — contains a cyclic cubic extension of $k$. This matters because cyclic cubic subfields are the structural handle that lets one decide whether an Albert algebra is reduced or a division algebra, and they connect the theory to algebraic groups of types $F_4$, $E_6$, and $^3D_4$.","feed_headline":"Albert division algebras get a cyclic cubic field in some isotope","feed_subtitle":"Albert's 1965 question answered up to isotopy, in every characteristic, by Tits constructions and invariants.","key_machinery":"The proof is carried by the second Tits construction $A = J(B,\\sigma,u,\\mu)$, which builds an Albert algebra from a degree-$3$ central simple algebra with unitary involution $(B,\\sigma)$ and an admissible pair $(u,\\mu)$ with $N_B(u)=N_K(\\mu)$. After normalizing so both norms equal $1$, the proof produces $v \\in B^\\times$ such that the conjugate involution $\\sigma_v$ is distinguished, and forms $A' = J(B,\\sigma_v,1,\\mu)$. The mod-$3$ invariant $g_3$ is unchanged by this passage, the mod-$2$ invariant $f_3$ becomes hyperbolic, and a theorem on Jordan algebras of degree three says $A'$ is a first Tits construction, the split form that visibly contains a cyclic cubic subfield. The invariant $g_3$ then transfers that conclusion back to an isotope of the original $A$.","core_discovery":"The central claim is the theorem: to every Albert division algebra $A$ over a field $k$ of arbitrary characteristic there exists an isotope of $A$ that contains a cyclic cubic extension $L/k$; equivalently, for some cyclic cubic extension $L$, the base change $A_L$ is reduced. A corollary of the proof is that if $k$ has no cyclic cubic extensions, then every Albert algebra over $k$ is reduced, which generalizes an earlier cyclicity result for local fields. When $\\operatorname{char}(k) \\neq 2,3$, the same argument shows that the structure group scheme of $A$ contains a subgroup of type $^3D_4$ defined over $k$.","pith_inferences":["A concrete next question suggested by the proof: can the isotope in the theorem be replaced by an isomorphic copy whenever the base field already contains a cubic extension, effectively measuring the isotopy obstruction to Albert's original question?","The unproved characteristic-free normalization could be checked directly in characteristic 2 or 3 on explicit Tits construction inputs; if it holds, the proof becomes self-contained, and if it fails, the theorem still might be true but needs a new route.","Because isotopes leave the structure group scheme unchanged, the $^3D_4$ subgroup guaranteed by the corollary may be reachable in constructions of exceptional groups of type $E_8$, where cyclic cubic subfields have been used in the Tits-Weiss conjecture arguments."],"forward_implications":["Over any field with no cyclic cubic extensions, every Albert algebra is reduced: Albert division algebras cannot exist there.","Every Albert division algebra has an isotope that contains a cyclic cubic subfield and whose mod-2 5-invariant $f_5$ is hyperbolic.","When the characteristic is not 2 or 3, the structure group scheme of every Albert division algebra contains a subgroup of type $^3D_4$ defined over $k$.","The earlier cyclicity result for Albert division algebras over local fields, and its dependence on the classification of such algebras, is subsumed by a uniform proof in all characteristics."],"supporting_citations":[{"why":"Supplies the norm-one normalization of the admissible pair $(u,\\mu)$ in the second Tits construction.","marker":"[6, (39.2)(2)]"},{"why":"Exhibits an element $v$ making the conjugate involution $\\sigma_v$ distinguished.","marker":"[9, 2.10]"},{"why":"Provides the passage to the new Albert algebra $A' = J(B,\\sigma_v,1,\\mu)$.","marker":"[20, 3.4]"},{"why":"Shows the mod-3 invariant is independent of $\\sigma$, a step used to identify $g_3(A')$ with $g_3(A)$.","marker":"[16, 3.5]"},{"why":"Covers the characteristic-3 analogue of independence of the mod-3 invariant.","marker":"[15, 8.]"},{"why":"Concludes that a hyperbolic $f_3$ forces $A'$ to be a first Tits construction.","marker":"[9, 4.10]"},{"why":"Forces a cyclic cubic subfield into an isotope of $A$ once the base change $A_L$ is reduced.","marker":"[12, Thm. 2]"},{"why":"Gives the subgroup of type $^3D_4$ in the automorphism group scheme when a cyclic cubic subfield is present.","marker":"[4, Cor. 3.2]"}],"fun_headline_variants":["Albert algebras: cyclic cubic subfield via isotope","Cyclic cubic subfield after isotopy for Albert algebras","Albert's cyclicity problem: solved up to isotopy","Cyclicity in Albert algebras: isotope suffices","Isotope yields cyclic cubic subfield for Albert algebras"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is the unproved characteristic-free normalization, taken from [6, (39.2)(2)], that every second Tits construction input can be scaled so both norms equal 1; if that fails in characteristic 2 or 3, the isotope built in the proof need not be admissible.","fun_headline_variants_meta":{"raw":{"variants":["Albert algebras: cyclic cubic subfield via isotope","Cyclic cubic subfield after isotopy for Albert algebras","Albert's cyclicity problem: solved up to isotopy","Cyclicity in Albert algebras: isotope suffices","Isotope yields cyclic cubic subfield for Albert algebras"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001718,"raw_usage":{"total_tokens":6708,"prompt_tokens":770,"completion_tokens":5938,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":386,"completion_tokens_details":{"reasoning_tokens":5860}},"tokens_in":386,"tokens_out":5938,"duration_ms":40685,"temperature":1.0,"reasoning_tokens":5860,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T14:29:19.274193+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Exhibit an Albert division algebra over a field that has no cyclic cubic extensions; the paper's corollary says none can exist, so any such construction refutes the theorem. Alternatively, in characteristic 2 or 3, produce a second Tits construction input whose admissible pair cannot be scaled to make both norms equal 1, which would break the proof's opening normalization.","supporting_citations":[],"review_version":1}