By OpenAISourceGitHub

Manuscripts from openai/math

Fork
Clone
git clone https://openversions.com/git/index/openai-math.git
Cite
Version link
https://openversions.com/index/openai-math/commit/dfd8c46870b48511e697cbb43a3c8c91d375fa2f
Version identifier (commit hash)
dfd8c46870b48511e697cbb43a3c8c91d375fa2f

Fork to create your own copy with credit to the source. Clone to get a Git copy on your computer.

All papers

Claims

Findings

The Margulis–Platonov conjecture over number fields

Let kk be a number field and GG an absolutely almost simple simply connected algebraic kk-group. If N◃G(k)N \triangleleft G(k) is not contained in Z(G(k))Z(G(k)), then there is an open normal subgroup W◃HAW \triangleleft H_A such that N=δA−1(W)N = \delta_A^{-1}(W).

Under the hypotheses of Theorem 1.1, if M/kM/k is the minimal Galois extension over which GG becomes an inner form of the split group, and SS is a set of places of kk containing all archimedean places, no nonarchimedean place vv with rankkvG=0\mathrm{rank}_{k_v} G = 0, and satisfying dk(S∩Spl(M/k))>0d_k(S \cap \mathrm{Spl}(M/k)) > 0, then the SS-congruence kernel CS(G)C^S(G) is trivial.

Every abstract noncentral normal subgroup of G(k)G(k) has finite index, and to prove (MP) for GG, it suffices to show V=V0/R=1V = V_0/R = 1 for every e≥1e \ge 1, where R=G(k)eR = G(k)^e, P0=δ(R)‾⊂GAP_0 = \overline{\delta(R)} \subset G_A, and V0=δ−1(P0)V_0 = \delta^{-1}(P_0).

Under the hypotheses of Proposition 2.6, for a kk-anisotropic group GG and rational coset C=x∗γR⊂X(k)\mathcal{C} = x_* \gamma R \subset X(k), there exists x′∈Cx' \in \mathcal{C} satisfying prescribed local open conditions at BB and belonging to Uv\mathcal{U}_v for every finite v∉Bv \notin B.

Let L/kL/k be quadratic, F/kF/k be separable of degree m≥3m \ge 3, and suppose the normal closure of FF jointly with LL has Galois group Sm×C2S_m \times C_2. Then T=ker⁡(ResF/k(ResLF/F1Gm)→ResL/k1Gm)T = \ker(\mathrm{Res}_{F/k}(\mathrm{Res}^1_{LF/F}\mathbb{G}_m) \to \mathrm{Res}^1_{L/k}\mathbb{G}_m) is a torus, every everywhere locally soluble torsor under TT has a rational point, and its rational points are dense at any prescribed finite set of completions.

For an odd-degree (n≥3n \ge 3) central division algebra DD over L/kL/k with unitary involution ∗*, if the finite defect V=V0/RV = V_0/R of S=SU(D,∗)S = \mathrm{SU}(D, *) is nontrivial, then U(k)U(k) admits either a homomorphism χ:U(k)→C\chi: U(k) \to C to a finite abelian group with χ(S(k))≠1\chi(S(k)) \neq 1, or a homomorphism ϕ:U(k)→F\phi: U(k) \to \mathcal{F} to a finite nonabelian simple group with ϕ(V0)=F\phi(V_0) = \mathcal{F} and R⊂ker⁡ϕR \subset \ker \phi.

For an odd-degree (n≥3n \ge 3) central division algebra DD over L/kL/k with unitary involution ∗*, every homomorphism from U(k)U(k) to a finite abelian group is trivial on S(k)S(k).

For an odd-degree (n≥3n \ge 3) central division algebra DD over L/kL/k with unitary involution ∗*, there is no homomorphism ϕ:U(k)→F\phi: U(k) \to \mathcal{F} to a finite nonabelian simple group with R⊂ker⁡ϕR \subset \ker \phi and ϕ(V0)=F\phi(V_0) = \mathcal{F}.

For every finite nonabelian simple group F\mathcal{F}, there is a nonempty proper conjugacy-invariant subset S⊂F\mathcal{S} \subset \mathcal{F} such that if W∈SW \in \mathcal{S}, Z∉SZ \notin \mathcal{S}, and WZ=ZWWZ = ZW, then ZW∈C(B)(ZW−1)C(BZ2)ZW \in C(B)(ZW^{-1})C(BZ^2) cannot hold for every B∈FB \in \mathcal{F}.

If D=M3(L)D = M_3(L), then V0/R=1V_0/R = 1 for S=SU(D,∗)S = \mathrm{SU}(D, *).

The Margulis–Platonov assertion holds for every simply connected absolutely almost simple group of outer type AA over a number field.

Let kk be a number field and let GG be an absolutely almost simple simply connected kk-group of type D4D_4, including the trialitarian forms. Every noncentral abstract normal subgroup of G(k)G(k) is G(k)G(k).

Let kk be a number field and let GG be an absolutely almost simple simply connected kk-group of type E6E_6. Every noncentral abstract normal subgroup of G(k)G(k) is G(k)G(k).

Methods

Finite-defect reduction: isolating the discrepancy between an abstract power subgroup R=G(k)eR = G(k)^e and its closure V0=δ−1(δ(R)‾)V_0 = \delta^{-1}(\overline{\delta(R)}) as a finite quotient V=V0/RV = V_0/R using the SS-arithmetic normal subgroup theorem and strong approximation.

Source excerpts

“Fix e ≥ 1, and put R = G(k)e, P0 = δ(R) ⊂ GA, V0 = δ−1(P0), V = V0/R. Then R has finite index, P0 is open and normal in GA, and V is finite.” · Show in paper

Power-factor approximation with affine codimension-two avoidance: constructing representatives in a fixed rational coset C=x∗γR\mathcal{C} = x_* \gamma R satisfying prescribed local open conditions and avoiding a codimension-two locus at almost all finite places by combining Lemma 2.5 with anisotropy of GG.

Source excerpts

“We now use anisotropy to make the unrestricted direction in Lemma 2.5 harmless at v0. Assume G is k-anisotropic. Fix a maximal k-torus T0 ⊂ G, a finite Galois extension P/k splitting T0 and any fixed auxiliary data, and a regular cocharacter λ : Gm,P → T0,P .” · Show in paper

Cyclic detection for norm-torus cohomology: reducing the Hasse principle and weak approximation for a torus TT to the splitting of lattice extensions over cyclic subgroups via Shaω2(k,M)\mathrm{Sha}^2_\omega(k, M) and central sign elements.

Source excerpts

“Lemma 2.7 (Cyclic detection). Let M be a k-lattice split by a finite Galois extension P/k, and put Γ = Gal(P/k). Inflation identifies Sha2ω(k, M) with ker(H2(Γ, M) → ∏C⊂Γ cyclic H2(C, M)).” · Show in paper

Noncommutative difference identity and real-signature interpolation: converting an abstract finite abelian character on U(k)U(k) into a difference function Fd(a,c)F_d(a,c) on DD, proving invariance under elementary transformations, and interpolating continuity to A∪ΩA \cup \Omega via the inner-type-AA theorem over LL.

Source excerpts

“The proof first converts χ into a difference function on D. Exact elementary transformations make this difference independent of its arguments. A real-signature argument then converts that algebraic identity into continuity on S(k), after which local commutators finish the proof.” · Show in paper

Conformal uniform-marginal law and multidimensional recurrence on palettes: constructing a translation-invariant probability measure with uniform color marginals using Eisenstein-series weights, Howe–Moore matrix-coefficient decay, and additive averaging, then applying Furstenberg–Katznelson recurrence.

Source excerpts

“We construct a conformal joint law, identify its local projection, prove that its conditional color marginals are uniform, and only then average additively.” · Show in paper

Châtelet-surface and quaternion-factorization reductions: resolving ternary split-algebra unitary groups via Brauer–Manin on Châtelet surfaces, and even-degree unitary groups via a quadratic Hermitian subfield S0S_0 and quaternion factorization.

Source excerpts

“A smooth projective model of NL/k(ξ) = f1(t)f2(t)f3(t) (7.2) is a Châtelet surface. The theorem of Colliot-Thélène–Sansuc–Swinnerton-Dyer identifies the closure of its rational points with its Brauer–Manin set [8]” · Show in paper

“We shall factor a representative u into a quaternion norm-one element and an element of U(Cs).” · Show in paper