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 Bogomolov-Pop reconstruction theorem

Let kk and ll be arbitrary algebraically closed fields of characteristics different from ℓ\ell, and let K/kK/k and L/lL/l be function fields with trdeg⁡(K/k),trdeg⁡(L/l)≥2\operatorname{trdeg}(K/k), \operatorname{trdeg}(L/l) \ge 2. Then ΦK,L:Isom⁡Fi(K,L)⟶Isom⁡c(ΠLa,ΠKa)/Zℓ×\Phi_{K,L} : \operatorname{Isom}^i_F(K, L) \longrightarrow \operatorname{Isom}^c(\Pi^a_L, \Pi^a_K)/\mathbb{Z}_\ell^\times, [α]⟼[α∗][\alpha] \longmapsto [\alpha^*], is a bijection.

For every t∈K∖kt \in K \setminus k, there is a finite constant BtB_t such that ∑D∈Supp⁡X(Θ[t−a])deg⁡D≤Bt\sum_{D \in \operatorname{Supp}_X(\Theta[t-a])} \deg D \le B_t for every a∈ka \in k.

The map Θ\Theta gives a bijection between the completed curve subfields of K/kK/k and those of L/lL/l. More precisely, for every curve subfield E⊂KE \subset K there is a unique curve subfield P⊂LP \subset L such that Θ(E^)=P^\Theta(\widehat{E}) = \widehat{P}.

Every bracket-compatible isomorphism in Theorem 1.1 is the abelianization of a continuous isomorphism ΠLc→ΠKc\Pi^c_L \to \Pi^c_K. This holds also for ℓ=2\ell = 2.

Let K/kK/k and L/lL/l be function fields over algebraically closed fields of characteristic different from ℓ\ell, each of transcendence degree at least two. Let φ:WL→WK\varphi : W_L \to W_K be a continuous bracket-compatible Zℓ\mathbb{Z}_\ell-module isomorphism. Then their transcendence degrees agree, say they are dd. For every 1≤r<d1 \le r < d, the map φ\varphi carries exactly the pairs Iw⊂DwI_w \subset D_w of quasi-prime rr-divisors of L/lL/l onto the corresponding pairs of K/kK/k. For matched pairs it induces a continuous isomorphism Dw/Iw≅WLw→Dv/Iv≅WKvD_w/I_w \cong W_{Lw} \to D_v/I_v \cong W_{Kv} that preserves alternating pairs.

There is an integer GX≥0G_X \ge 0, depending only on the embedded variety XX, such that given g1,…,gr∈L×g_1, \dots, g_r \in L^\times and q=ℓeq = \ell^e, e≥1e \ge 1, there is a quasi-prime (d−1)(d-1)-divisor ww of L/lL/l satisfying: (1) lwlw is an algebraic closure of a finite field of characteristic different from ℓ\ell and Lw=lw(C)Lw = lw(C) for a smooth projective integral curve C/lwC/lw with g(C)≤GXg(C) \le G_X; (2) w(gj)=0w(g_j) = 0 for every jj and ∏jgjnj∈L×q  ⟺  ∏jgˉjnj∈(Lw)×q\prod_j g_j^{n_j} \in L^{\times q} \iff \prod_j \bar{g}_j^{n_j} \in (Lw)^{\times q}; (3) for m=ℓm = \ell and m=qm = q, #{P∈C(lw):ord⁡P(gˉj)≢0(modm)}=sm(gj)\#\{P \in C(lw) : \operatorname{ord}_P(\bar{g}_j) \not\equiv 0 \pmod m\} = s_m(g_j).

Let ll be algebraically closed, X/lX/l a normal integral projective variety of dimension d≥2d \ge 2, and S\mathcal{S} an infinite set of prime divisors of bounded degree in a fixed projective embedding. Let A\mathcal{A} be any collection of pairs (g,q)(g, q) with g∈l(X)×g \in l(X)^\times and q≥1q \ge 1 prime to char⁡l\operatorname{char} l such that for each pair, all but finitely many D∈SD \in \mathcal{S} satisfy ord⁡D(g)=0\operatorname{ord}_D(g) = 0 and g∣D∈l(D)×qg|_D \in l(D)^{\times q}. Then there exist a finite extension M/l(X)M/l(X) and a curve field P0/lP_0/l contained in MM such that M/P0M/P_0 is regular and g∈(MP0‾)×qg \in (M\overline{P_0})^{\times q} for every (g,q)∈A(g, q) \in \mathcal{A}.

Methods

Lifting bracket-compatible isomorphisms of abelianized pro-ℓℓ Galois groups to class-two pro-ℓℓ isomorphisms via fiber products and projectivity of Zℓ\mathbb{Z}_\ell-products.

Source excerpts

“In the fiber product P = E1 ×A E2, quotient by the diagonal central subgroup {(b, b) : b ∈ B}. The resulting group Q is abelian: the commutator of two pairs has identical components by bracket compatibility.” · Show in paper

Constructing finite residual tests by spreading out algebraic data and root covers to a finitely generated domain, specializing to a finite-characteristic residue field, and cutting down with Bertini hyperplane sections.

Source excerpts

“All these data descend to a finitely generated domain R ⊂ l in which ℓ is invertible.” · Show in paper

“Over λ, choose a general flag of d − 1 hyperplanes in the fixed projective embedding. Make the final section C avoid the omitted codimension-two set and meet the Di,s transversely.” · Show in paper

Bounding the divisor-support degree of an entire pencil uniformly by testing five fixed anchors at finite Kummer levels and applying the Riemann–Hurwitz different inequality on residual curves.

Source excerpts

“For a separable morphism of smooth curves, the different exponent at P is at least eP − 1, including in the presence of wild ramification. The five fibers are disjoint. Riemann–Hurwitz therefore yields 2g(C′) − 2 + 2n ≥ ∑5i=1 ∑P|β′i (eP − 1) ≥ 5(1 − 1/ℓ)n − ∑5i=1 Ni.” · Show in paper

Recovering curve subfields from infinitely many bounded-degree prime divisors by fixing an incidence curve in a Hilbert scheme family and using normal-base-change descent for arbitrary collections of Kummer tests.

Source excerpts

“Choose an integral locally closed curve T′ ⊂ T now, before making any of the tests in A, and put Z′ = Z ×T T′.” · Show in paper

“To pass to the fixed curve, descend the generic splitting to a finite extension of l(T) and normalize T in it, obtaining a finite surjective morphism Te → T.” · Show in paper