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.

Latest version

Checks complete

Sync OpenAI Math manuscripts from 4eaee87548250b2682affda7d05d947c05554bae

Committed

All papers

Claims

Findings

Nagata's conjecture for plane curves

For every integer r≥10r \ge 10, there is a countable union ErE_r of proper Zariski-closed subsets of Ur={(p1,…,pr)∈(PC2)r:pi≠pj for i≠j}U_r = \{(p_1, \dots, p_r) \in (\mathbb{P}^2_{\mathbb{C}})^r : p_i \ne p_j \text{ for } i \ne j\}, with nonempty complement, such that for every tuple (p1,…,pr)∉Er(p_1, \dots, p_r) \notin E_r, if CC is a nonzero effective plane curve of degree d≥1d \ge 1 and m1,…,mr∈Z≥0m_1, \dots, m_r \in \mathbb{Z}_{\ge 0} satisfy mult⁡piC≥mi\operatorname{mult}_{p_i} C \ge m_i, then dr>∑i=1rmid\sqrt{r} > \sum_{i=1}^r m_i.

For each r≥10r \ge 10, there is a countable union ErE_r of proper Zariski-closed subsets of UrU_r, all defined over Q\mathbb{Q}, with Ur∖Er≠∅U_r \setminus E_r \ne \emptyset, such that ruling out every universal system Ur(d,m)\mathsf{U}_r(d, m) with d/m≤rd/m \le \sqrt{r} suffices to prove Theorem 1.1 for that rr.

If Ur(d,m)\mathsf{U}_r(d, m) holds, then for every smooth irreducible plane curve B⊆P2B \subseteq \mathbb{P}^2 of degree k≥1k \ge 1, every choice of distinct points a1,…,ar∈Ba_1, \dots, a_r \in B, and arbitrary vectors ci∈Maic_i \in M_{a_i} (where L=OB(1)L = \mathcal{O}_B(1) and M=LkM = L^k), there is a nonzero polynomial F(w)=∑j=0⌊d/k⌋fjwjF(w) = \sum_{j=0}^{\lfloor d/k \rfloor} f_j w^j with fj∈H0(B,Ld−kj)f_j \in H^0(B, L^{d-kj}) having multiplicity at least mm at each (ai,ci)∈Tot⁡(M)(a_i, c_i) \in \operatorname{Tot}(M).

Let r=k2r = k^2 with k≥4k \ge 4. There are no positive integers d,md, m such that d≤kmd \le km and Ur(d,m)\mathsf{U}_r(d, m) holds.

If r≥10r \ge 10 and Ur(d,m)\mathsf{U}_r(d, m) holds for positive integers d,md, m, then d>3md > 3m.

If r≥10r \ge 10, d>3md > 3m, and Ur(d,m)\mathsf{U}_r(d, m) holds, then for any smooth plane cubic BB, nine distinct points P=a1+⋯+a9P = a_1 + \dots + a_9 on BB, and a holomorphic frame of A=M⊗OB(−P)A = M \otimes \mathcal{O}_B(-P) on a coordinate disk in B∖PB \setminus P, every choice of q=r−9q = r - 9 distinct points b1,…,bqb_1, \dots, b_q in this disk admits a nonzero F~∈W\widetilde{F} \in \mathcal{W} of multiplicity at least mm at each (bα,1)∈Tot⁡(A)(b_\alpha, 1) \in \operatorname{Tot}(A).

If Ur(d,m)\mathsf{U}_r(d, m) holds, then ker⁡Jτ≠{0}\ker J_\tau \ne \{0\} for every fixed τ∈(0,1)\tau \in (0, 1).

In the normalized theta basis {Ej,K,τ:(j,K)∈S}\{E_{j,K,\tau} : (j, K) \in \mathcal{S}\} of Wτ\mathcal{W}_\tau, the matrix BτB_\tau of JτJ_\tau converges entrywise as τ→0+\tau \to 0^+ to B0=(Kℓjb)(ℓ,b)∈I,(j,K)∈SB_0 = (K^\ell j^b)_{(\ell, b) \in \mathcal{I}, (j, K) \in \mathcal{S}}; if B0B_0 has full column rank, then JτJ_\tau is injective for all sufficiently small positive τ\tau.

If q(1−λ)m≥1q(1 - \lambda)m \ge 1, then the limiting matrix B0B_0 has full column rank, and consequently JτJ_\tau is injective for all sufficiently small positive τ\tau.

Let kk be an uncountable algebraically closed field of characteristic zero, and let r≥10r \ge 10. There is a countable union of proper Zariski-closed subsets of the space of distinct ordered rr-tuples in Pk2\mathbb{P}^2_k, with nonempty complement, outside which the conclusion of Theorem 1.1 holds with the same degree and multiplicity conventions.

On the blowup XX of the plane at a tuple (p1,…,pr)(p_1, \dots, p_r) covered by Theorem 1.1 or Corollary 6.1, the class Nr=rH−∑iEiN_r = \sqrt{r}H - \sum_i E_i is strictly nef and satisfies Nr2=0N_r^2 = 0; in particular, the multipoint Seshadri constant is ε(O(1);p1,…,pr)=1/r\varepsilon(\mathcal{O}(1); p_1, \dots, p_r) = 1/\sqrt{r}.

Methods

Cyclic-product reduction of a violation of Nagata's inequality at very general points to a universal equal-multiplicity system Ur(d,m)\mathsf{U}_r(d, m) existing at every configuration in UrU_r.

Source excerpts

“We use the cyclic-product reduction of Nagata [13]. Permuting the ordered points gives the same assertion for each cyclic shift of m. At any fixed tuple, choose a nonzero form for each of the r cyclic shifts and take their product.” · Show in paper

Specialization of a universal equal-multiplicity system to a polynomial on the normal bundle Tot⁡(M)\operatorname{Tot}(M) of a smooth plane curve BB via affine-linear motions pi(s)=ai+sVip_i(s) = a_i + s V_i and normal coordinate rescaling t=swt = sw.

Source excerpts

“We move the marked points in the plane, rescale the normal coordinate, and retain the first nonzero term of a family of plane forms.” · Show in paper

Elimination of nine prescribed zero normal displacements on a smooth plane cubic BB by dividing out σPm\sigma_P^m and changing the fiber coordinate via w=σPvw = \sigma_P v to obtain a finite-dimensional space W\mathcal{W} of polynomial sections on Tot⁡(A)\operatorname{Tot}(A).

Source excerpts

“Nine zero normal displacements impose zeros on the coefficients of a normal polynomial. Dividing out these zeros puts its first m + 1 coefficients in bundles of the same positive degree.” · Show in paper

Pointwise collision of the remaining q=r−9q = r - 9 points on a fixed elliptic curve XτX_\tau using Rouché's theorem (the argument principle) to convert multiplicity-mm conditions into a nontrivial kernel of the finite jet map JτJ_\tau.

Source excerpts

“For all sufficiently large ν, every ξa,ν is inside this disk, and uniform convergence on its boundary implies that hb,ν and hb,0 have the same number of zeros there, counted with multiplicity. The argument principle, or Rouché’s Theorem, therefore gives ord0 hb,0 ≥ q(m − b).” · Show in paper

Explicit normalized multiplicative theta-function bases Ej,K,τE_{j,K,\tau} on Xτ=C∗/τZX_\tau = \mathbb{C}^*/\tau^{\mathbb{Z}} whose local expressions converge uniformly with all mixed derivatives to eKx+jye^{Kx+jy} as τ→0+\tau \to 0^+.

Source excerpts

“In the local covering-space frame their expressions are sK,τ (z0ex) = ∑ p∈Z ϵ−pτ p(K−U)+np(p−1)/2e(K+np)x −→ eKx. (4.2) The convergence is uniform on each compact subset of the complex x-plane, and the same is true after any fixed number of x-derivatives.” · Show in paper

Horizontal-row polynomial interpolation on the finite exponent set S\mathcal{S} enclosed in a scaled, translated polygon mDλ+(0,U0)m\mathcal{D}_\lambda + (0, U_0) to prove full column rank of the limiting jet matrix B0B_0.

Source excerpts

“For the exponent set S, it remains to prove the two row bounds in this lemma. We will obtain them by enclosing S in a polygon and using its horizontal slice lengths to count the rows.” · Show in paper