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

Integral points on character varieties of curves

For any smooth connected complex algebraic curve XX, any rank r≥1r \ge 1, and prescribed quasi-unipotent conjugacy classes C1,…,Cs⊂SLr(C)C_1, \dots, C_s \subset \mathrm{SL}_r(\mathbb{C}) at the punctures, if K⊂CK \subset \mathbb{C} is a number field containing their eigenvalues, then there is a finite extension L/KL/K (explicitly F=K(2,ζpq)F = K(\sqrt{2}, \zeta_{pq}) and L=F(ζr,{u1/r:u∈OF×})L = F(\zeta_r, \{u^{1/r} : u \in \mathcal{O}_F^\times\}) for distinct primes p,q>rp, q > r) such that Mr(X)(OL)∩Y(C)M_r(X)(\mathcal{O}_L) \cap Y(\mathbb{C}) is Zariski dense in YY and in every irreducible component of YY, where YY is the reduced locus in Mr(X)CM_r(X)_\mathbb{C} of characters whose semisimple representations have ρ(γi)∈Ci\rho(\gamma_i) \in C_i. The same conclusion holds when only the characteristic polynomial of each boundary monodromy is prescribed (with all roots roots of unity), and for the entire character variety without boundary conditions (with K=QK = \mathbb{Q} in the projective case).

The locus YY of characters whose semisimple representations have exactly the prescribed boundary classes is Y=q(D)∖q(Dsm)Y = q(D) \setminus q(D_{\mathrm{sm}}), which is open in the closed subset q(D)q(D) and defined over any number field containing the boundary eigenvalues.

For every surface diagram DD with ranks at most rr and prescribed scalar puncture monodromies in FF, its integral solutions over R=OFR = \mathcal{O}_F are Zariski dense in S(D)\mathcal{S}(D).

Given two flags F∙,H∙F_\bullet, H_\bullet in a complex vector space VV, there are nonnegative integers mijm_{ij} and a sequence of allowed split, interchange, and merge operations refining F∙F_\bullet into blocks of ranks mijm_{ij} in row-first order, sorting them into column-first order by adjacent interchanges, and merging them to H∙H_\bullet, such that all flags and identified quotients admit a common splitting, and prescribed isomorphisms on FF-graded and HH-graded modules extend to an affine space of full isomorphisms (with free integral parameters over RR when graded modules are free).

Lifting a filtered exact sequence 0→Ui/Ui−1→Wi→Qi/Qi−1→00 \to U_i/U_{i-1} \to W_i \to Q_i/Q_{i-1} \to 0 to V=U⊕QV = U \oplus Q forms a nonempty homogeneous space for H=Hom(Q,U)H = \mathrm{Hom}(Q, U), and forgetting coarse identifications at a vertex gives an affine surjection H/Kcoarse→H/KintervalH/K_{\mathrm{coarse}} \to H/K_{\mathrm{interval}} with affine fiber Kinterval/KcoarseK_{\mathrm{interval}}/K_{\mathrm{coarse}} (having free integral parameters over RR when graded modules are free).

Let 2≤n≤r2 \le n \le r, and suppose integral solutions are dense for every surface diagram of ranks less than nn. Then they are dense for every normalized diagram of ranks at most nn whose rank-nn facets have nonempty boundary (closed facets of smaller rank are allowed).

Let R⊂CR \subset \mathbb{C} be an infinite domain with infinitely many units. If c∈Rc \in R is a root of unity and cn=1c^n = 1, then Rn,1(c)(R)\mathcal{R}_{n,1}(c)(R) is Zariski dense in Rn,1(c)\mathcal{R}_{n,1}(c).

For n≥2n \ge 2, g≥2g \ge 2, and cn=1c^n = 1, the variety Rn,g(c)\mathcal{R}_{n,g}(c) is irreducible of dimension (2g−1)n2+1(2g-1)n^2 + 1.

Let 2≤n≤r2 \le n \le r, g≥2g \ge 2, and let c∈Fc \in F satisfy cn=1c^n = 1. Assume that integral solutions are dense for every normalized diagram of ranks at most nn whose rank-nn facets have nonempty boundary. Then Rn,g(c)(R)\mathcal{R}_{n,g}(c)(R) is Zariski dense in Rn,g(c)\mathcal{R}_{n,g}(c).

Methods

Surface diagrams with filtered local systems, seams, and punctured disk caps to reduce rank by cutting maximal-rank facets to disks and passing to a subspace and quotient.

Source excerpts

“A surface diagram has finitely many compact oriented surfaces, called its facets. Each facet has an assigned rank between 0 and r and finitely many interior punctures.” · Show in paper

“There are two geometric stages. First, removing bands cuts the rank-n facets to disks; the two-flag lemma recovers each band through affine matching choices. Next, a proper subspace and its quotient replace each disk.” · Show in paper

Two-flag reconstruction and lifting of filtered exact sequences over the ring of integers R=OFR = \mathcal{O}_F using free graded modules.

Source excerpts

“The last two assertions hold over R when the labeled graded modules are free. The affine spaces then have free integral parameters.” · Show in paper

“These assertions hold over R when all named graded modules are free. After free splittings, all choices and all fibers have free integral parameters.” · Show in paper

Finite-field character counting via the Frobenius commutator formula combined with the Lang–Weil estimate and spreading out to prove geometric irreducibility of Rn,g(c)\mathcal{R}_{n,g}(c) in characteristic zero.

Source excerpts

“The proof below follows the character-count and Lang–Weil method of Liebeck–Shalev [12, Section 7, Lemma 7.1 and the proof of Theorem 1.8].” · Show in paper

Dehn twist saturation and a submersive family of centralizer twists at an explicit smooth representation to spread integral density from a single spectrum to the entire closed-surface variety Rn,g(c)\mathcal{R}_{n,g}(c).

Source excerpts

“Torus saturation by Dehn twists. The automorphisms τ±1(A1, B1, A2, B2, . . .) = (A1B1±1, B1, A2, B2, . . .) preserve V , since [A1B1, B1] = [A1, B1], and preserve its integral points.” · Show in paper

“Hence dΨ is surjective. Its image contains an analytic open subset of V . Such a subset is Zariski dense in an irreducible complex variety, proving D = V .” · Show in paper

Uniform determinant-one twisting over a single finite extension L=F(ζr,{u1/r:u∈OF×})L = F(\zeta_r, \{u^{1/r} : u \in \mathcal{O}_F^\times\}) via Dirichlet's unit theorem, followed by open selection of exact Jordan classes on the affine character quotient.

Source excerpts

“The field L = F(ζr, {u1/r : u ∈ R×}) is a finite extension of F. It contains every r-th root of every unit of R, and every such root belongs to O×L .” · Show in paper

“Scalar twisting defines f : S × T −→ Z, (s, t) ↦−→ t · s. It is the base change of the power map [r] : T → T.” · Show in paper