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

The Deligne-Drinfeld conjecture

There are homogeneous elements σ2k+1∈W2k+1\sigma_{2k+1} \in W_{2k+1}, one for each k≥1k \ge 1, for which LieQ⟨e3,e5,e7,… ⟩→(W,{,})\mathrm{Lie}_{\mathbb{Q}}\langle e_3, e_5, e_7, \dots \rangle \to (W, \{,\}), e2k+1↦σ2k+1e_{2k+1} \mapsto \sigma_{2k+1}, is an isomorphism of graded Lie algebras; in particular, WW is closed under the Ihara bracket, and the induced map on completions by weight is a continuous graded isomorphism.

For every nn and r≥0r \ge 0, there is an injection grFrL‾n↪fodd,n,r\mathrm{gr}_F^r \overline{L}_n \hookrightarrow \mathfrak{f}_{\mathrm{odd},n,r} obtained by the leading projection and the identification gk↔zkg_k \leftrightarrow z_k; in particular, dim⁡QWn=dim⁡F2L‾n≤∑r≥0dn,r=dn\dim_{\mathbb{Q}} W_n = \dim_{\mathbb{F}_2} \overline{L}_n \le \sum_{r \ge 0} d_{n,r} = d_n.

For n>2n > 2 and every r≥0r \ge 0, the leading projection grFrL‾n→LieF2⟨C,B⟩\mathrm{gr}_F^r \overline{L}_n \to \mathrm{Lie}_{\mathbb{F}_2}\langle C, B \rangle, χ↦χ(0,C,B)\chi \mapsto \chi(0, C, B) is injective, and its source is zero unless n=2m+rn = 2m + r for an integer m≥0m \ge 0, in which case its image has CC-count mm and BB-count rr.

The elements gk=adCkBg_k = \mathrm{ad}_C^k B for k≥1k \ge 1 freely generate an ordinary free Lie algebra inside LieF2⟨C,B⟩\mathrm{Lie}_{\mathbb{F}_2}\langle C, B \rangle, and for n>2n > 2, the image of the leading projection of grFrL‾n\mathrm{gr}_F^r \overline{L}_n belongs to its weight-nn, length-rr component, where wt(gk)=2k+1\mathrm{wt}(g_k) = 2k + 1 and each gkg_k has length one; the associated graded piece with r=0r = 0 is zero.

For n≥2n \ge 2, evaluation δ↦δ(α)\delta \mapsto \delta(\alpha) takes Dn(Q)\mathcal{D}_n(\mathbb{Q}) into WnW_n, and its images Vn={δ(α):δ∈Dn(Q)}V_n = \{\delta(\alpha) : \delta \in \mathcal{D}_n(\mathbb{Q})\}, V=⨁n≥2VnV = \bigoplus_{n \ge 2} V_n, form a rational Lie subalgebra under the Ihara bracket.

For every odd n≥3n \ge 3, there exists ψn∈Vn\psi_n \in V_n for which the coefficient of xn−1yx^{n-1}y is nonzero, or equivalently, its ordinary yy-depth-one part is a nonzero multiple of adxn−1y\mathrm{ad}_x^{n-1} y.

Let ψ∈FrL‾n\psi \in F^r \overline{L}_n and ϕ∈FsL‾n′\phi \in F^s \overline{L}_{n'}, where n,n′>2n, n' > 2 and r,s≥1r, s \ge 1, and write f,gf, g for their leading projections at counts r,sr, s. Their Ihara bracket, computed in the ambient Lie algebra in x,yx, y, belongs to Fr+sF^{r+s} in the A,C,BA, C, B presentation, and the projection of its component of count r+sr + s is [f,g][f, g].

Over Q\mathbb{Q}, the zeroth cohomology of Kontsevich's graph complex satisfies H0(GC2)≅LieQ⟨e3,e5,e7,… ⟩^H^0(\mathrm{GC}_2) \cong \widehat{\mathrm{Lie}_{\mathbb{Q}}\langle e_3, e_5, e_7, \dots \rangle}.

Methods

Characteristic-two filtered degeneration of the pentagon: embedding LieF2⟨x,y⟩\mathrm{Lie}_{\mathbb{F}_2}\langle x, y \rangle (in weight >1> 1) into LieF2⟨A,C,B⟩\mathrm{Lie}_{\mathbb{F}_2}\langle A, C, B \rangle via A↦x2,C↦[x,y],B↦yA \mapsto x^2, C \mapsto [x,y], B \mapsto y, filtering by BB-count, and transferring the pentagon relation to the level-two cyclotomic hyperplane arrangement enveloping algebra U(h)U(\mathfrak{h}) via a deterministic ordering algorithm.

Source excerpts

“The main step is a degeneration of the pentagon in characteristic two. Inside the free associative algebra, put A = x 2 , C = [x, y], B = y, and filter by the number of occurrences of B. The pentagon for a leading part becomes a relation in the enveloping algebra of the level-two cyclotomic hyperplane arrangement [7, Section 1.4], modulo terms of lower count.” · Show in paper

Integral deletion operators on the ratio alphabet {0,e,p,w}\{0, e, p, w\}: specializing the integral algebraic differential operators of Hirose and Sato to obtain an integral representation of hZ\mathfrak{h}_{\mathbb{Z}} on ENE_N and scalar deletion tests modulo two.

Source excerpts

“The operators below specialize the integral algebraic differential operators of Hirose and Sato [10, Section 2]. Their logarithmic-form and cyclotomic braid interpretation follows Hirose’s construction [9, Section 4.3]. Here we construct an integral representation of the signed-hyperplane algebra of Section 3. After reduction modulo two it will turn the leading pentagon into individual coefficient equations.” · Show in paper

Ore extension polynomial shift invariance and Vandermonde determinant confinement: using E=S[X;D]/(X2)\mathcal{E} = S[X; D]/(X^2) to prove f(C,B+s(C))=f(C,B)f(C, B + s(C)) = f(C, B) and a Vandermonde matrix over F2(t0,…,tr−1)\mathbb{F}_2(t_0, \dots, t_{r-1}) to confine the leading projection to LieF2⟨adCkB:k≥1⟩\mathrm{Lie}_{\mathbb{F}_2}\langle \mathrm{ad}_C^k B : k \ge 1 \rangle.

Source excerpts

“Consequently the Ore extension, followed by the indicated quotient, E = S[X; D]/(X2 ), Xq = qX + Dq (q ∈ S), has underlying left S-module S ⊕ SX.” · Show in paper

“Taking j = 0, . . . , r − 1 produces a Vandermonde matrix whose determinant is Y 0≤a<b≤r−1 (tb − ta). It is a nonzero polynomial over F2, because the ti are independent indeterminates.” · Show in paper

Parenthesized-chord categorical derivations and regularized braid holonomy: defining compatible derivations Dn(Q)\mathcal{D}_n(\mathbb{Q}) on parenthesized-chord categories C(I;K)\mathcal{C}(I; K) up to arity four, and constructing odd-weight depth-one elements via the operator logarithm δ=log⁡(H−H+−1)\delta = \log(H_- H_+^{-1}) of regularized transport for the universal logarithmic braid connection.

Source excerpts

“We use the parenthesized-chord formalism of Bar-Natan [2, Sections 2–3], restricted to the finite arities and operations specified below.” · Show in paper

“The proposed regularized rule is H+(γ) = lim ϵ→0+ Eq(ϵ) −1G(γϵ)Ep(ϵ).” · Show in paper

Saturated Z(2)\mathbb{Z}_{(2)}-lattice dimension squeeze: comparing the F2\mathbb{F}_2-dimension of saturated reductions M‾n⊂L‾n\overline{M}_n \subset \overline{L}_n with the free Lie dimension dnd_n to inductively lift integral generators σ2k+1\sigma_{2k+1} with prescribed leading BB-count-one projection.

Source excerpts

“We construct σ2k+1 ∈ M2k+1 inductively so that its reduction has leading B-count one, with projection π in1 σ2k+1 = adk C B.” · Show in paper