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

An irrational-trace counterexample to reduced Baum–Connes

There exist a finitely generated discrete group GsurG_{\mathrm{sur}} and a projection b0∈Cr∗(Gsur)b_0 \in C_r^*(G_{\mathrm{sur}}) with irrational canonical trace, such that [b0]∉μ0Gsur(K0Gsur(E‾Gsur))[b_0] \notin \mu_0^{G_{\mathrm{sur}}}(K_0^{G_{\mathrm{sur}}}(\underline{E}G_{\mathrm{sur}})), which disproves the coefficient-free reduced Baum--Connes conjecture for countable discrete groups and persists after rationalization.

In the coned Cayley graph X^\widehat{X} of the graphical presentation Γ\Gamma, for any group vertex or center RR and any center A=cP≠RA = c_P \neq R with a chosen geodesic from AA to RR whose first group vertex is a∈Pa \in P, every vertex shared by PP and any different copy QQ with dX^(cQ,R)≤dX^(cP,R)d_{\widehat{X}}(c_Q, R) \le d_{\widehat{X}}(c_P, R) lies in the intrinsic induced ball BPB_P about aa of radius rP=⌊12(L0+1)λσ(P)⌋r_P = \lfloor 12(L_0 + 1)\lambda \sigma(P) \rfloor.

There exist labelled dd-regular expander graphs (Θi)i≥1(\Theta_i)_{i \ge 1} satisfying the hypotheses of Section 2 with uniform spectral gap ρ>0\rho > 0, edge-expansion constant κ>0\kappa > 0, girth σi>max⁡{16,20/c}\sigma_i > \max\{16, 20/c\}, strictly increasing ⌊λσi⌋\lfloor \lambda \sigma_i \rfloor, and ki+1−ki→∞k_{i+1} - k_i \to \infty (where ki=⌊ni/2⌋k_i = \lfloor n_i/2 \rfloor), such that every overlap ball BPB_P in Lemma 2.4 is an induced tree with ∣BP∣<η∣P∣|B_P| < \eta |P| and the inclusions of the copies give an internal algebraic direct sum Z1(X;F2)=⨁PZ1(P;F2)Z_1(X; \mathbb{F}_2) = \bigoplus_P Z_1(P; \mathbb{F}_2).

For every ii, there exists a kik_i-dimensional code Ci⊆FpVert(Θi)C_i \subseteq \mathbb{F}_p^{\mathrm{Vert}(\Theta_i)} containing wiw_i such that ∣supp z∣>ηni|\mathrm{supp}\, z| > \eta n_i for every nonzero z∈Ci∪Ci⊥z \in C_i \cup C_i^\perp; moreover, in V=Fp(Γ)/W0V = \mathbb{F}_p^{(\Gamma)} / W_0 where W0=∑g∈Γ,i≥1CPg,i⊥W_0 = \sum_{g \in \Gamma, i \ge 1} C_{P_{g,i}}^\perp, we have W0∩FpP=CP⊥W_0 \cap \mathbb{F}_p^P = C_P^\perp for every copy PP, and the pushforward of mΩm_\Omega under resP(ω)=Xω∣P\mathrm{res}_P(\omega) = X_\omega|_P is the uniform measure on CPC_P, giving mΩ{ω:Xω∣P=wP}=p−kim_\Omega\{\omega : X_\omega|_P = w_P\} = p^{-k_i}.

The self-adjoint operator T=dI−∑s∈S±1fsusfs−1∈Cr∗(Gsur)T = dI - \sum_{s \in S^{\pm 1}} f_s u_s f_{s^{-1}} \in C_r^*(G_{\mathrm{sur}}) satisfies specCr∗(Gsur)(T)⊆{0}∪[ε,2d]\mathrm{spec}_{C_r^*(G_{\mathrm{sur}})}(T) \subseteq \{0\} \cup [\varepsilon, 2d] with ε=min⁡{ρ,c2/(8d)}>0\varepsilon = \min\{\rho, c^2/(8d)\} > 0, and continuous functional calculus yields a projection b0=χ(T)∈Cr∗(Gsur)b_0 = \chi(T) \in C_r^*(G_{\mathrm{sur}}) such that πω(b0)\pi_\omega(b_0) is the orthogonal projection onto ker⁡πω(T)\ker \pi_\omega(T) for every ω∈Ω\omega \in \Omega.

The projection b0∈Cr∗(Gsur)b_0 \in C_r^*(G_{\mathrm{sur}}) has canonical trace τGsur(b0)=∑i≥1p−ki∉Q\tau_{G_{\mathrm{sur}}}(b_0) = \sum_{i \ge 1} p^{-k_i} \notin \mathbb{Q}.

Methods

Coned Cayley graph construction and minimal van Kampen diagrams with adjustable center-passage sides to bound arc lengths and establish uniform thinness and simultaneous overlap balls.

Source excerpts

“Start a nonempty closed edge walk in Xb at a group vertex and partition it into m sides. Each side is a generator edge in X or a two-edge passage through a center.” · Show in paper

Selection of a subsequence of Lubotzky-Phillips-Sarnak d-regular expanders with growing girth and diverging dimension gaps, labelled over a single finite alphabet via Osajda's labelling theorem to obtain isometrically embedded relators.

Source excerpts

“Osajda’s labelling theorem [22, Theorem 2.7], applied to the already chosen λ, gives a reduced labelling over one finite alphabet with no repeated long path word.” · Show in paper

Probabilistic Gilbert-Varshamov counting argument over F_p to construct k_i-dimensional linear codes C_i containing a prescribed prototype pattern w_i with both primal and dual minimum support size greater than eta * n_i.

Source excerpts

“The following Gilbert–Varshamov type counting argument [9, 24] obtains both bounds while keeping wi in the code.” · Show in paper

Construction of the finitely generated semidirect product G_sur = V x| Gamma from a quotient finite-field module V = F_p(Gamma)p^{(Gamma)} / W_0, and identification of C*_r(G_sur) with C(Omega) x|_{alpha, r} Gamma via Fourier transformation on the compact dual Omega.

Source excerpts

“Define the discrete additive group and distinguished element V = F(Γ)p /W0, v∗ = [e1] ∈ V, with the induced Γ-action, and put Gsur = V ⋊ Γ.” · Show in paper

“Fourier transformation on the discrete abelian group V identifies C∗r (V ) with C(Ω).” · Show in paper

Elimination of copy coordinates in the algebraic direct sum Z_1(X; F_2) = bigoplus_P Z_1(P; F_2) ordered by distance to a root, charging cycle ranks to disjoint exterior vertex sets, combined with a discrete Cheeger and coarea estimate to prove a uniform spectral gap at 0 for T in C*_r(G_sur).

Source excerpts

“We will obtain this bound by charging the ranks of successive copy coordinates to pairwise disjoint subsets of Y .” · Show in paper

“We give the coarea and Cauchy–Schwarz calculation to account for the vertex potential d − degω(x) and the unnormalized Laplacian used here.” · Show in paper