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

Vanishing of higher bounded Hochschild cohomology

Let MM be a complex von Neumann algebra and let k≥2k \ge 2. Every f∈Cbk(M,M)f \in C_b^k(M, M) satisfying df=0df = 0 has the form f=dgf = dg for some g∈Cbk−1(M,M)g \in C_b^{k-1}(M, M), and consequently Hbk(M,M)=0H_b^k(M, M) = 0 (with the actual image of dd and no complete-boundedness hypothesis). Together with the degree-one inner-derivation theorem, this resolves the Kadison–Ringrose cohomology conjecture positively.

Let MM be a finite type II1\mathrm{II}_1 von Neumann algebra with separable predual and faithful normal tracial state τ\tau. There exist a symmetric, countably supported probability measure μ\mu on U(M)\mathcal{U}(M) and a strictly increasing sequence nℓ→∞n_\ell \to \infty such that the positive-mass atoms of μ\mu contain a countable 22-norm dense subset of U(M)\mathcal{U}(M) (and hence generate MM as a von Neumann algebra), and for every r∈Nr \in \mathbb{N}, every nonempty finite set W⊂Fr∖{e}\mathcal{W} \subset \mathbb{F}_r \setminus \{e\}, and every η>0\eta > 0, independent products G1(ℓ),…,Gr(ℓ)G_1^{(\ell)}, \dots, G_r^{(\ell)} with law μ∗nℓ\mu^{*n_\ell} satisfy Pr⁡(max⁡w∈W∣τ(w(G1(ℓ),…,Gr(ℓ)))∣≥η)→0\Pr(\max_{w \in \mathcal{W}} |\tau(w(G_1^{(\ell)}, \dots, G_r^{(\ell)}))| \ge \eta) \to 0.

Let (M,τ)(M, \tau) be a type II1\mathrm{II}_1 von Neumann algebra with separable predual and a faithful normal tracial state, and let μ\mu be the measure supplied by Theorem 2.1. If T:M→MT : M \to M is bounded, linear, and bounded-ball 22-continuous, and satisfies T(x)=EU∼μ[T(xU)U∗]T(x) = \mathbb{E}_{U \sim \mu}[T(xU)U^*] for all x∈Mx \in M, then T(x)=T(1)xT(x) = T(1)x for all x∈Mx \in M.

Let (N,τ)(N, \tau) be a finite von Neumann algebra with a faithful normal tracial state, and let R:N→NR : N \to N be bounded, complex linear, and bounded-ball 22-continuous. Let (gj)j≥1(g_j)_{j \ge 1} be a free Haar family in NN, and let a,b∈Na, b \in N. If R(w)=awR(w) = aw when ε1=1\varepsilon_1 = 1 and R(w)=bwR(w) = bw when ε1=−1\varepsilon_1 = -1 for every nonempty signed word w=gi1ε1⋯giqεqw = g_{i_1}^{\varepsilon_1} \cdots g_{i_q}^{\varepsilon_q} with εj∈{1,−1}\varepsilon_j \in \{1, -1\} and pairwise distinct indices i1,…,iqi_1, \dots, i_q, then a=ba = b.

Let MM be of type II1\mathrm{II}_1 with separable predual and faithful normal tracial state τ\tau. If k≥2k \ge 2 and f∈Cbk(M,M)f \in C_b^k(M, M) is a separately normal cocycle, then there is g∈Cbk−1(M,M)g \in C_b^{k-1}(M, M) such that dg=fdg = f and ∥g∥≤∥f∥\|g\| \le \|f\|.

Let MM be any type II1\mathrm{II}_1 von Neumann algebra admitting a faithful normal tracial state. Every separately normal cocycle f∈Cbk(M,M)f \in C_b^k(M, M), k≥2k \ge 2, has a bounded primitive gg with ∥g∥≤∥f∥\|g\| \le \|f\|.

For a central projection z∈Mz \in M, q=1−zq = 1 - z, n≥2n \ge 2, and ψ∈Cbn(M,zM)\psi \in C_b^n(M, zM), the operator JzJ_z defined by (Jzψ)(a1,…,an−1)=∑i=1n−1(−1)iψ(za1,…,zai−1,q,qai,ai+1,…,an−1)(J_z\psi)(a_1, \dots, a_{n-1}) = \sum_{i=1}^{n-1} (-1)^i \psi(za_1, \dots, za_{i-1}, q, qa_i, a_{i+1}, \dots, a_{n-1}) (with Jz=0J_z = 0 in degrees 0,10, 1) satisfies dJz+Jzd=id−cutzdJ_z + J_z d = \mathrm{id} - \mathrm{cut}_z in every degree and ∥Jzψ∥≤(n−1)∥ψ∥\|J_z \psi\| \le (n-1)\|\psi\| for n≥1n \ge 1.

Methods

Inductive construction of a symmetric, countably supported probability measure μ=p0μ0+∑ℓ≥1pℓμℓ\mu = p_0 \mu_0 + \sum_{\ell \ge 1} p_\ell \mu_\ell on U(M)\mathcal{U}(M) whose independent products have joint free Haar moments in probability.

Source excerpts

“We construct finite symmetric measures μℓ = 1/(2dℓ) ∑dℓj=1(δuℓ,j + δu−1ℓ,j) successively, and finally set μ = p0μ0 + ∑ℓ≥1 pℓμℓ.” · Show in paper

Harmonic-martingale construction Zn=D(Wn)=T(Wn)Wn∗Z_n = D(W_n) = T(W_n)W_n^* in L2(M,τ)L^2(M, \tau) combined with diagonal selection into a tracial ultrapower Mω1M^{\omega_1} to produce free Haar unitaries with first-letter multiplier identities.

Source excerpts

“Let U1, U2, . . . be independent with law μ, and set W0 = 1, Wn = U1 · · · Un, and Fn = σ(U1, . . . , Un). The H-valued random variables Zn = D(Wn) have countable range and satisfy ∥Zn∥2 ≤ ∥Zn∥ ≤ C.” · Show in paper

Iterated tracial ultrapower Nω2N^{\omega_2} and polar approximation of averaged free generators sr=r−1/2∑j=1rgjs_r = r^{-1/2} \sum_{j=1}^r g_j using Haagerup's free-group length estimate and Catalan moment computation.

Source excerpts

“Under the hypotheses of Theorem 4.1, set sr = r−1/2 ∑rj=1 gj ∈ N. (4.5) Lemma 4.2 gives ∥sr∥2 = 1 and ∥sr∥ ≤ 2 for every r. Fix a free ultrafilter ω2 on N and put Nˆ = Nω2, τˆ = τω2, s = (sr)ω2 ∈ Nˆ, S = Rω2. (4.6)” · Show in paper

Spectral arc partitioning of T\mathbb{T} with uniformly bounded sine sums Hm(z)=∑k=1mzk/kH_m(z) = \sum_{k=1}^m z^k / k to force equality of left multipliers in a finite von Neumann algebra.

Source excerpts

“We use analytic polynomials whose imaginary parts remain uniformly bounded on the unit circle while their real parts grow near 1. Rotating these polynomials will bound a − b on each short spectral arc; the trace will then combine the compression estimates without a loss depending on the number of arcs.” · Show in paper

Cesàro averaging of the last cocycle input with respect to μ\mu and passing to a pointwise ultraweak cluster point.

Source excerpts

“For a bounded k-cochain Φ, define (PΦ)(a, x) = ∑s≥1 psΦ(a, xus)u∗s, a = (a1, . . . , ak−1). (5.2) ... Set Fn = 1/n ∑n−1j=0 Pjf.” · Show in paper

Explicit central-cut homotopy JzJ_z to correct mixed inputs outside a central summand zMzM with a uniform norm bound independent of zz.

Source excerpts

“In the present setting we use an explicit homotopy to correct all inputs outside a given central summand before assembling the primitives.” · Show in paper