OctoInformBeta

Cuntz comparison, nuclear dimension, and equivariant Jiang–Su stability

For a unital C*C^*-algebra DD, write T(D)T(D) for its tracial state space. For d∈Dd\in D and τ∈T(D)\tau\in T(D), write ∥d∥2,τ=τ(d*d)1/2\|d\|_{2,\tau}=\tau(d^*d)^{1/2}. Let AA be a unital C*C^*-algebra with T(A)≠⌀T(A)\neq\varnothing. The uniform 22-seminorm on AA is ∥a∥2,T(A)=sup⁡τ∈T(A)∥a∥2,τ\|a\|_{2,T(A)}=\sup_{\tau\in T(A)}\|a\|_{2,\tau}. The uniform tracial completion BB of AA is the quotient of the C*C^*-algebra of norm-bounded sequences in AA that are Cauchy for ∥⋅∥2,T(A)\|\cdot\|_{2,T(A)} by the ideal of those that tend to zero for it. For τ∈T(A)\tau\in T(A), write τ‾([(aj)])=lim⁡jτ(aj)\bar\tau([(a_j)])=\lim_j\tau(a_j) for the trace it defines on BB. Fix a free ultrafilter ω\omega on ℕ\mathbb N. The uniform tracial ultrapower BωB^\omega is the quotient of the C*C^*-algebra of norm-bounded sequences (bj)(b_j) in BB by the ideal of those with lim⁡j→ωsup⁡τ∈T(A)∥bj∥2,τ‾=0\lim_{j\to\omega}\sup_{\tau\in T(A)}\|b_j\|_{2,\bar\tau}=0. Each sequence (τj)(\tau_j) in T(A)T(A) defines a limit trace λ([(bj)])=lim⁡j→ωτ‾j(bj)\lambda([(b_j)])=\lim_{j\to\omega}\bar\tau_j(b_j) on BωB^\omega. Write Λω\Lambda_\omega for the family of these limit traces. The algebra AA has uniform property Γ\Gamma at ω\omega when BωB^\omega contains a projection pp that commutes with BB, embedded as the constant sequences, and satisfies λ(px)=12λ(x)\lambda(px)=\tfrac12\lambda(x) for every x∈Bx\in B and every λ∈Λω\lambda\in\Lambda_\omega. Whether real rank zero of BωB^\omega implies uniform property Γ\Gamma is Problem XXI of Schafhauser, Tikuisis and White.

Vaccaro derived uniform property Γ\Gamma from real rank zero of the tracial ultrapower together with tracially locally finite nuclear dimension, and Kessinger and Toms obtained it for classes of inductive limits with subquadratic growth. Nuclearity alone does not ensure uniform property Γ\Gamma, since Toms constructed a simple, separable, unital AH algebra without it.

The preprint answers Problem XXI affirmatively: if AA is unital, separable, simple, infinite-dimensional, nuclear and stably finite with T(A)≠⌀T(A)\neq\varnothing, and BωB^\omega has real rank zero, then AA has uniform property Γ\Gamma at ω\omega (Theorem 1.1). A projection is full when the closed two-sided ideal it generates is the whole algebra. The preprint also proves a finite-set central splitting (Theorem 1.3): if A⊆DA\subseteq D is a unital inclusion of C*C^*-algebras with AA nuclear, DD of real rank zero and T(D)≠⌀T(D)\neq\varnothing, and for every δ>0\delta>0 there is a full projection g∈Dg\in D with sup⁡τ∈T(D)τ(g)<δ\sup_{\tau\in T(D)}\tau(g)<\delta, then for every finite set KK of unitaries of AA and every ε>0\varepsilon>0 there is a projection P∈DP\in D with sup⁡τ∈T(D)∥[P,x]∥2,τ<ε\sup_{\tau\in T(D)}\|[P,x]\|_{2,\tau}<\varepsilon and sup⁡τ∈T(D)|τ(Px)−12τ(x)|<ε\sup_{\tau\in T(D)}|\tau(Px)-\tfrac12\tau(x)|<\varepsilon for every x∈Kx\in K. From the hypotheses of Theorem 1.1 the preprint further derives that restriction along the map A→BA\to B sending aa to the class of the constant sequence (a)(a) is an affine homeomorphism of T(B)T(B) onto T(A)T(A), which the Lean library does not contain.

The Lean library TracialSplitting proves Theorem 1.1 and Theorem 1.3 in the forms stated above. For a finite set KK of unitaries of AA and μ>0\mu>0, the approximate diagonals of the preprint’s Lemma 4.1 are the finite convex combinations d=∑ltlvl⊗vl*d=\sum_l t_l v_l\otimes v_l^* over contractions vl∈Av_l\in A with ∥x⋅d−d⋅x∥π<μ\|x\cdot d-d\cdot x\|_\pi<\mu for every x∈Kx\in K and ∥1−∑ltlvlvl*∥<μ\|1-\sum_l t_l v_lv_l^*\|<\mu, where ∥⋅∥π\|\cdot\|_\pi is the projective tensor norm on A⊗̂AA\widehat\otimes A. The Lean library also proves Theorem 1.3 with nuclearity replaced by the hypothesis that AA has approximate diagonals for every such KK and μ\mu. Haagerup’s virtual diagonal theorem provides them for every nuclear AA. Under the same hypothesis, the Lean library proves the conclusion of Theorem 1.1 without nuclearity or stable finiteness.

The results on these pages are those of the Lean library, each stated as the Lean library states it and linked to the Lean declaration behind it. Lean has checked every one of them, so the Lean library is the reference: where the preprint states a result differently, the statement on the page is the Lean library’s, and the preprint’s is given beside it. The preprint gives the results their order and their names.

The Lean library: TracialSplitting

TracialSplitting is a Lean library: 23 items, in six groups. Each has a page written by OctoInform, listed below.

§1. Introduction 3 items
  1. §1 (setting): the hypotheses of Theorem 1.1 Five notions behind the hypotheses of Theorem 1.1. §1
  2. Theorem 1.1 [Real rank zero and uniform property Γ] When real rank zero yields uniform property Γ\Gamma. §1
  3. Theorem 1.3 [Finite-set central splitting] Tracially halving projections almost commuting with finitely many unitaries. §1
§2. Notation and the tracial completion 5 items
  1. §2 (setting): tracial seminorms, fullness and subequivalence Trace norms, fullness and subequivalence in a C*^*-algebra. §2
  2. §2 (2.1): tracial inequalities Cauchy-Schwarz and multiplication bounds for the tracial 22-norm. §2
  3. §2 (setting): contractions as combinations of unitaries Commutators with contractions are tested on finitely many unitaries. §2
  4. §2.1 (setting): the uniform tracial completion, the ultrapower and the limit traces (2.2), (2.3) Constructing the uniform tracial completion, its ultrapower and limit traces. §2.1
  5. §2.1 (2.4): uniform property Γ Uniform property Γ\Gamma as a halving projection in the ultrapower. §2.1
§3. Projection tools 7 items
  1. Lemma 3.1 [Scalar compressions] Real rank zero passes to corners and gives scalar compressions. §3
  2. Lemma 3.2 [Full positive elements with small tracial support] Positive elements of small trace in infinite-dimensional simple C*-algebras. §3
  3. Lemma 3.3 [Small full projections] Full projections of arbitrarily small trace. §3
  4. Lemma 3.4 [Splitting a column] Splitting a projection into orthogonal pieces along a column. §3
  5. Proposition 3.5 [Uniform proportional subprojections] Subprojections carrying nearly a given fraction of every trace at once. §3
  6. Corollary 3.6 [Splitting finite linear tests] Splitting a projection by any fraction on finitely many tests. §3
  7. Lemma 3.7 [Orthogonal insertion] Replacing a projection by one orthogonal to another. §3
§4. Nuclear averaging and weighted matrix blocks 3 items
  1. Lemma 4.1 [Averaging with trace control] Averaging by conjugation with control of traces. §4
  2. Lemma 4.2 [A square-root trace estimate] The trace one-norm bounds the distance between square roots. §4
  3. Proposition 4.3 [Weighted matrix blocks] Nuclear averaging cuts projections into weighted matrix blocks. §4
§5. Sampling and orthogonal packing 3 items
  1. Lemma 5.1 [Sampling a matrix block] Sampling a matrix block approximately scales traces and commutator norms. §5
  2. Lemma 5.2 [Energy of orthogonal sums] Squared commutator 2-norms are subadditive over orthogonal sums of projections. §5
  3. §5 (proof of Theorem 1.3): the packing rounds Packing rounds build orthogonal projection pairs with explicit bounds. §5
§6. The projection in the prescribed ultrapower 2 items
  1. Lemma 6.1 [Detection by limit traces] Limit traces compute uniform tracial norms in an ultrapower. §6
  2. §6 (proof of Theorem 1.1): diagonal argument and density From approximate to exact halving projections in the ultrapower. §6

No item matches.