Gabor frames for quasicrystals, \(K\)-theory, and twisted gap labeling (Q906312): Difference between revisions

\textit{Quasicrystals} \(\Lambda \subset \mathbb R^{2d}\) are generalizations of lattices which have enough lattice-like structure: are relatively dense (\(\rho(\Lambda) := \sup_{z\in \mathbb R^{2d}}\inf_{\lambda\in\Lambda} |z-\lambda|< \infty\)), unifomly discrete (up to translation there are only finitely many \(r\)-patches \(B_r(z) \cap \Lambda\) for all ball \(B_r(z)\) with radius \(r\)), have finite local complexity (FLC: there are only finitely many patches up to translation, for any real \(r\)) and have uniform cluster frequencies (UCF: there exist limits \(\mathrm{freq}(P,\Lambda) = \lim_{k\to \infty}\frac{L_P(B_{r_k}-z)}{vol(B_{r_k})}\) uniformly in \(z\) and independently of the choice of balls \(B_{r_k}\), where for each \(r\)-patch \(P\) and subset \(A\subset \mathbb R^{2d}\), \(L_P(A) = \{ z\in \mathbb R^{2d} | P-z \subset A \cap \Lambda \}\)). Two quasicrystals \(\lambda\) and \(\Lambda'\) are \textit{locally isomorphic} iff any \(r\)-patch that appeared in \(\Lambda\) also appears as an \(r\)-patch in \(\Lambda\) and conversely. The \textit{hull} \(\Omega_\Lambda\) of \(\Lambda\) is the equivalence class of all quasicrystals \(\lambda'\), locally isomorphic with \(\Lambda\). It is the closure of the \textit{orbit} \(O_\Lambda= \{ \Lambda -z| z\in \mathbb R^{2d}\}\). The \textit{canonical transversal} \(\Omega_{\mathrm{trans}}=\{ \Lambda' \in \Omega_\Lambda | 0 \in \Lambda'\}\) is the closure (a Kantor set) with measure \(\mu\), \(\mu(\Omega_\Lambda =\mathrm{freq}(P,\Lambda)\)) of the \(O^\Lambda_{\mathrm{trans}} := \{\Lambda -z | z\in \Lambda \}\). The groupoid \(C^*\)-algebra structure of \(\Lambda\) is introduced the groupoid structure on \(\Omega_{\mathrm{trans}}\) with the convolution product as: two elements \((T-z,T)\) and \((T'-w,T')\) can be composed iff \(T'=T-z\) and in that case \((T-z-w,T-z) * (T-z,T) := (T-z-w,T)\). If \(\theta\) is a 2-cocycle on \(\mathbb R^{2d}\) then on defines a \textit{standard 2-cocycle} \(\theta_\Lambda((T-z,T),(T'-w,T')) := \theta(z,w).\) For the symplectic cocycle \(\sigma\) on \(\mathbb R^{2d}\), given by \(\sigma((x,\omega),(x'\omega')) := e^{-2\pi i x\omega'}\), one defines the \(C^*\)-algebra \(\mathcal A_\theta = C^*(R_\lambda,\theta)\) of functions on \(R_\Lambda\) by the convolution product \[ f*g(T-z,T) := \sum_{w} f(T-z,T-w)g(T-w,T)\theta((T-z,T-w),(T-w,T)) \] and the involution \(f^*(T-z,T) = \overline{f(T,T-z)}\theta((T-z,T),(T,T-z))\). The \(C^*\)-algebra \(\mathcal A_\theta\) is Morita equivalent to the cross product \(C^*(C(\Omega_\Lambda)\rtimes \mathbb R^{2d}, \theta)\). The condition guarantees that the action of \(\mathbb R^{2d}\) on \(\Omega_\lambda\) is minimal and ergodic, the both \(C^*\)-algebras are simple and have a unique normalized trace given by \[ \mathrm{Tr}(f) = \int_{\Omega_{\mathrm{trans}}} f(T,T)dT, \quad f\in C_c(\Omega_\lambda) \] and following the Birkhoff's ergodic theorem, it can be written as \[ \mathrm{Tr}(f) = \lim_{k\to \infty} \frac{1}{|\Lambda \cap C_k|}\sum_{z\in\Lambda \cap C_k} f(T-z,T-z). \] If \(\Lambda \subset \mathbb R^d\) is a ``lattice with an aperiodic coloring satisfying the definition of a quasicrystal'', then \(\Omega_\lambda\) is a fiber bundle \(p: \Omega_\Lambda \twoheadrightarrow \mathbb T^d\) over torus \(\mathbb T^d\). The paper under reviewing is devoted to the problem of description of the image of trace map \(\mathrm{Tr}_* : K_0(\mathcal A_\theta) \to \mathbb R\). The main result of the paper is Theorem 3: The induced map \(p^*: K^0(\mathbb T^d) \rightarrowtail K^0(\Omega_\Lambda)\) is injective and the intersection of the images of \(p^*\) and of \(r_*: K_0(C(\Omega_{\mathrm{trans}}))\to K_0(C(\Omega_{\mathrm{trans}})\rtimes \mathbb Z^d)\cong K_0(C(\Omega_{\Lambda})\rtimes \mathbb R^d)\cong K^0(\Omega_\Lambda)\) induced by \(r: C(\Omega_{\mathrm{trans}}) \to C(\Omega_{\mathrm{trans}}) \rtimes Z^d\), is generated by the class of the trivial bundle. From author's summary: ``In particular, we construct a finitely generated projective module over this algebra, and any multiwindow Gabor frame for \(\Lambda\) can be used to construct an idempotent representing this module in K-theory. For lattice subsets in dimension two, this allows us to prove a twisted version of Bellissard's gap labeling theorem.''