We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:


Current browse context:


Change to browse by:

References & Citations


(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Mathematics > Analysis of PDEs

Title: On a quantum-classical correspondence: from graphs to manifolds

Authors: Akshat Kumar
Abstract: We establish conditions for which graph Laplacians $\Delta_{\lambda,\epsilon}$ on compact, boundaryless, smooth submanifolds $\mathcal{M}$ of Euclidean space are semiclassical pseudodifferential operators ($\Psi$DOs): essentially, that the graph Laplacian's kernel bandwidth ($\textit{bias term}$) $\sqrt{\epsilon}$ decays faster than the semiclassical parameter $h$, $\textit{i.e.}$, $h \gg \sqrt{\epsilon}$ and we compute the symbol. Coupling this with Egorov's theorem and coherent states $\psi_h$ localized at $(x_0, \xi_0) \in T^*\mathcal{M}$, we show that with $U_{\lambda,\epsilon}^t := e^{-i t \sqrt{\Delta}_{\lambda,\epsilon}}$ spectrally defined, the (co-)geodesic flow $\Gamma^t$ on $T^*\mathcal{M}$ is approximated by $\langle U_{\lambda,\epsilon}^{-t} \operatorname{Op}_h(a) U_{\lambda,\epsilon}^t \psi_h, \psi_h \rangle = a \circ \Gamma^t(x_0, \xi_0) + O(h)$. Then, we turn to the discrete setting: for $\Delta_{\lambda,\epsilon,N}$ a normalized graph Laplacian defined on a set of $N$ points $x_1, \ldots, x_N$ sampled $\textit{i.i.d.}$ from a probability distribution with smooth density, we establish Bernstein-type lower bounds on the probability that $||U_{\lambda,\epsilon,N}^t[u] - U_{\lambda,\epsilon}^t[u]||_{L^{\infty}} \leq \delta$ with $U_{\lambda,\epsilon,N}^t := e^{-i t \sqrt{\Delta}_{\lambda,\epsilon,N}}$. We apply this to coherent states to show that the geodesic flow on $\mathcal{M}$ can be approximated by matrix dynamics on the discrete sample set, namely that $\textit{with high probability}$, $c_{t,N}^{-1} \sum_{j=1}^N |U_{\lambda,\epsilon,N}^t[\psi_h](x_j)|^2 u(x_j) = u(x_t) + O(h)$ for $c_{t,N} := \sum_{j=1}^N |U_{\lambda,\epsilon,N}^t[\psi_h](x_j)|^2$ and $x_t$ the projection of $\Gamma^t(x_0, \xi_0)$ onto $\mathcal{M}$.
Comments: This is a companion paper to "Manifold learning via quantum dynamics"
Subjects: Analysis of PDEs (math.AP); Differential Geometry (math.DG)
Cite as: arXiv:2112.10748 [math.AP]
  (or arXiv:2112.10748v2 [math.AP] for this version)

Submission history

From: Akshat Kumar [view email]
[v1] Mon, 20 Dec 2021 18:54:06 GMT (94kb)
[v2] Tue, 13 Dec 2022 19:14:06 GMT (343kb,D)

Link back to: arXiv, form interface, contact.