Convergence of random graph Laplacian eigenmaps on an interval

Let ϕk\phi_k denote the kthk^\text{th} eigenfunction of the Laplacian on [1,1][-1,1], and let ϕϵ,n,k\phi_{\epsilon,n,k} denote the kthk^\text{th} eigenfunction of a graph Laplacian Lϵ,n\mathcal{L}_{\epsilon,n}. Convergence conjecture. If 0<β<10<\beta<1 and ϵ(n)nβ\epsilon(n)\sim n^{-\beta}, then

ϕkϕϵ(n),n,kL0\|\phi_k-\phi_{\epsilon(n),n,k}\|_{L^\infty}\xrightarrow[]{}0

as nn\to\infty. For 0<β<2/30<\beta<2/3, the convergence is improved to convergence in energy, and for 0<β<1/30<\beta<1/3, it is further improved to convergence of the Laplacians. These conjectured convergence rates would substantially improve the bounds supplied by the preceding theorems for low-frequency eigenfunctions.

Sources & referencesView supporting material

Primary source

Bernard Akwei, Bobita Atkins, Rachel Bailey, Ashka Dalal, Natalie Dinin, Jonathan Kerby-White, Tess McGuinness, Tonya Patricks, Luke Rogers, Genevieve Romanelli, Yiheng Su and Alexander Teplyaev, “Convergence, optimization and stability of singular eigenmaps”, arXiv:2406.19510 (2024).

Progress summary

Never refreshed

Nothing recorded yet. Refresh searches the literature and the public web for attempts on this problem, and writes the first summary here.

Solutions 0

No solutions have been posted yet.