Asymptotics of greedy energy points

  • García A
  • Saff E
15Citations
Citations of this article
6Readers
Mendeley users who have this article in their library.

Abstract

For a symmetric kernel k : X × X → R ∪ { + ∞ } k:X\times X \rightarrow \mathbb {R}\cup \{+\infty \} on a locally compact metric space X X , we investigate the asymptotic behavior of greedy k k -energy points { a i } 1 ∞ \{a_{i}\}_{1}^{\infty } for a compact subset A ⊂ X A\subset X that are defined inductively by selecting a 1 ∈ A a_{1}\in A arbitrarily and a n + 1 a_{n+1} so that ∑ i = 1 n k ( a n + 1 , a i ) = inf x ∈ A ∑ i = 1 n k ( x , a i ) \sum _{i=1}^{n}k(a_{n+1},a_{i})=\inf _{x\in A}\sum _{i=1}^{n}k(x,a_{i}) . We give sufficient conditions under which these points (also known as Leja points) are asymptotically energy minimizing (i.e. have energy ∑ i ≠ j N k ( a i , a j ) \sum _{ieq j}^{N}k(a_{i},a_{j}) as N → ∞ N\rightarrow \infty that is asymptotically the same as E ( A , N ) := min { ∑ i ≠ j k ( x i , x j ) : x 1 , … , x N ∈ A } \mathcal {E}(A,N):=\min \{\sum _{ieq j}k(x_{i},x_{j}):x_{1},\ldots ,x_{N}\in A\} ), and have asymptotic distribution equal to the equilibrium measure for A A . For the case of Riesz kernels k s ( x , y ) := | x − y | − s k_{s}(x,y):=|x-y|^{-s} , s > 0 s>0 , we show that if A A is a rectifiable Jordan arc or closed curve in R p \mathbb {R}^{p} and s > 1 s>1 , then greedy k s k_{s} -energy points are not asymptotically energy minimizing, in contrast to the case s > 1 s>1 . (In fact, we show that no sequence of points can be asymptotically energy minimizing for s > 1 s>1 .) Additional results are obtained for greedy k s k_{s} -energy points on a sphere, for greedy best-packing points (the case s = ∞ s=\infty ), and for weighted Riesz kernels. For greedy best-packing points we provide a simple counterexample to a conjecture attributed to L. Bos.

Cite

CITATION STYLE

APA

García, A. L., & Saff, E. (2010). Asymptotics of greedy energy points. Mathematics of Computation, 79(272), 2287–2316. https://doi.org/10.1090/s0025-5718-10-02358-6

Register to see more suggestions

Mendeley helps you to discover research relevant for your work.

Already have an account?

Save time finding and organizing research with Mendeley

Sign up for free