Mollification formulas and implicit smoothing

Mollification formulas and implicit smoothing
复制标题

DOI:
10.1007/s10444-005-7512-3
复制
发表时间:
2007-08-01
影响因子:
1.7
通讯作者:
Bui, H.-Q.
Bui, H.-Q.
中科院分区:
数学4区
文献类型:
--
作者:
Beatson, R. K.;Bui, H.-Q.

文献摘要

被引文献

相似文献

本文开发了一些涉及流行的径向基函数 (RBF) 基本函数 Phi 和合适的缓和器之间的卷积的缓和公式。考虑了多调和样条、缩放贝塞尔核(Matern 函数)和紧支持的基本函数。典型的结果是,在 R-d 中,|.|(beta) 和 (.(2) + c(2))-((beta+2d)/2) 的卷积是广义多重二次函数 (.(2) + c(2))(beta/2) 直至乘法常数。该常数取决于 c > 0、beta,其中 R(beta) > - d 和 d。推动公式发展的一个应用是一种称为隐式平滑的技术。这种计算效率高的技术通过在评估过程中用更平滑的版本 Psi 替换基本函数 Phi 来平滑先前获得的 RBF 拟合。
This paper develops some mollification formulas involving convolutions between popular radial basis function (RBF) basic functions Phi, and suitable mollifiers. Polyharmonic splines, scaled Bessel kernels (Matern functions) and compactly supported basic functions are considered. A typical result is that in R-d the convolution of |.|(beta) and (.(2) + c(2))-((beta+2d)/2) is the generalized multiquadric (.(2) + c(2))(beta/2) up to a multiplicative constant. The constant depends on c > 0, beta, where R(beta) > - d, and d. An application which motivated the development of the formulas is a technique called implicit smoothing. This computationally efficient technique smooths a previously obtained RBF fit by replacing the basic function Phi with a smoother version Psi during evaluation.