Sergei Solodky
Institute of Mathematics of NASU, Ukraine,
University of Giessen, Germany
solodky@imath.kiev.ua
Optimization of Numerical Differentiation Methods.
Approximation and Information Aspects
Friday, 20 October 2023 at 12:00
On line: https://meet.google.com/afw-vnwo-dfb
Abstract
In the talk, I will incorporate the so-called self-regularization into the numerical differentiation of multivariable functions. The proposed approach is a combination of the truncation method and a discretization scheme using the idea of a hyperbolic cross. It will be shown that numerical differentiation methods constructed in this way not only have a simple implementation but are also optimal in terms of the accuracy and volume of discrete information used.
Seminar video report
About the author