Discrete Version of an Optimal Partitioning Problem

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Many recent works deal with problems concerning optimal partitions related to spectral quantities of domains in Euclidean spaces or on manifolds. Due to the complexity of these problems, few explicit solutions are known. Therefore, numerical algorithms have been developed in order to find approximations of optimal partitions. Such algorithms are based on discretizations of the domain and lead to finite dimensional difference equations. In the following, the coupling of the gradient descent method with a projection algorithm leads to a non-linear difference equation. Various properties of the discrete problem are discussed and numerical results illustrating the behaviour of the discretization scheme are shown.

Original languageEnglish
Title of host publicationDifference Equations, Discrete Dynamical Systems and Applications - ICDEA 23, 2017
EditorsSaber Elaydi, Christian Pötzsche, Adina Luminiţa Sasu
PublisherSpringer New York LLC
Pages247-256
Number of pages10
ISBN (Print)9783030200152
DOIs
Publication statusPublished - 1 Jan 2019
Event23rd International Conference on Difference Equations and Applications, ICDEA 2017 - Timişoara, Romania
Duration: 24 Jul 201728 Jul 2017

Publication series

NameSpringer Proceedings in Mathematics and Statistics
Volume287
ISSN (Print)2194-1009
ISSN (Electronic)2194-1017

Conference

Conference23rd International Conference on Difference Equations and Applications, ICDEA 2017
Country/TerritoryRomania
CityTimişoara
Period24/07/1728/07/17

Keywords

  • Eigenvalues
  • Finite differences
  • Numerical simulations
  • Optimal partitions

Fingerprint

Dive into the research topics of 'Discrete Version of an Optimal Partitioning Problem'. Together they form a unique fingerprint.

Cite this