MULTILEVEL AND ADAPTIVE METHODS FOR NONLINEAR OPTIMIZATION PROBLEMS: From quantization to materials design

MULTILEVEL AND ADAPTIVE METHODS FOR NONLINEAR OPTIMIZATION PROBLEMS: From quantization to materials design
Автор
 
Год
 
Страниц
 
116
ISBN
 
3639224361
Категория
 
Новинки академической литературы Америки - 2010

Описание:

This work discusses some novel multilevel and adaptive methods for nonlinear non-convex optimization, focusing on two particular problems that come from the fields of materials science and quantization, with numerous applications, including image/signal compression, mesh generation and optimal placement of resources in the context of centroidal Voronoi tessellations. In particular, several acceleration methods for quantization are developed, including Newton-type and multilevel algorithms, both of which yield significant speedup comparing to traditional methods. Rigorous convergence analysis is provided and uniform convergence of the multilevel scheme with respect to the grid size and the number of grid levels is demonstrated. A new adaptive scheme for automating phase diagram construction in complex multicomponent materials systems is proposed. The new method utilizes the geometric properties of the energy surfaces together with effective sampling techniques to improve on the...

Похожие книги

Higher order components for grid programming making grids more usableHigher order components for grid programming making grids more usable
Автор: Jan Dunnweber,Sergej Gorlac
Год: 2009
Euro-Par 2007 Workshops: Parallel Processing: HPPC 2007, UNICORE Summit 2007, and VHPC 2007, Rennes, France, August 28-31, 2007, Revised Selected Papers ... Computer Science and General Issues)Euro-Par 2007 Workshops: Parallel Processing: HPPC 2007, UNICORE Summit 2007, and VHPC 2007, Rennes, France, August 28-31, 2007, Revised Selected Papers ... Computer Science and General Issues)
Автор: Luc Bougé, Martti Forsell, Jesper Larsson Träff, Achim Streit, Wolfgang Ziegler, Michael Alexander, Stephen Childs
Год: 2008