logo SBA

ETD

Archivio digitale delle tesi discusse presso l’Università di Pisa

Tesi etd-03312014-192936


Tipo di tesi
Tesi di laurea magistrale
Autore
LADU, ROBERTO
URN
etd-03312014-192936
Titolo
Error Tolerant Descent Methods for Computing Equilibria
Dipartimento
INFORMATICA
Corso di studi
INFORMATICA E NETWORKING
Relatori
relatore Prof. Bigi, Giancarlo
Parole chiave
  • approximation
  • constrained
  • constraints
  • continuous
  • convex
  • descent methods
  • equilibrium
  • KKT
  • nash
  • nonlinear
  • optimality conditions
  • optimization
  • problem
Data inizio appello
24/04/2014
Consultabilità
Completa
Riassunto
Equilibrium problems naturally arise in the modelling of competitive agents systems in many fields of Engineering. Several descent methods have been developed to solve them, usually asking for an exact solution of auxiliary optimization problems at each step. We propose two algorithms that rely only on approximated solutions of these auxiliary problems. Some ideas on possible methods for computing these solutions efficiently controlling the error are also given. Preliminary numerical tests on cloud computing scenarios have been carried out.
File