Dimension reduction for large-scale networked systems - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2012

Dimension reduction for large-scale networked systems

Résumé

A methodology is proposed to approximate large-scale networked systems by a lower dimensional networked system. We first group the nodes into m communities which will form the m vertices of the reduced network. We then associate an appropriate scalar dynamics to each community; in that way, the dimension of the new model is equal to m. The main idea is to approximate each node trajectory by the trajectory of its community. The edges are derived by considering some linear combinations of the link strengths between the elements of each community. Finally, the initial conditions are selected to guarantee the asymptotic consistency of the reduced model with the original system. Thus, we prove the asymptotic convergence of any state of the original system to its corresponding community state according to some distance. It has to be emphasized that our approach is flexible as the user is free to select the reduced system dimension m.

Domaines

Automatique
Fichier principal
Vignette du fichier
ModelReduction.pdf (367.52 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)
Loading...

Dates et versions

hal-00733802 , version 1 (19-09-2012)

Identifiants

  • HAL Id : hal-00733802 , version 1

Citer

Irinel-Constantin Morarescu, Romain Postoyan. Dimension reduction for large-scale networked systems. 51st IEEE Conference on Decision and Control, CDC 2012, Dec 2012, Maui, Hawaii, United States. pp.CDROM. ⟨hal-00733802⟩
86 Consultations
124 Téléchargements

Partager

Gmail Facebook X LinkedIn More