Individual and collective learning in multi-agent Systems
Moduleco : a multi-agent object-designed software model

2.1. Object modeling concepts  and multi-agent Systems

 page précédente sommaire page suivante  

conceptual framework :

Object and encapsulation : data, methods (computation - behavior)

ModelInfo.jpg (84336 octets)

Inheritance (from abstraction towards implementation) and modular architecture

Diagramatic presentation of UML (= Unified Modeling Language) : to make easier the dialog between computer scientist and economist, and to enhance the economic model semantics.

Four (non exclusive) levels for multi-agent systems :

  1. "Cell within array" (or spreadsheet)
  2. Software object (i.e. objects in array list - this model)
  3. Active (software) objects (i.e. threads)
  4. Autonomous (software) objects (the real "agents") > threshold effects and individual deterministic auto-regulation (i.e. viability principle : Aubin [1991]) cf. Varela, Bourgine [1992])

Denis.Phan@enst-bretagne.frAntoine Beugnard@enst-bretagne.fr