Context

This work is part of a collaboration between computer scientists from I3S, University of Nice Sophia Antipolis and economists from SKEMA Business School. It entails the development of a forward-looking cognitive agent model in order to study the spontaneous emergence of money, with potential applications to cryptocurrencies. The main educational objecting of the internship is the implementation of the cutting edge artificial intelligence techniques for modeling decision-making of economic agents in complex environments.

Method
Agent based simulation is a tool for studying complex systems. Its basic element is called an agent: a discrete entity with a certain behaviour and its own objectives. Agents can be different (in terms of behaviour, for example) from each other and each agent can have the ability to be autonomous and therefore be able to adapt and change its behaviour.

Some examples of using agent-based simulation are in the following frameworks: interactions between people in a city, evacuation of a building, car traffic, urban segregation, etc. (see for example http://www.en.jakubas.net.pl/projects/multiagent-simulator).

In particular, agent-based modeling is well-suited to studying economic phenomena like the emergency of money. One of the original features of the model this project aims at building is that it will be based on BDI (Beliefs, Desires, Intentions) cognitive agents. These are agents endowed with reasoning capabilities, beliefs, desires and intentions as opposed to simple reactive agents which are usually employed in economic agent-based simulations.

Existing tools
A Java library, KOBDIG, providing a basic implementation of this kind of cognitive agents is already available (da Costa Pereira et Tettamanzi 2010).

Some Java code has been develpped by the economists to simulate the emergence of money in a simple setting using reactive agents.

Work required

During the internship period the student shall:

  • Get started with the KOBDIG library,

  • Familiarizing themselves with the existing Java code,

  • Integrate cognitive agents into the simulator using the KOBDIG library,

  • Improve the agent-based model as instructed by the supervisors,

  • Test the model and perform experiments,

  • Participate in the writing of a technical report.
References 
  1. C. da Costa Pereira,  A. Tettamanzi. Goal Relationships in Possibilistic Goal Generation. ECAI 2010: 641-646.
  2. C. da Costa Pereira, A. Mauri, A. Tettamanzi. Cognitive-Agent-Based Modeling of a Financial Market. IAT 2009: 20-27
  3. Iacopetta, M. (2018) The Emergence of Money: A Dynamic Analysis. Macroeconomic Dy- namics, forthcoming.

  4. Kiyotaki, N., and R. Wright (1989) On Money as a Medium of Exchange, Journal of Political Economy 97, 927–954.

 

 


Mots-clés
Interdisciplinarité; java; Multi-agent Systems; simulation multi-agents
Établissement
LABORATOIRE D'INFORMATIQUE SIGNAUX ET SYSTEMES DE SOPHIA-ANTIPOLIS (I3S)
6903 SOPHIA ANTIPOLIS  
Date de début souhaitée
01/03/2019
Langues obligatoires
Anglais; Français
Niveau
Bac +5
Prérequis

Good programming; knowledge of the java language

Durée
6 months
Indemnité
3,75 € per hour X 7 hours/day X number of working days
Informations de contact

C. da Costa Pereira, I3S (celia.pereira@unice.fr); M. Iacopeta, SKEMA Business School (maurizio.iacopetta@skema.edu)