Department of Mathematics, Computer Science, Control of the ISAE

Statistical Modeling and Application to Optimization

last modified 2 October 2009

All the versions of this article: [français] [English]


Industrial, governmental and academic partnerships

  • ONERA,
  • CERFACS (Toulouse),
  • IMT (Toulouse),
  • AIRBUS (Toulouse),
  • CNES.

Permanent staff: N. Bartoli (ONERA) Y. Caumel (ISAE), P. Klotz (ONERA), M. Samuelides (ISAE)

PhD students: A. Merval

Contact:, phone: 33 (0)5 61 33 81 06

Statistical surrogate models are used instead of linear predictions to approximate numerical functions and avoid complex computations. These methods are currently being developed and are applied to a large set of physical problems, Including structural computation, flow computation and information transmission via networks. Using statistical methods for non-linear regression and experiment design is natural as long as uncertainty has to be taken into account in the model. More recently, these methods have been successfully used in deterministic problems where they can replace interpolation methods advantageously, for example :

  • they can use irregular sampling, which is crucial in large dimension problems,
  • they can make up for modeling errors,
  • they can take results into account from zones with unsuitable computation.

We have used these methods within the framework of joint ONERA research projects (BUFET, HEPHAISTOS, STOK) and within the framework of the VIVACE European project.

Furthermore, within the framework of transdisciplinary research in DMIA on communication networks, in cooperation with CNES, the French space agency, THALES and NICTA (Australia), we are studying and developing new network architecture to meet users’ needs for mobile and wireless technology using waiting queues and networks.

Mentions légales | Plan d’accès | Site Map | Follow-up of the site's activity RSS 2.0 |