Pages

Wednesday, June 3, 2020

UQSay #10


The tenth UQSay seminar on Uncertainty Quantification and related topics, organized by L2S, MSSMAT, and EDF R&D, will take place online on Thursday afternoon, June 11, 2020.

14h–15h — Guillaume Perrin (CEA DIF) — [slides]

Point process-based approaches for the reliability analysis of complex systems

The conception of complex systems relies more and more on simulation. Numerical codes have therefore been introduced to optimize performance criteria for the system, but also to insure that undesirable events will not appear. Due to the presence of uncertainties on the inputs of the system, a probabilistic framework to the safety analysis of the system is generally considered. It generally relies on the estimation of the probability that a given real-valued output of the code does not exceed a specified threshold. This estimation may not be easy, as the numerical code can be very time-consuming and this probability may be very small (the system is supposed to be designed to operate in safe conditions).

In this presentation, we therefore propose an adaptive method to bound this probability with great confidence, while using as few code evaluations as possible. This method is based on the coupling of a particular Marked Poisson Process and the Gaussian process regression formalism. The interest of the proposed approach will finally be illustrated on a series of test cases.

Organizers: Julien Bect (L2S), Emmanuel Vazquez (L2S), Didier Clouteau (MSSMAT), Filippo Gatti (MSSMAT), Fernando Lopez Caballero (MSSMAT), Bertrand Iooss (EDF R&D).

Practical details: the seminar will be held online using Microsoft Teams.

If you want to attend this seminar (or any of the forthcoming online UQSay seminars), and if you do not already have access to the UQSay channel on Teams, simply to send an email and you will be invited. You will find the link to the seminar on this channel, approximately 15 minutes before the beginning.

The technical side of things: you can use Teams either directly from you web browser or using the "fat client" application, which is available for most platforms (Windows, Linux, Mac, Android & iOS). We strongly recommend the latter option whenever possible. Please give it a try before the seminar to anticipate potential problems.