BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:Data Analytic UQ Cascade in Optimization
DTSTART:20180717T161500
DTSTAMP:20260429T091028Z
UID:87c5a7415c33390ce7bc2c9f0c291a3128a24998a5b9549123165db5
CATEGORIES:Conferences - Seminars
DESCRIPTION:Prof. Bijan MOHAMMADI (Univ. de Montpellier)\nAbstract:   We 
 present an original framework for uncertainty quantification (UQ) in optim
 ization. It is based on a cascade of ingredients with growing computatio
 nal complexity for both forward and reverse uncertainty propagation. The a
 pproach is merely geometric. It starts with a complexity-based splitting o
 f the independent variables and the definition of a parametric optimizatio
 n problem. Geometric characterization of global sensitivity spaces through
  their dimensions and relative positions by the principal angles between g
 lobal search subspaces bring a first set of information on the impact of u
 ncertainties on the functioning parameters on the optimal solution. Joinin
 g the multi-point descent direction and the quantiles on the optimization 
 parameters permits to define the notion of Directional Extreme Scenarios (
 DES) without sampling of large dimension design spaces. One goes beyond DE
 S with Ensemble Kalman Filters (EnKF) after the multi-point optimization a
 lgorithm is cast into an ensemble simulation environment. This formulation
  accounts for the variability in large dimension. The UQ cascade ends w
 ith the joint application of the EnKF and DES leading to the concept of En
 semble Directional Extreme Scenarios (EDES) which provides more exhaustive
  possible extreme scenarios knowing the Probability Density Function of ou
 r optimization parameters. The different ingredients are illustrated on d
 ifferent industrial applications with particular emphasis on aircraft shap
 e design in the presence of operational and geometrical uncertainties.
LOCATION:MA A3 31 https://plan.epfl.ch/?room=MAA331
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR