Workshop on Stochastic Simulators

When & where

The workshop will take place online place on March 11, 2021, using Microsoft Teams.

Registered participants will receive a link by email on the day before the event.

Technicalities: participants can join using either their web browser (no account or installation required) or the "fat" Teams client (Teams account + installation required).

Presentation of the workshop

This workshop aims at presenting recent advances in the analysis of parameter-dependent stochastic numerical models (or stochastic simulators). Some usual questions of uncertainty quantification will be addressed for such models, such as sensitivity analysis, the construction of surrogate models or the solution of inverse problems. Optimization problems with stochastic simulators or noisy functions, which are related to the field of bandit optimization, will also be covered.

Organizers: Julien Bect, Anthony Nouy and Clémentine Prieur.


Registration is free but appreciated.



Long talks (40 minutes):

Mickaël Binois (Inria Sophia Antipolis - Méditerranée)
Multi-objective optimization of an agent-based simulator
Abstract / Slides

Aurélien Garivier (ENS Lyon)
The complexity of optimizing noisy functions on graphs
Abstract / Slides

Bruno Sudret (ETH Zürich)
Surrogate models for stochastic simulators: an overview and a focus on generalized lambda models
Abstract / Slides

Short talks (25 minutes):

Athénaïs Gautier (Univ. Berne)
Spatial logistic Gaussian process for density field modelling: an application to stochastic inverse problems
Abstract / Slides

Henri Mermoz Kouye (INRAE)
Global sensitivity analysis for models described by continuous time Markov chains with application to epidemic models
Abstract / Slides

Xujia Zhu (ETH Zürich)
Stochastic polynomial chaos expansions for emulating stochastic simulators

Bruno Barracosa (EDF R&D + L2S)
Bayesian methods for multi-objective simulation-based optimization
Abstract / Slides

Nora Lüthen (ETH Zürich)
Surrogating stochastic simulators using spectral methods and advanced statistical modeling