Guillaume Chennetier

Postdoctoral researcher in applied probability and statistics (in Paris area, France)


Curriculum vitae



CERMICS

École nationale des ponts et chaussées



Guillaume Chennetier

Postdoctoral researcher in applied probability and statistics (in Paris area, France)


Contact

Guillaume Chennetier

Postdoctoral researcher in applied probability and statistics (in Paris area, France)


Curriculum vitae



CERMICS

École nationale des ponts et chaussées




Welcome!


I am a postdoctoral fellow in applied probability and statistics at CERMICS (research center in applied mathematics of École nationale des ponts et chaussées, Paris area, France). I am recipient of the 2024 Cauchy fellowship and I work mainly with my advisor Julien Reygner.

Before that, I defended my PhD thesis in statistics in September 2024 under the supervision of Josselin Garnier at CMAP (Applied Mathematics Laboratory of École polytechnique at Palaiseau, France). This was an industrial collaboration with the French electricity company EDF, within the Probabilistic Safety Studies team of the R&D Lab Paris-Saclay, where I worked with Hassane Chraibi and Anne Dutfoy. 

About my research

My research generally falls within the field of probabilistic numerical simulation and uncertainty quantification. I am currently exploring the use of kernel methods for inference problems under covariate shift, and I continue to work on rare event simulation methods, building on my PhD research. 
About my PhD thesis
My thesis work focuse on the simulation of rare events for piecewise deterministic Markov processes (PDMPs). From an application perspective, PDMPs are used to model complex dynamic industrial systems where estimating the probability of failure is essential. Given the rarity of these failures and the high cost of simulating industrial systems, my research was dedicated to exploring methods that offer reduced variance compared to the standard Monte Carlo approach. Specifically, I have developed adaptive importance sampling methods informed by the approximation of the committor function of the underlying stochastic process. 

News

Share