Demi-journée « Distances for stochastic processes and applications »

Cette demi-journée aura lieu le 8 octobre 2021 à l’UFR des sciences de Versailles.

Les exposés auront lieu de 14h00 à 18h00, au bâtiment Fermat, dans l’amphi J.

Organisateurs : Ester Mariucci et Emmanuel Rio

Comment venir au laboratoire ?

 

Programme

  • 14h00-14h45 : Vlad Bally (Université de Marne-la-Vallée) : Small jumps versus Brownian motion: error in total variation

Abstract: We consider a stochastic equation with jumps and we replace the small jumps by a Brownian motion. This is a usual operation in stochastic numerics but not only. An easy exercise (Taylor development up to order three for the Infinitesimal operator) allows to estimate the error if one takes three time differentiable test functions. But the problem becomes much more challenging if the test functions are just measurable and bounded (the total variation distance). Our job is to obtain such estimates by using some Malliavin type techniques for jump processes.

 

  • 14h50-15h35 : Arnaud Gloter (Université d’Évry Val d’Essonne) : Approximation of law for S.D.E. driven by a Lévy process and applications to statistical problems   (Abstract)

 

  • 15h40-16h10 : Pause café
  • 16h10-16h55 : Mark Podolskij (Université du Luxembourg) : Semiparametric estimation of McKean-Vlasov SDEs

Abstract: In this paper we study the problem of semiparametric estimation for a class of McKean-Vlasov stochastic differential equations. Our aim is to estimate the drift coefficient of  a MV-SDE based on observations of the corresponding particle system. We propose a semiparametric estimation procedure and derive the rates of convergence for the resulting estimator.  We further prove that the obtained rates are essentially optimal in the minimax sense. This is a joint work with D. Belomestny and V. Pilipauskaite.

 

  • 17h00-17h45 : Emmanuel Rio (Université de Versailles Saint-Quentin) : Quadratic transportation cost in the conditional central limit theorem

Abstract: In this talk we give new estimates of the quadratic cost in the conditional central limit theorem for strictly stationary dependent sequences of real-valued bounded random variables. In particular we obtain rates of convergence of the order of (1/n) for the quadratic cost in the conditional central limit theorem under power-type decay conditions on the dependence  coefficients. This is a joint work with J. Dedecker and F. Merlevède.

Cet événement est soumis au contrôle du pass sanitaire : les participants, y compris les participants de l’UVSQ seront contrôlés par les organisateurs.

Demi-journée « Distances for stochastic processes and applications »