Exposé
GDT "EDP et Calcul Scientifique" du mardi 31 mars 2020
(annulé - COVID19)
Atelier des doctorants du mardi 11 février
On kernel estimation for spatial data.
In this talk, we present a central limit theorem for the well-known Nadaraya-Watson regression estimator in the context of strongly mixing and weakly dependent random fields in the sense of Rosenblatt (1956) and Wu (2005) respectively. Our main motivation is to provide mild conditions on the mixing coefficients and bandwidth parameters for the estimator to be asymptotically normal. We also present our current research concerning the recursive version of this estimator under the same conditions.
GDT "EDP et Calcul Scientifique" du mardi 7 avril 2020
(annulé - COVID19)
GdTProbaTESD20200217
Rigidité et disjonction de Möbius de systèmes dynamiques
La conjecture de Sarnak dit que tout système dynamique déterministe $(X,T)$ est disjoint (au sens arithmétique) de la fonction de Möbius $\mu$: $$\lim_{N\to\infty}\frac1N\sum_{n\leq N}f(T^nx)\mu(n)=0$$ pour toute fonction continue $f$ et tout $x\in X$. Les systèmes rigides sont déterministes, mais la rigidité peut être définie soit de façon topologique, soit métrique en utilisant les systèmes dynamiques métriques $(X,\nu,T)$ où $\nu$ parcurt l'ensemble des mesures $T$-invariantes.
Non-asymptotic sharp oracle inequalities for high dimensional ergodic diffusion models
In this talk we study high dimensional ergodic diffusion models in nonparametric setting on the basis of discrete data, when the diffusion coefficients are unknown. For this problem, by using efficient sequential point-wise estimators we construct a model selection procedure and then we show sharp oracle inequalities, i.e. the inequalities in which the main term coefficient is closed to one. This means that the proposed sequential model selection procedure is optimal in this sense.
GDT "EDP et Calcul Scientifique" du mardi 28 avril 2020
(annulé - COVID19)
GDT "EDP et Calcul Scientifique" du mardi 10 mars 2020
Sur l'inégalité isopérimétrique quantitative dans le plan
GdTProbaTESD20200127
Limiting spectrum of sparse graphs
In this introductory talk, I will present a general limiting
theory for the spectrum of large networks. The models I will consider
are quite general, but they share a common feature : all of them are
studied in their very sparse regime where the number of connections has
the same order as the number of nodes (Erdös-Rényi with fixed mean
degree, regular graphs, uniform trees, uniform triangulations,
preferential attachments). The spectrum of such networks is notoriously
Drawdown and speed of market crash on high frequency data
Financial markets are characterized by continuous upward or downward fluctuations in prices, caused by the vast amount of information they receive. A strong price instability historically and cyclically caused strong market collapses that prompted investors to control the risk related to the excessive fluctuation of the prices in order to prevent significant portfolio losses.