no. 12
Probability Theory
A functional Hungarian construction for the sequential empirical process
[Une construction hongroise fonctionnelle pour le processus empirique séquentiel]

Présenté par : Michel Talagrand

Jähnisch, Michael1 ; Nussbaum, Michael2
1 SAP AG, Neurottstrasse 16, 69190 Walldorf, Germany
2 Department of Mathematics, Cornell University, Ithaca, NY 14853, USA
Comptes Rendus. Mathématique, Tome 341 (2005) no. 12, pp. 761-763.

Nous établissons un couplage KMT pour le processus empirique séquentiel et le processus de Kiefer–Müller. Les processus sont indexés par des fonctions f appartenant à une classe de Hölder H, mais le supremum est pris en dehors de la probabilité. Comparé au couplage en norme sup, ceci permet des classes de fonctions H plus larges. Le résultat peut être utilisé pour démontrer l'équivalence asymptotique de certaines expériences statistiques non-paramétriques.

We establish a KMT coupling for the sequential empirical process and the Kiefer–Müller process. The processes are indexed by functions f from a Hölder class H, but the supremum over f∈H is taken outside the probability. Compared to the coupling in sup-norm, this allows for larger functional classes H. The result is useful for proving asymptotic equivalence of certain nonparametric statistical experiments.

Reçu le :
Accepté le :
Publié le :
DOI : 10.1016/j.crma.2005.10.022
Jähnisch, Michael 1 ; Nussbaum, Michael 2

1 SAP AG, Neurottstrasse 16, 69190 Walldorf, Germany
2 Department of Mathematics, Cornell University, Ithaca, NY 14853, USA 
@article{CRMATH_2005__341_12_761_0,
     author = {J\"ahnisch, Michael and Nussbaum, Michael},
     title = {A functional {Hungarian} construction for the sequential empirical process},
     journal = {Comptes Rendus. Math\'ematique},
     pages = {761--763},
     publisher = {Elsevier},
     volume = {341},
     number = {12},
     year = {2005},
     doi = {10.1016/j.crma.2005.10.022},
     language = {en},
     url = {http://archive.numdam.org/articles/10.1016/j.crma.2005.10.022/}
}
TY  - JOUR
AU  - Jähnisch, Michael
AU  - Nussbaum, Michael
TI  - A functional Hungarian construction for the sequential empirical process
JO  - Comptes Rendus. Mathématique
PY  - 2005
SP  - 761
EP  - 763
VL  - 341
IS  - 12
PB  - Elsevier
UR  - http://archive.numdam.org/articles/10.1016/j.crma.2005.10.022/
DO  - 10.1016/j.crma.2005.10.022
LA  - en
ID  - CRMATH_2005__341_12_761_0
ER  - 
%0 Journal Article
%A Jähnisch, Michael
%A Nussbaum, Michael
%T A functional Hungarian construction for the sequential empirical process
%J Comptes Rendus. Mathématique
%D 2005
%P 761-763
%V 341
%N 12
%I Elsevier
%U http://archive.numdam.org/articles/10.1016/j.crma.2005.10.022/
%R 10.1016/j.crma.2005.10.022
%G en
%F CRMATH_2005__341_12_761_0
Jähnisch, Michael; Nussbaum, Michael. A functional Hungarian construction for the sequential empirical process. Comptes Rendus. Mathématique, Tome 341 (2005) no. 12, pp. 761-763. doi : 10.1016/j.crma.2005.10.022. http://archive.numdam.org/articles/10.1016/j.crma.2005.10.022/

[1] Bretagnolle, J.; Massart, P. Hungarian constructions from the non-asymptotic viewpoint, Ann. Probab., Volume 17 (1989), pp. 239-256

[2] Castelle, N.; Laurent-Bonvalot, F. Strong approximation of bivariate uniform empirical processes, Ann. Inst. H. Poincaré Probab. Statist., Volume 34 (1998), pp. 425-480

[3] Dudley, R. Real Analysis and Probability, Wadsworth Brooks/Cole, Pacific Grove, CA, 1989

[4] Grama, I.; Nussbaum, M. A functional Hungarian construction for sums of independent random variables, Ann. Inst. H. Poincaré Probab. Statist., Volume 38 (2002), pp. 923-957

[5] Grama, I.; Nussbaum, M. Asymptotic equivalence for nonparametric generalized linear models, Probab. Theory Related Fields, Volume 111 (1998), pp. 167-214

[6] Grama, I.; Nussbaum, M. Asymptotic equivalence for nonparametric regression, Math. Methods Statist., Volume 11 (2002) no. 1, pp. 1-36

[7] M. Jähnisch, Asymptotische Äquivalenz für ein Modell unabhängiger nicht identisch verteilter Daten, Doctoral Thesis, Humboldt Universität zu Berlin, 1999

[8] Jähnisch, M.; Nussbaum, M. Asymptotic equivalence for a model of independent non identically distributed observations, Statistics and Decisions, Volume 21 (2003) no. 3, pp. 197-218

[9] Koltchinskii, V. Komlós–Major–Tusnády approximation for the general empirical process and Haar expansions of classes of functions, J. Theoret. Probab., Volume 7 (1994) no. 1, pp. 73-118

[10] Komlós, J.; Major, P.; Tusnády, G. An approximation of partial sums of independent r.v.'s and the sample df. I, Z. Wahr. Verw. Gebiete, Volume 32 (1975), pp. 111-131

[11] Nussbaum, M. Asymptotic equivalence of density estimation and Gaussian white noise, Ann. Statist., Volume 24 (1996), pp. 2399-2430

[12] Rio, E. Local invariance principles and their application to density estimation, Probab. Theory Related Fields, Volume 98 (1994), pp. 21-45

[13] Shorack, G.; Wellner, J. Empirical Processes with Applications to Statistics, Wiley, New York, 1986

Cité par Sources :