Conference on non-stationarity

04.06.2018 - 06.06.2018


Time series methods are traditionally used for modeling data observed over time. These methods have been successfully applied to various scientific disciplines, including finance, environmetrics, risk management, medicine and others. A crucial assumption underlying most time series methods is that of stationarity. Stationarity implies that the long-term behavior of the process is stable and enables deriving ergodic theorems which are important tools for studying properties of estimators such as their asymptotic behavior.

The aim of this conference is to bring together scientists to discuss models and methods that are not based on the assumption of stationarity. Towards this goal, the conference will not be limited to local stationary models (in the sense of Dahlhaus). Other important topics will be covered such as nonstationary periodic models, inclusion of covariates and regression modeling based on shape constraints. In addition, we will bring together scientists from applied fields to show the wide applicability of the new methodology. The following speakers, who have already agreed to give a talk, have eminent contributions to this rapidly developing field : Jean Marc Bardet, Arnaud Belletoile, Rainer Dahlhaus, Didier Dacunha-Castelle, Herold Dehling, Anna Dudek, Gilles Durrieu, Konstantinos Fokianos, Pierre Gaillard, Liudas Giraitis, Rainer Höpfner, Emilie Lebarbier, Nikolai Leonenko, Jacek Leskow, Kerlyns Martınez, Guy Mélard, Florence Merlevède, Michael Neumann, Nathalie Picard, Stefan Richter, Mathieu Rosenbaum, François Roueff, Suhasini Suba Rao, Lionel Truquet, Oliver Wintenberger, Wei Biao Wu.

 Participants are welcome to submit a poster: paul.doukhan @

Selected posters will also lead to constructive discussions.


This workshop is sponsored by the LABEX MME-DII, the Institute for Advances Studies, the laboratory AGM, the ANR Elitisme, in University of Cergy-Pontoise, and the laboratory SAMM Paris 1 Sorbonne, CDiscount Bordeaux and the Mathamsud SaSMoTiDep project.




As this is clear from the current research activities, a wide part of the statistical activity is related to this central question :

"extending models for non stationary and real life data"

This means that non only local stationarity or monotonic tendencies should be developed, but certainly periods and exogenous data and much more questions need to be taken care of... Both the theoretical aspects and applied issues will be considered. Many theoretical and computational developments of this important question are now on progress, check e.g. ,


 The precise schedule of the conference is attached below :



The following interface allows a reader to online registration, participants are offered the possibility to get a lunch buffet (details to be announced and registration is mandatory)

Days of presence (Multiple choice) 
Participation to the lunch buffet (details to be announced) 
Partipation to the conference dinner (details to be announced) - June 5
Registration will close May 20 or earlier if they a no more available seats in the conference rooms

In case of any question please contact paul.doukhan @ or (jennifer.denis @

Applications will be related to various questions: 

  1. Corals reefs vivacity and its relation to the main question of non stationarity, the global warming problem are a main question; the extreme biological diversity of corals makes useful to develop high dimensional non-stationary time series models
  2. Astronomical data sets are often supposed to fit a long range dependent model, but definitely alternative nonstationary models should be checked too.
  3. e-commerce admit analogue high dimensional needs that corals in point 1). Prediction of such high dimensional data models need to be performed and periodic behaviours are patent here.
  4. Economy, financial or actuarial data have analogously periodic behaviours. In the field of economy one may also think of electricity consumption, also subject to pics and trends.
  5. For biology or genomic nonstationary behaviours are also relevant, eg for modelling DNA chains.
  6. In the area of mechanic, cyclo-stationarity considered natural for models of gearing also include periodic data sets.



Suggested hotels: : 8 rue Pierre de Coubertin 95300 Pontoise, tel. +33 1 30 38 55 44 : 13, rue de Neuville 95000 CERGY- (Ham), tel +33 1 30 30 09 17 : 8 Rue Caillaux 75013 Paris  Tel:(+33) 1 44 24 22 00 :  1 Rue Maryse Bastie, 75013 Paris, France  tel : +33 1 42174620

Paul Doukhan
paul.doukhan @

The organization committee is :

Paul Doukhan paul.doukhan @

Eva Loecherbach eva.loecherbach @

Jean-Luc Prigent jean-luc.prigent @ prigent @

Nathalie Picard nathalie.picard @

Joseph Rynkiewicz

Jennifer Denis jennifer.denis @

 Flora koukiou (floria.koukiou @

The conference will be at IEA on June 4-5 «Site_MIR.pdf» and in the site les chênes (June 6) salle de conférences «Site_Chenes.pdf»

Partager :