Logarithmic Sobolev inequality for diffusion semigroups

Ivan Gentil

Description

Logarithmic Sobolev inequality for diffusion semigroups Ivan Gentil June 5, 2009 Abstract Course given in Grenoble at the 2009 summer school : Optimal transportation, Theory and applications 1 Introduction The goal of this course is to introduce inequalities as Poincare or logarithmic Sobolev for diffusion semigroups. We will focus more on examples than on the general theory of diffusion semigroups. A main tool to obtain those inequalities is the so called the Bakry-Emery ?2-criterium. This criterium is well known to prove such inequalities and as been used many times for other problems (see for example [BE85, ABC+00, Bak06]). In Section 4, we will explain an alternative method to get a logarithmic Sobolev in- equality under the ?2-crierium. It is called the Mass transportation method and has been introduced recently (see [CE02, OV00, CENV04, Vil09]). By this way we will also get a another inequality called the Talagrand inequality. 2 The Ornstein-Uhlenbeck semigroup and the Gaussian measure A Markov semigroup on Rn (for n > 0) is associated to a Markov processes, there are two famous example of diffusion semigroups. The first one is the heat semigroup which is associated to the Brownian motion on Rn. In this course we will study the second one which is the Ornstein-Uhlenbeck semigroup. As we will see in the next section, the Ornstein- Uhlenbeck semigroup is associated to a linear stochastic differential equation driven by a Brownian motion.

• commutation property


• gaussian distribution


• curvature criterium


• sobolev inequality


• interpolation between


• ?f ·


• smooth functions


• ?the logarithmic


• gaussian measure

Sujets

• Grenoble
• Gentil
• Normal distribution
• Sobolev inequality
• Smooth function
• Gaussian measure

Informations

• Publié par: pefav
• Nombre de lectures: 29
• Langue: English

Extrait

LogarithmicSobolevinequalityfordiffusionsemigroups IvanGentil June5,2009 Abstract CoursegiveninGrenobleatthe2009summerschool:Optimaltransportation,Theory andapplications 1Introduction ThegoalofthiscourseistointroduceinequalitiesasPoincare´orlogarithmicSobolevfor diffusionsemigroups.Wewillfocusmoreonexamplesthanonthegeneraltheoryofdiffusion semigroups. AmaintooltoobtainthoseinequalitiesisthesocalledtheBakry-EmeryΓ 2 -criterium. Thiscriteriumiswellknowntoprovesuchinequalitiesandasbeenusedmanytimesfor otherproblems(seeforexample[BE ´ 85,ABC + 00,Bak06]). InSection4,wewillexplainanalternativemethodtogetalogarithmicSobolevin- equalityundertheΓ 2 -crierium.ItiscalledtheMasstransportationmethodandhasbeen introducedrecently(see[CE02,OV00,CENV04,Vil09]).Bythiswaywewillalsogeta anotherinequalitycalledtheTalagrandinequality. 2TheOrnstein-UhlenbecksemigroupandtheGaussian measure AMarkovsemigroupon R n (forn>0)isassociatedtoaMarkovprocesses,thereare twofamousexampleofdiffusionsemigroups.Thefirstoneistheheatsemigroupwhich isassociatedtotheBrownianmotionon R n .Inthiscoursewewillstudythesecondone whichistheOrnstein-Uhlenbecksemigroup.Aswewillseeinthenextsection,theOrnstein- Uhlenbecksemigroupisassociatedtoalinearstochasticdifferentialequationdrivenbya Brownianmotion. Inthisnoteasmoothfunctionfin R n isafunctionsuchthatallcomputationdoneas integrationbypartsarejustify. 2.1Definitionandgeneralproperties Definition2.1Letdefinethefamilyofoperator(P t ) t > 0 :iff∈C b ( R n )then P t f(x)=f(e −t x+1−e −2t y)dγ(y),(1) Z p where | y | 2 2 − 2 e dγ(y)= n/2 dy )nπ2( isthestandardGaussiandistributionin R n and|
|istheEuclideannormon R n . Thefamilyofoperator(P t ) t > 0 iscalledtheOrnstein-Uhlenbecksemigroup.