a bilateral version of the shannon-mcmillan-breiman theoremnew

a bilateral version of the shannon-mcmillan-breiman theoremnew

ID:34458410

大小:121.32 KB

页数:7页

时间:2019-03-06

a bilateral version of the shannon-mcmillan-breiman theoremnew_第1页
a bilateral version of the shannon-mcmillan-breiman theoremnew_第2页
a bilateral version of the shannon-mcmillan-breiman theoremnew_第3页
a bilateral version of the shannon-mcmillan-breiman theoremnew_第4页
a bilateral version of the shannon-mcmillan-breiman theoremnew_第5页
资源描述:

《a bilateral version of the shannon-mcmillan-breiman theoremnew》由会员上传分享,免费在线阅读,更多相关内容在教育资源-天天文库

1、ABilateralVersionoftheShannon-McMillan-BreimanTheoremPierreTisseurLaboratoireG´enomeetInformatiqueUniversit´ed’Evry,TourEvry2.December5,2003AbstractWegiveanewversionoftheShannon-McMillan-Breimantheoreminthecaseofabijectiveaction.ForafinitepartitionαofacompactsetXandameasurableaction

2、TonX,wedenotebyCn,m,αT(x)theelementofthepartitionα∨T1α∨...∨Tmα∨T−1α∨...∨T−nαwhichcontainsapointx.Weprovethatforµ-almostallx,−1Tlimlogµ(Cn,m,α(x))=hµ(T,α),n+m→∞n+mwhereµisaT-ergodicprobabilitymeasureandhµ(T,α)isthemetricentropyofTwithrespecttothepartitionα.1IntroductionTheShannon-

3、McMillan-Breimantheorem[2],[3]isusedinmanyproblemsrelatedtothemetricentropymapofanergodicmeasure.Weextendthiswell-knownresulttothecaseofabijectivedynamicalsystem.OurprooffollowsthelineofPetersen’sproof[3].Weillustratethisnewresultwithanexamplethatgivesaninequalitybetweenshiftsandce

4、llularautomataentropiesandsomeanalogoftheLyapunovexponents.OurbilateralversionoftheShannon-McMillan-Breimantheoremisexpectedtobeusefulinotherareasofdynamicalsystems.12BackgroundmaterialLetXbeacompactspace,µaprobabilitymeasureonXandTameasurablemapfromXtoX.Wedenotebyαafinitepartitiono

5、fXandbyCT(x)n,αtheelementofthepartitionα∨T−1α∨...∨T−nαwhichcontainsthepointx.ForallpointxtheinformationmapIisdefinedbyXTI(α)(x)=−logµ(Cα(x))=−logµ(A)χA(x),A∈αwhereCT(x)istheelementofαwhichcontainsxandχisthecharacteristicαAfunctiondefinedby1ifx∈A,χA(x)=0otherwise.Theinformationmapsat

6、isfiesI(α∨β)=I(α)+I(β

7、α),(1)−1I(Tα

8、Tβ)=I(α

9、β)◦T.(2)ThesetwopropertiesareeasilyprovedfromthedefinitionofIandthefactthatTisasurjectivemap.Wereferto[3,p.238],[4,Chap.8]foradetailedproofof(1)and(2).AsimpleformulationofthemetricentropywithrespecttothepartitionαisgivenbyZnkhµ(T,α)=limI(α

10、∨

11、k=1Tα)(x)dµ(x),n→∞XwhereXµ(A∩B)I(α

12、β)(x)=−χA∩B(x)logµ(B)A∈α,B∈βistheconditionalinformationmaprepresentingthequantityofinformationgivenbythepartitionαknowingthepartitionβaboutthepointx.WerecalltheShannon-McMillan-Breimantheorem[2][3].Theorem1(Shannon-McMillan-Breiman’stheorem)Ifµi

13、saT-ergodicmeasure,thenfor

当前文档最多预览五页,下载文档查看全文

此文档下载收益归作者所有

当前文档最多预览五页,下载文档查看全文
温馨提示:
1. 部分包含数学公式或PPT动画的文件,查看预览时可能会显示错乱或异常,文件下载后无此问题,请放心下载。
2. 本文档由用户上传,版权归属用户,天天文库负责整理代发布。如果您对本文档版权有争议请及时联系客服。
3. 下载前请仔细阅读文档内容,确认文档内容符合您的需求后进行下载,若出现内容与标题不符可向本站投诉处理。
4. 下载文档时可能由于网络波动等原因无法下载或下载错误,付费完成后未能成功下载的用户请联系客服处理。