Web-Books
im Austria-Forum
Austria-Forum
Web-Books
Naturwissenschaften
Physik
Differential Geometrical Theory of Statistics
Seite - 52 -
  • Benutzer
  • Version
    • Vollversion
    • Textversion
  • Sprache
    • Deutsch
    • English - Englisch

Seite - 52 - in Differential Geometrical Theory of Statistics

Bild der Seite - 52 -

Bild der Seite - 52 - in Differential Geometrical Theory of Statistics

Text der Seite - 52 -

Entropy2016,18, 386 • Wehaveestablished that theaffinerepresentationofLiegroupandLiealgebraby Jean-Marie Souriau isequivalent to Jean-LouisKoszul’saffinerepresentationdevelopedin the frameworkof hessiangeometryof convexsharpcones. BothSouriauandKoszulhaveelaboratedequations requestedforLiegroupandLiealgebratoensuretheexistenceofanaffinerepresentation.Wehave comparedbothapproachesofSouriauandKoszul ina table. • Wehave applied the Souriaumodel for exponential families and especially formultivariate Gaussiandensities. • Wehaveapplied theSouriau-KoszulmodelGibbsdensity to compute themaximumentropy density forsymmetricpositivedefinitematrices,usingthe innerproduct〈η,ξ〉=Tr(ηTξ),∀η,ξ∈ Sym(n)givenbyCartan-Killingform.TheGibbsdensity(generalizationofGaussianlawfortheses matricesanddefinedasmaximumentropydensity): pξˆ(ξ)= e −〈Θ−1(ξˆ),ξ〉+Φ(Θ−1(ξˆ)) =ψΩ(Id) · [ det ( αξˆ−1 )] ·e−Tr(αξˆ−1ξ) withα= n+1 2 (5) • For thecaseofmultivariateGaussiandensities,wehaveconsideredGA(n)asub-groupofaffine group, thatwedefinedbya(n+1)× (n+1)embedding inmatrixLiegroupGaf f , andthatacts formultivariateGaussian lawsby: [ Y 1 ] = [ R1/2 m 0 1 ][ X 1 ] = [ R1/2X+m 1 ] , ⎧⎪⎪⎨⎪⎪⎩ (m,R)∈Rn×Sym+(n) M= [ R1/2 m 0 1 ] ∈Gaf f X≈ℵ(0, I)→Y≈ℵ(m,R) (6) • FormultivariateGaussiandensities,aswehave identifiedtheactingsub-groupofaffinegroup M,wehavealsodevelopedthecomputationof theassociatedLiealgebrasηL andηR, adjointand coadjointoperators,andespecially theSouriau“momentmap”ΠR:〈 nL,M−1nRM 〉 = 〈ΠR,nR〉 withM= [ R1/2 m 0 1 ] , nL= ⎡⎣ R−1/2 .R1/2 R−1/2 .m 0 0 ⎤⎦ andηR= ⎡⎣ R−1/2 .R1/2 .m−R−1/2 .R1/2 .m 0 0 ⎤⎦ ⇒ΠR= ⎡⎣ R−1/2 .R1/2+R−1 .mmT R−1 .m 0 0 ⎤⎦ (7) Using Souriau Theorem (geometrization ofNoether theorem), weuse the property that this momentmapΠR is constant (its componentsareequal toNoether invariants): dΠR dt =0⇒ ⎧⎨⎩ R −1 .R+R−1 .mmT=B= cste R−1 .m= b= cste (8) to reduce theEuler-LagrangeequationofgeodesicsbetweentwomultivariateGaussiandensities:⎧⎨⎩ .. R+ . m . mT− .RR−1 .R=0 .. m− .RR−1 .m=0 (9) to this reducedequationofgeodesics: 52
zurück zum  Buch Differential Geometrical Theory of Statistics"
Differential Geometrical Theory of Statistics
Titel
Differential Geometrical Theory of Statistics
Autoren
Frédéric Barbaresco
Frank Nielsen
Herausgeber
MDPI
Ort
Basel
Datum
2017
Sprache
englisch
Lizenz
CC BY-NC-ND 4.0
ISBN
978-3-03842-425-3
Abmessungen
17.0 x 24.4 cm
Seiten
476
Schlagwörter
Entropy, Coding Theory, Maximum entropy, Information geometry, Computational Information Geometry, Hessian Geometry, Divergence Geometry, Information topology, Cohomology, Shape Space, Statistical physics, Thermodynamics
Kategorien
Naturwissenschaften Physik
Web-Books
Bibliothek
Datenschutz
Impressum
Austria-Forum
Austria-Forum
Web-Books
Differential Geometrical Theory of Statistics