Seite - 79 - in Differential Geometrical Theory of Statistics
Bild der Seite - 79 -
Text der Seite - 79 -
Entropy2016,18, 386
LetÏbean invariantvolumeelementonG/K inanafïŹne local coordinatesystem { x1,x2,...,xn }
inaneighborhoodofo:
Ï=Ί ·dx1â§ ...â§dxn (128)
WecanwriteXâ=â
i Ïi â
âxi anddeveloptheLiederivativeof thevolumeelementÏ:
LXâÏ=(LXâΊ) .dx1â§ ...â§dxn+â
j Ί.dx1â§Â·Â·Â·â§LXâdxjâ§Â·Â·Â·â§dxn= (
XâΊ+ (
â
j âÏj
âxj )
Ί )
dx1â§ ...â§dxn (129)
Since thevolumeelementÏ is invariantbyG:
LXâÏ=0âXâΊ+ (
â
j âÏj
âxj )
Ί=0âXâlogΊ=ââ
j âÏj
âxj (130)
ByusingAXâYâ=âDYâXâ,wehave:(
D â
âxi (AXâ) )(
â
âxj )
=D â
âxi (
AXâ (
â
âxj ))
âAXâ (
D â
âxi â
âxj )
=âD â
âxi D â
âxj (
â
k Ïk â
âxk )
=ââ
k â2Ïk
âxiâxj â
âxk (131)
ButasD is locallyïŹatandXâ isan inïŹnitesimalafïŹnetransformationwithrespect toD:
D â
âxi (AXâ)=0â â 2Ïk
âxiâxj =0 (132)
TheKoszul formandcanonicalbilinear formaregivenby:
α=â
i âlogΊ
âxi dxi=DlogΊ (133)
Dα=â
i,j â2logΊ
âxiâxj dxidxj=DdlogΊ (134)
LXâα=LXâDlogΊ=DLXâlogΊ=DXâlogΊ=âD (
â
j âÏj
âxj )
=ââ
,j â2Ïj
âxiâxj dxi=0 (135)
Then,LXâα=0âXâ g.
ByusingXâlogΊ=ââ
j âÏj
âxj ,wecanobtain:
α(Xâ)= (DlogΊ)(Xâ) â
LXâα=0 DXâlogΊ=ââ
j âÏj
âxj (136)
ByusingAXâYâ=âDYâXâ,wecandevelop:
AXâ (
â
âxj )
=âD â
âxj Xâ=ââ
i âÏi
âxj â
âxi (137)
As f(X)=AXâ,o andq(X)=Xâo:
Tr(f(X))=Tr(AXâ,o)=ââ
i âÏi
âxi (o)=α(Xâ0)=α0(q(X)) (138)
Ifweuse thatLXâα=0âXâ g, thenweobtain:
(Dα)(Xâ,Yâ)=(DYâα)(Xâ)=â(AYâα)(Xâ)=âAYâ (α(Xâ))+α(AYâXâ)=α(AYâXâ) (139)
79
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