Page - 83 - in Differential Geometrical Theory of Statistics
Image of the Page - 83 -
Text of the Page - 83 -
Entropy2016,18, 386
Figure9.GenerationofKoszulelements fromCartan innerproduct.
This informationgeometryhasbeen intensivelystudiedforstructuredmatrices [151ā166]andin
statistics [167]andis linkedto theseminalworkofSiegel [168]onsymmetricboundeddomains.
Wecanapply thisKoszulgeometry frameworkforconesofsymmetricpositivedeļ¬nitematrices.
Let the innerproduct ćĪ·,ξć=Tr(Ī·Tξ) ,āĪ·,ξāSym(n)givenbyCartan-Killingform,Ī©bethesetof
symmetricpositivedeļ¬nitematrices isanopenconvexconeandisself-dualĪ©ā=Ī©.
ćĪ·,ξć=Tr(Ī·Tξ) ,āĪ·,ξāSym(n)
ĻĪ©(β)=
Ī©ā eāćβ,ξćdξ=det(β)ān+12 ĻĪ©(Id)
ξĖ= āΦ(β)
āβ = ā(ālogĻĪ©(β))
āβ = n+1
2 βā1 (158)
pξĖ(ξ)= e āćĪā1(ξĖ),ξć+Φ(Īā1(ξĖ)) =ĻĪ©(Id) Ā· [
det (
αξĖā1 )] Ā·eāTr(αξĖā1ξ)
withα= n+1
2 (159)
Wewill in the following illustrate informationgeometry formultivariateGaussiandensity [169]:
pξĖ(ξ)= 1
(2Ļ)n/2det(R)1/2 eā 1
2(zām)TRā1(zām) (160)
Ifwedevelop:
1
2 (zām)TRā1(zām) = 1
2 [ zTRā1zāmTRā1zāzTRā1m+mTRā1m]
= 1
2 zTRā1zāmTRā1z+ 12mTRā1m (161)
Wecanwrite thedensityasaGibbsdensity:
pξĖ(ξ)= 1
(2Ļ)n/2det(R)1/2e 1
2m TRā1m eā[āmTRā1z+12zTRā1z] = 1
Z eāćξ,βć
ξ= [
z
zzT ]
andβ= ā”⣠āRā1m1
2 Rā1 ā¤ā¦=[ a
H ]
with ćξ,βć= aTz+zTHz=Tr[zaT+HTzzT] (162)
83
Differential Geometrical Theory of Statistics
- Title
- Differential Geometrical Theory of Statistics
- Authors
- FrƩdƩric Barbaresco
- Frank Nielsen
- Editor
- MDPI
- Location
- Basel
- Date
- 2017
- Language
- English
- License
- CC BY-NC-ND 4.0
- ISBN
- 978-3-03842-425-3
- Size
- 17.0 x 24.4 cm
- Pages
- 476
- Keywords
- Entropy, Coding Theory, Maximum entropy, Information geometry, Computational Information Geometry, Hessian Geometry, Divergence Geometry, Information topology, Cohomology, Shape Space, Statistical physics, Thermodynamics
- Categories
- Naturwissenschaften Physik