Page - 97 - in Differential Geometrical Theory of Statistics
Image of the Page - 97 -
Text of the Page - 97 -
Entropy2016,18, 386
Figure14.GeodesicsShootingbetweentwomultivariateGaussian incasen=2.
9. SouriauRiemannianMetric forMultivariateGaussianDensities
To illustrate the Souriau-Fishermetric, wewill consider the family ofmultivariateGaussian
densitiesandwilldevelopsomeelements thatwehavepreviouslydevelopedpurely theoretically.
For the families ofmultivariateGaussiandensities, thatwehave identified as homogeneous
manifoldwith theassociated sub-groupof theaffinegroup [
R1/2 m
0 1 ]
,wehave seen that ifwe
consider themaselementsofexponential families,wecanwrite ξˆ (elementof thedualLiealgebra)
thatplay the roleofgeometricheatQ inSouriauLiegroupthermodynamics, andβ thegeometric
(Planck) temperature.
ξˆ= [
E [z]
E [ zzT ] ]
= [
m
R+mmT ]
, β= ⎡⎢⎣ −R−1m1
2 R−1 ⎤⎥⎦ (245)
These elements are homeomorphic to the matrix elements in matrix Lie algebra and dual
Liealgebra:
ξˆ= [ R+mmT m
0 0 ]
∈ g∗ , β= ⎡⎢⎣ 12R−1 −R−1m
0 0 ⎤⎥⎦ ∈ g (246)
IfweconsiderM= [
R′1/2 m′
0 1 ]
, thenwecancompute theco-adjointoperator:
Ad∗Mξˆ= [
R+mmT−mm′T R′1/2m
0 0 ]
(247)
Wecanalsocompute theadjointoperator:
AdMβ=M ·β ·M−1= [
R′1/2 m′
0 1 ]⎡⎣ 12R−1 −R−1m
0 0 ⎤⎦[ R′−1/2 −R′−1/2m′
0 1 ]
AdMβ= ⎡⎣ 12R′1/2R−1R′−1/2 −12R′1/2R−1R′−1/2m′−R′1/2R−1m
0 0 ⎤⎦ (248)
97
Differential Geometrical Theory of Statistics
- Title
- Differential Geometrical Theory of Statistics
- Authors
- Frédéric 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