Seite - 350 - in Differential Geometrical Theory of Statistics
Bild der Seite - 350 -
Text der Seite - 350 -
Entropy2016,18, 396
whereA stands for theconjugateofA. Thepoint iI inHn is identifiedwith thenullmatrixnoted0 in
Dn. LetZ∈Dn. ThereexistsPadiagonalmatrixwithdecreasingpositiverealentriesandUaunitary
matrixsuchthatZ=UPUt. Letτ1≥ ...≥τnbesuchthat
P= ⎛
⎜⎝ th(τ1)
...
th(τn) ⎞
⎟⎠ .
Let
A0= ⎛
⎜⎝ ch(τ1)
...
ch(τn) ⎞
⎟⎠ ,B0= ⎛
⎜⎝ sh(τ1)
...
sh(τn) ⎞
⎟⎠
and
gZ= (
U 0
0 U )
. (
A0 B0
A0 B0 )
.
It canbeshownthat
gZ∈Sp(n,C)andgZ.0=Z. (3)
WeprovidenowacorrespondencebetweentheelementsofDn andtheelementsofpdefinedin
Equation(1). Let
HZ= ⎛
⎜⎜⎜⎜⎜⎜⎜⎜⎜⎝ τ1
...
τn
−τ1
...
−τn ⎞
⎟⎟⎟⎟⎟⎟⎟⎟⎟⎠ ∈a, (4)
and
aZ= ⎛
⎜⎜⎜⎜⎜⎜⎜⎜⎜⎝ eτ1
...
eτn
e−τ1
...
e−τn ⎞
⎟⎟⎟⎟⎟⎟⎟⎟⎟⎠ ∈A= exp(a).
It canbeshownthat thereexistsk∈K suchthat
Cexp(Adk(HZ))C−1.0=Z,
orequivalently
CkaZkC−1.0=Z.
Recall thatEquation(2)givesAdk(H)∈pandkak∈ exp(p). ThedistancebetweenZand0 inDn
isgivenby
d(0,Z)= (
2∑τ2i )1/2
(5)
(seep. 292 in [20] ).
350
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