Seite - 89 - in Differential Geometrical Theory of Statistics
Bild der Seite - 89 -
Text der Seite - 89 -
Entropy2016,18, 386
theautomorphisms, theactionbyconjugationof theLiegrouponitself thatallowsthisoperator to
carryamemberof thegroup.
AD :G×G→G
M,N →ADMN=M.N.M−1 (192)
⎧⎪⎪⎪⎪⎪⎨⎪⎪⎪⎪⎪⎩ M1= [
R1/21 m1
0 1 ]
, M2= [
R1/22 m2
0 1 ]
ADM1M2= [
R1/22 −R1/22 m1+R1/21 m2+m1
0 1 ] (193)
IfnowweconsideracurveN(t)curveonthemanifoldvia the identityat t=0. Its imagebythe
previousoperatorwill be thencurveγ=M ·N(t) ·M−1 passing through identityelementat t=0.
As .
N(0) isanelementof theLiealgebraandits imagebypreviousconjugationoperator iscalledthe
Adjointoperator:
Ad :G×g→ g
M,n →AdMn=M.n.M−1= ddt ∣∣∣
t=0 (ADMN(t))with {
N(0)= I
.
N(0)=n∈ g (194)
Wecanthencompute theAdjointoperator for thepreviousLiegroup:
⎧⎪⎪⎪⎪⎪⎪⎪⎨⎪⎪⎪⎪⎪⎪⎪⎩ n2L= ⎡⎣ R−1/22 .R1/22 R−1/22 .m2
0 0 ⎤⎦ , n2R= ⎡⎣ R−1/22 .R1/22 −R−1/22 .R1/22 m2+ .m2
0 0 ⎤⎦
AdM1n2L=n2R andAdM2n2R= ⎡⎣ R−1/22 .R1/22 −R−1/22 .R1/22 m2+ .R1/22 m2+R1/22 .m2
0 0 ⎤⎦ , AdM−11 n2R=n2L (195)
Werecall that theLiealgebrahasbeendefinedas the tangentspaceat the identityofaLiegroup.
Wewill then introduceaLiebracket [., .], theexpressionof theoperatorassociatedwith thecombined
actionof theLiealgebraon itself, calledanadjointoperator. Theadjointoperator represents theaction
byconjugationof theLiealgebraonitselfandisdefinedby:
ad : g×g→ g
n,m → admn=m ·n−n ·m= ddt ∣∣∣
t=0 (AdMn(t))= [m,n]with { .
N(0)=n∈ g
.
M(0)=m∈ g (196)
Wecanthencompute thisoperator forourusecase:
n1L= ⎡⎣ R−1/21 .R1/21 R−1/21 .m1
0 0 ⎤⎦ , n2L= ⎡⎣ R−1/22 .R1/22 R−1/22 .m2
0 0 ⎤⎦ (197)
adn1Ln2L=[n1L,n2L]= ⎡⎢⎣ 0 R−1/21 (
.
R 1/2
1 .
m2− .
R 1/2
2 .
m1 )
R−1/22
0 0 ⎤⎥⎦ (198)
adn1Rn2R=[n1R,n2R]= ⎡⎣ 0 R−1/21 .R1/21 (
−R−1/22 .
R 1/2
2 m2+ .
m2 )
−R−1/22 .
R 1/2
2 (
−R−1/21 .
R 1/2
1 m1+ .
m1 )
0 0 ⎤⎦ (199)
89
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