Page - 192 - in Differential Geometrical Theory of Statistics
Image of the Page - 192 -
Text of the Page - 192 -
Entropy2016,18, 433
Nowweassumeamodel (Θ,P) is regular. Then theChristoffel symbols and theFisher informationare
relatedby the formula
Γαij,k=g(∇α∂i∂j,∂k).
Further everyquadruple (Θ,g,∇α,∇−α) is a statisticalmanifold [18,48].
Thuswehavea familyof splittingshort exact sequences
0→Ω∇α(Θ)→M(∇−α,∇α)→S∇α2 (Θ)→0.
Sothemachinerywehavedeveloped in theprecedingsectionscanbeperformed to explore thedifferential
topologyof regular local statisticalmodels. For thatpurpose the crucial tool is the familyofvector space
Sα2(Θ)=S∇
α
2 (Θ).
Weconsider theabstract trivial bundle
∪α[Sα×{α}]→R
whosefiberoverα∈R isSα(Θ). To everyB∈Sα(Θ)weassign theuniqueψ+∈Σ(g)definedby
g(ψ+α(X),Y)=B(X,Y).
Themachinery in theprecedingsubsection leads to the followingproposition.
Proposition8. Weassume (Θ,P) is regular.
(1) Everynonzero singular section
R α→Bα∈Sα(Θ)
gives rise the familyof (g-orthogonal)2-web
TΘ=Ker(ψ+α)⊕ im(ψ+α).
Furtheraccording to thenotationusedpreviously (Bα) is a familyofRiemannian foliationsas in [39,40].
(2) ByreplacingSα(Θ)byΩ∇2 (Θ) everynonzero singular section
R α→ωα∈Ω∇2 (Θ)
yieldsa familyof symplectic foliationsωα.
Reminder.
(i) α→Bα is calleda singular section if eachBα isnon inversible.
(ii) α→ωα is calleda simple section if eachωα is simple.
Wehaveusedsomegaugemorphisms toconstructRiemanniansubmersionsof statisticalmanifoldsover
symplecticmanifolds. Thenotionswe just introduced lead to similar situations.
Theorem15. Let (Θ,P)bea regular statisticalmodelwhoseFisher information isdenotedbyg. Everysimple
nonzero singular section
R α→ωα∈Ωα(Θ)
definesanα-familyofRiemanniansubmersionsof (Θ,g)onto symplecticmanifolds.
192
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