Page - 220 - in Differential Geometrical Theory of Statistics
Image of the Page - 220 -
Text of the Page - 220 -
Entropy2016,18, 433
Accordingtoourconstructiononehas the followingfunctor
qΨ= qp2◦ψ∗.
This functorqΨ is theHessianfunctorof themodel
MΨ=[E1,π,M1,D1,p2◦Ψ].
FurtherΨ×ψ isan isomorphismofMΨontoM2.
Letusprove thatassertion(2) impliesassertion(1).
Byof thedefinitionofmorphismofmodels, thepairΨ×ψ isan isomorphismofM1 ontoM2 if
andonly if
p2◦Ψ= p1.
Hereweset theexplicit formulas. Let∇∈LC(M1). ForallvectorfieldsX,Y inM2wehave
X ·(Y · log(p1))−∇XY · log(p1)=X ·(Y · log(p2◦Ψ))−∇XY · log(p2◦Ψ)
=ψ∗(X) ·(ψ∗(Y) · log(p2))−ψ∗(∇XY) · log(p2).
nowweobserve that
ψ∗(∇XY)= [ψ∗(∇)][ψ∗(X)]ψ∗(Y).
Therefore (2) impies theequality
ψ∗[q[p2](ψ∗(∇))]= q[p2◦Ψ](∇)= qp1.
Thisshowsthe implication (2)→ (1).
Letusprove thatassertion(1) impliesassertion(2).
Nowweassumethat that (1)holds,viz
q[p2◦Ψ] = qp1.
Then bothM1 andMΨ have the sameHessian functor. By the virtue of Lemma8 abovewe
deduce that
p2◦Ψ= p1.
Thisends thedemonstration.
Reminder.
(i) Objects ofGM(Γ,Ξ)arequintuplets
M=[E,π,M,D,p].
Theyare called statisticalmodels for themeasurable set (Ξ,Ω).
(ii) Objects ofFB(MSE)are functors
[E,p]→ [M,D].
Theyare calledMSE-fibrations.
220
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