Web-Books
in the Austria-Forum
Austria-Forum
Web-Books
Naturwissenschaften
Physik
Differential Geometrical Theory of Statistics
Page - 189 -
  • User
  • Version
    • full version
    • text only version
  • Language
    • Deutsch - German
    • English

Page - 189 - in Differential Geometrical Theory of Statistics

Image of the Page - 189 -

Image of the Page - 189 - in Differential Geometrical Theory of Statistics

Text of the Page - 189 -

Entropy2016,18, 433 Theconsiderationswejustdiscussedmayhaveremarkableimpactsonthetopological-geometrical structureofstatisticalmanifolds. Fromourbriefdiscussionweconclude Theorem12. Consider a statisticalmanifold (M,g,∇,∇∗). Every non trivial simple symplectic foliation ω∈Ω∇2 (M) isdefinedbyaRiemanniansubmersiononasymplecticmanifold. Corollary4. Everynontrivial simple symplecticweb [ωj, j∈ J]⊂Ω∇2 (M) isdefinedby familyofRiemannian submersionsonsymplecticmanifols. 5.2. TheHessian InformationGeometry,Continued In [52]Shimapointedout that theFisher informationsofmanyclassical statisticalmodels are Hessianmetric tensors. Atanother side theexponentialmodels (orexponential family)maybeconsideredasoptimal Statisticalmodels. Asalreadymentionedtheredoesnotexistsanycriterionforknowingwhetheragivenstatistical model is isomorphic toanexponentialmodel [22], [13] In thecategoryof regularmodels,vizmodelswhoseFisher information isaRiemannianmetric, it isknownthat theFisher informationofanexponentialmodel isaHessianRiemannianmetric [18,52]. In this subsectionweaddress thegeneral situation.Wegiveacohomological characterizationof exponentialmodels.Wealso introduceanewnumerical invariant rbwhichmeasureshowfar from beinganexponential family isagivenstatisticalmodel. See theAppendixAtothispaper. Werecall that themetric tensorgofaHessianstructure (M,D,g) isa2-cocycleof theKVcomplex [C∗KV(A,R),δKV]. Tononspecialistswegotorecallsomeconstructions inthegeometryofKoszul [2,52], seealso[53]. Let ((M,x∗),D)beapointed locallyflatmanifoldwhoseuniversal covering isdenotedby (M˜,D˜). Here the topological space M˜ is thesetofhomotopyclassofcontinuouspaths {([0,1],0)→ (M,x∗)}. Its topology is thecompact-opentopology. Let cbeacontinuouspathwith c(0)=x∗. For s∈ [0,1] theparallel transportofTx∗M inTc(s)M isdenotedbyτs. OnedefinesQ(c)∈Tx∗Mby Q(c)= ∫ 1 0 τ−1s ( dc(s) ds )ds. The tangentvectorQ(c)dependsonlyonthehomotopyclassof thepath c. Therefore,Qdefines a localhomeomorphism Q˜ : M˜→Tx∗M. Letπ1(x∗)bethe fundamentalgroupatx∗. Let [γ]∈π1(x∗). Thenatural leftaction π1(x∗)×M˜→ M˜ isgivenbythecompositionofpaths,viz [γ].[c]= [γ◦c]. Theparallel transportalonga loopγ(t)yieldsa linearactionofπ1(x∗) inTx∗Mwhich isdenoted by f([γ]). 189
back to the  book Differential Geometrical Theory of Statistics"
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
Web-Books
Library
Privacy
Imprint
Austria-Forum
Austria-Forum
Web-Books
Differential Geometrical Theory of Statistics