Web-Books
in the Austria-Forum
Austria-Forum
Web-Books
Naturwissenschaften
Physik
Differential Geometrical Theory of Statistics
Page - 246 -
  • User
  • Version
    • full version
    • text only version
  • Language
    • Deutsch - German
    • English

Page - 246 - in Differential Geometrical Theory of Statistics

Image of the Page - 246 -

Image of the Page - 246 - in Differential Geometrical Theory of Statistics

Text of the Page - 246 -

Entropy2016,18, 383 When s1>āˆ’2, s2>āˆ’3/2and s3>āˆ’1, the integral formula (35)holdswith: γV(s)=(2Ļ€)3/2Ī“(s1+2)Ī“(s2+3/2)Ī“(s3+1). Furthermore,when s1>1, s2>1/2and s3>0, the integral formula (37)holdswith: Ī“V(s)=(2Ļ€)3/2Ī“(s1āˆ’1)Ī“(s2āˆ’1/2)Ī“(s3). 5.MultiplicativeLegendreTransformofGeneralizedPowerFunctions Fors∈Rr>0,weseethat logĪ”āˆ’s isastrictlyconvexfunctionontheconePV. Infact,Ī”āˆ’s isdefined naturallyonPn asaproductofpowersofprincipalminors,andit iswellknownthatsuchlogĪ”āˆ’s is strictlyconvexonthewholePn. In thissection,weshall showthat logĪ”Vāˆ’sandlogĪ“Vāˆ’sarerelatedby theFenchel–Legendre transform. Forx∈PV,wedenotebyIVs (x) theminusgradientāˆ’āˆ‡logĪ”āˆ’s(x)atxwithrespect to the inner product.Namely,IVs (x) isanelementofZV forwhich: (IVs (x)|y)=āˆ’ (d dt ) t=0 logĪ”āˆ’s(x+ ty) (y∈ZV). Similarly,JVs (ξ) :=āˆ’āˆ‡logĪ“āˆ’s(ξ) isdefinedforξ∈Pāˆ—V. Ifq1>0, thenforanyB∈W,wehave: IVs ◦τB=Ļ„āˆ—Bā—¦IVs , (41) JVs ā—¦Ļ„āˆ—B=Ļ„Bā—¦JVs (42) owingto (30)and(34), respectively. Theorem5. For any s∈Rr>0, themapIVs :PV→ZV gives adiffeomorphism fromPV ontoPāˆ—V, andJVs gives the inversemap. Proof. Weshallprove the statementby inductionon the rank. When r= 1,wehaveIVs (x11In1)= s1 x11 In1 =JVs (x11In1) forx11>0. Thus, thestatement is true in thiscase. When r>1,assumethat thestatementholds for thesystemofrank rāˆ’1. LetZ0V be thesubspace ofZV definedby: Z0V := {( x11In1 0 0 x′ ) ; x11∈R, x′ ∈ZV′ } . Bydirect computationwith (31)and(33),wehave: IVs ( x11In1 0 0 x′ ) = ( s1 x11 In1 0 0 IV′s′ (x′) ) , (43) JVs ( ξ11In1 0 0 ξ′ ) = ( s1 ξ11 In1 0 0 JV′s′ (ξ′) ) (44) for x11, ξ11> 0, x′ ∈ PV′ and ξ′ ∈ Pāˆ—V′. By the inductionhypothesis,we see thatIVs :PV∩Z0V → Pāˆ—V∩Z0V isbijectivewith the inversemapJVs :Pāˆ—V∩Z0V→PV∩Z0V. Ifq1=0, thestatementholdsbecauseZV=Z0V. Assumeq1>0. Lemma1(ii) tellsus that, for x∈PV, thereexistuniquex0∈Z0V∩PV andB∈W forwhichx=Ļ„Bx0. Similarly,wesee from(32) that, for ξ ∈Pāˆ—V, there exist unique ξ0 ∈Z0V∩Pāˆ—V andC∈W forwhich ξ = Ļ„āˆ—Cξ0. Therefore,we deduce from(41)and(42) thatIVs :PV →Pāˆ—V isabijectionwithJVs the inversemap. 246
back to the  book Differential Geometrical Theory of Statistics"
Differential Geometrical Theory of Statistics
Title
Differential Geometrical Theory of Statistics
Authors
FrƩdƩric 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