Page - 236 - in Differential Geometrical Theory of Statistics
Image of the Page - 236 -
Text of the Page - 236 -
Entropy2016,18, 383
decomposition. Basedonsuchresults,ourmatrixrealizationmethod[15,17,18]hasbeendeveloped
for thepurposeof theefïŹcient studyofhomogeneouscones. In thispaper,wepresentageneralization
ofmatrix realizationdealingwith awide class of convex cones,which turns out to include cones
associatedwithchordalgraphs.Actually, itwasanenigmafor theauthor thatsomeformulas in[11,19]
for thechordalgraphresembletheformulas in[8,17] forhomogeneousconessomuch,andthemystery
isnowsolvedbytheuniïŹedmethodinthispaper toget the formulas. Furthermore, the techniques
andideas in the theoryofhomogeneouscones, suchasRieszdistributions [8,20,21]andhomogeneous
Hessianmetrics [4,18,22],willbeapplied tovariouscones toobtainnewresults inour futureresearch.
Here,weïŹx somenotationused in thispaper. WedenotebyMat(p,q,R) thevector spaceof
pĂq realmatrices. ForamatrixA,wewrite tA for the transposeofA. The identitymatrixof size p is
denotedby Ip.
2.NewConesPV andPâV
2.1. Setting
WeïŹxapartitionn=n1+n2+ · · ·+nrofapositive integern. LetV={Vlk}1â€k<lâ€rbeasystem
ofvectorspacesVlkâMat(nl,nk,R) satisfying
(V1)AâVlkâAtAâRInl (1†k< l†r),
(V2)AâVlj,BâVkjâAtBâVlk (1†j< k< l†r).
The integer r is called therankof thesystemV.Wedenotebynlk thedimensionofVlk.Note that
somenlk canbezero. LetZV bethespaceof real symmetricmatricesxâSym(n,R)of the form:
x= ââââââ X11 tX21 . . . tXr1
X21 X22 tXr2
... ...
Xr1 Xr2 . . . Xrr ââââââ (
Xkk= xkkInk, xkkâR, k=1,. . . ,r
XlkâVlk, 1†k< l†r )
, (1)
andPV thesubsetofZV consistingofpositivedeïŹnitematrices. Then,PV isaregularopenconvex
cone inZV.
Example 1. Let r= 3, and setV21 := {(
a 0 )
; aâR }
,V31 := {(
0 a )
; aâR }
, andV32 :=R.
Then,ZV is the spaceof symmetricmatrices x of the form:
x= âââââ x1 0 x4 0
0 x1 0 x5
x4 0 x2 x6
0 x5 x6 x3 âââââ . (2)
Weshall see later that the conePV=ZVâ©P4 isnothomogeneous in this case, but admitsvarious integral
formulas, aswell as explicit expressionof theKoszulâVinbergcharacteristic function.
2.2. InductiveDescriptionofPV
If the systemV = {Vlk}1â€k<lâ€r satisïŹes (V1) and (V2), any subsystemVI := {Vlk}k,lâI with
Iâ{1,. . . ,r}alsosatisïŹesthesameconditions. Inparticular, theconecorrespondingtothesubsystem
236
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