The Consistent Criteria of Hypotheses for Gaussian Homogeneous Fields Statistical Structures

Z.Zerakidze.et al. Int. Journal of Engineering Research and Applications www.ijera.com
ISSN: 2248-9622, Vol. 6, Issue 4, (Part - 3) April 2016, pp.01-04
www.ijera.com 1|P a g e
The Consistent Criteria of Hypotheses for Gaussian
Homogeneous Fields Statistical Structures
Z.Zerakidze, L.Aleksidze, L. Eliauri
Gori University, Gori, Georgia
ABSTRACT
The present theory of consistent criteria for testing hypothesis of statistical structures of homogeneous Gaussian
fields can be used, for example, in the reliability prediction of different engineering designs. In the paper there
are discussed Gaussian homogeneous fields statistical structures  IiSE ii
,,,  in Hilbert space of
measures. We define the consistent criteria for checking hypotheses such as the probability of any kind of errors
is zero for given criteria. We prove necessary and sufficient conditions for the existence of such criteria.
Keywords: Consistent criteria, Orthogonal, weakly separable, strongly separable statisticalstructures.
Classification cocles62H05, 62H12
Let there is given  SE , measurable
space and on this space there is given
 Iii
, family of probability measures
depended on Ii  parameter.Let bring some
definition     .91 see
Definition 1: A statistical structure is called
object IiSE ii
,,,  .
Definition 2: A statistical
structure IiSE ii
,,,  is called orthogonal
(singular) if i
 and j
 are orthogonal for each
.,, IjIiji 
Definition 3: A statistical structure
 IiSE ii
,,,  is called separable, if there
exists family S-measurable sets  IiX i
, such
thatthe relations are fulfilled:
1.  






ji
ji
XIjIi ji
if
if
,0
,1
, 
2.   ,, cXXcardIjIi ji
 Wh
ere c denotes power continuum.
 IiSE ii
,,,  is called weakly separable if
there exists family S-measurable sets  IiX i
,
such that the relations are
fulfilled:  






ji
ji
X ji
if
if
,0
,1
 .
 IiSE ii
,,,  is called strongly separable is
There exists disjoint family S-measurable sets
 IiX i
, such that the relations are fulfilled
  .1 ji
XIi 
Remark 1: A strong separable there follows
weakly separable. From weakly separable there
follows orthogonal but not vice versa.
Example 1: Let    1,01,0 E and S be
Borel -algebra of parts of E. Take the S-
measurable sets
    .1,0,,10,  iiyxyxX i
Let li be are linear Lebesgue probability measures
on  .1,0, iX i
and 0
l Lebesgue plane
probability measures on    1,01,0 E then the
statistical structure   1,0,,, ilSE ii
is
orthogonal, but not weakly separable.
Remark 2: The countable family of probability
measures    ,...2,1,,  NNkk
 strongly
separable, weakly separable, separable and
orthogonal are equivalent, (see 2,3,4).
Remark 3: On an arbitrary set E of continuum
power one can define Gaussian homogeneous
orthogonal statistical structure having maximal
possible power equal to ,2
2
c
Gaussian
homogeneous weakly statistical structure having
maximal possible power equal to ,2
c
where C is
continuum power (see 3).
Remark 4: A.V Ckorokhod proved (in ZFC&CH
theory) that continual weakly separable statistical
structure as much strongly separable Z.Zerakidze
proved (in ZFC&MA theory) that continuum
power Borelweakly separable statistical structure
as much strongly separable. We deote be (MA) the
Martins axiom.
RESEARCH ARTICLE OPEN ACCESS

Let  be set of Hypotheses and   be  -
algebra of subsets of  which contains all finite
subsets of  .
 HSE H
,,,  will be said to admit a
consistent criteria for checking hypotheses, if there
exists at least one measurable map
    ,,,: SE such that
   .1:  HHxxH

Remark 5. By Z.Zerekidze was introduced
definition a consistent criterion for checking
hypotheses (see 2).
Definition 7. The following probability
    HxxPHH
  : is called the
probability of error of the H-th kind for a given
criterion  .
Known the following theorem (see3)
Theorem 1. The statistical structure
 HSE H
,,,  admits consistent criteria for
checking hypotheses if and only if the probability
of error of all kind is equal to zero for the
criterion .
Let

M be a real linear space of all alternating
finite measurable on S.
Definition 8. A linear subset

MM H
 is
called a Hilbert space of measurable if.
(1) One can introduce on H
M a scale product
 , , H
M , such that H
M is the
Hilbert space , and for every mutually singular
measurable  and  , H
M , the scale
product ;0, 
(2) If H
M and ,1f then
      
A
Hf
MdxxfA  where  xf
is the S-measurable real function and
.,,  ff
Let   ,,...,, 21
Ttttt n
 where T be a
closed bounded subset of
n
R ,
  .,, IiTti
t Gaussian real homogenous
field on T with zero means
   ,,0 IitE  and correlation function
       .,,, IiTsTtStRstE iii

 Iii
, be the corresponding probability
measures given on S and
  IiRf
n
i
 ,,  be spectral densities.
We becalled the Fourier transformation of
generalized function in the sense of Schwartz as
generation Fourier transformation.
Let
 
   
,,
,
~ 2
Iidd
ff
b
n n
R R ii
  


where
  n
Rb  ,,,
~
the Generalization Fourier
transformation of the following function
      .,,,,,, IjiTttsRtsRtsb ji

Then i
 and j
 are pairwise orthogonal (see 9,10)
and  IiSE ii
,,,  are the Gaussian orthogonal
homogenous fields statistical structures. Next, we
consider S-measurable   Iixg i
, functions
such that     


Ii E
ii
dxxg .
2
 Let H
M
the set measures defined by the formula
      


1
,
Ii B
ii
dxxgB  where II 1
a
countable subsets in I is and
    


1
.
2
Ii E
ii
dxxg  define a scale product
on H
M by formula
      


21
,,
21
21
IIi E
iii
dxxgxg 
Where
      SBjdxxgB
jIi B
i
j
ij
  

B,2,1, t
hen H
M is a Hilbert space of measures and
 ,iH
Ti
H
MM 

 where  iH
M  is the
Hilbert space of elements the
form       


1
,,
Ii B
ii
SBdxxgB 
    
E
i
dxxf ,
2
 with the scale product on
 iH
M  by formula
     
E
i
dxxfxf ,, 2121
 where
      .2,1,   jdxxfB
E
ijj
 (see 7).
Let  i
H be a countable family of
hypotheses, denote  B
MFF  the set of real

functions f for which    
E
H
dxxf
i
 is defined
for all ,HH
M
i
 where  .iH
Ni
H
MM 


Theorem 2.Let  iH
Ni
H
MM 

 be a Hilbert
space of measures. The Gaussian homogenous
fields orthogonal statistical structures
 NiSE
iH
,,,  admits a consistent criteria for
checking hypotheses if and only if the
correspondence j
f  defined by the equality
    HHHf
E
H
Mdxxf   ,, is
one –to-one , where  .B
MFf 
Proof. Sufficiency. For  B
MFf  we define
the linear functional f
l by the equality
       ., HfHfH
ldxxf  Denote
as f
l a countable subset in N, for which
    0
E
H
dxxf
i
 for f
Ii  .
Let us consider the functional
iHf
l on  HH
M 
to which it corresponds Then for
 iHHH
M  
21
, we have
             .
211 21  
E E
HHH
E
HH
dxfxfdxxfxfdxxf
i

Therefore 11
ff H
 a.e. with respect to the
measure iH
 Let 0
iH
f a.e. with respect to the
measures iH
 and     
E
HH
dxxf
ii
 ,
     ,
c
HHH
dxxfc
iii
 then
    .,0, jidxxf
jiji HH
E
HH
 
Denote   ,0: xfxC
ii HH then
    .,0 jidxxf
E
HH ji
  Hence if follows,
that   jiC
ij HH
 ,0 . On the other hand
  .0
ij HH
CE Therefore the statistical
structure  NiSE
iH
,,,  is weakly separable
and from Remark 2 follows that the statistical
iH
,,,  is strongly separable.
It is obvious that  NiC
iH
, is a disjunctive
family of S-measurable sets and
  .,1 NiC
ii HH
 Let us definea
mappch     HBHSE ,, 
Likethet   ., NiHC iH i
 We have
   .,1: NiHxx iH j

Necessity. Since the statistical structure
 NiSE
iH
,,,  admits a consistent criterion
for checking hypotheses then the statistical
iH
,,,  is strongly separable
(see 2,4), so there exists S-measurable sets
 NiX i
, such that   .,1 NiX iH i

We put the measure iH
 unto the correspondence
to a function  BH
MFI
i
 . Then
          .,  
E
HHHHHHH iiiiiji
dxxIxIdxxI 
The function        BHH
MFxIxfxf
i
 11
we put the
measures  .
1 iHHH
M   Then
                 
E E E
HHHHH
xIxfxfdxxIxfdxxf
iii 21 211

 .
2 iHH
M   We put the measures
,H
M where     


NIi
Hi
dxxg
i
1
 into
the correspondence to a function
       .
1



NIi
BHi
MFxIxgxf
i
Then
    


NIi
HiH
dxxgM
i
2
1
11
,  we
have
                   
 

E IIi IIi
HiiHii
dxxgxgdxxgxgdxxf
ii
21 21
., 1
11
1

It follows from the proven theorem that the
indicated above correspondence puts some
functions  B
MFf  into the correspondence to
each linear continuous functional .f
l If in
 B
MF we identify function considering with
respect the measures  NiSE
iH
,,,  the
correspondence will be bijecfive. The Theorem 2 is
proved.

Example 2. Let  ,; RE S be
Borel -algebra of parts of R and mH
 are
Lévesque measure on   .1, Zmmm  The
statistical structure  ZmSR m
,,,  is strongly
separable. Let is define  map
    HBHSR ,,  like that
   .,1, ZmHmm m
 We have
   ZmHx mH m
 ,1 e. i.  x is
consistent criteria for checking hypotheses such as
the probability of any Kind of errors is zero for
given criteria .
REFERENCES
[1] I. Ibramlalilov A. Skoroklod. Consistent
criteria of parameters of zaclom
processes. Kiev 1980.
[2] Z. Zerakidze, Generalization of Neimann-
Pearson criterion. Collected scientific of
work (In Georgia), P. 63-69 ISNN 1512-
2271. The ministry of Education and
Science of Georgia. Gory state university.
Tbilisi 2005
[3] L.Aleksidze, Z.Zerekidze Construction of
Gaussian Homogenous isotropic statistical
structures. Bull Acad. Sci. Georgia SSR
169 #3 2004. 456-457.
[4] L.EliauriM.Mumladze, Z.Zerekidze.
Consistent criteria for checking
hypotheses. Joznal of Mathemaics and
System science # 3 #10 2013 p. 514-518
[5] Z.Zerekidze. On weakly divisible and
divisible families of probability measures.
Bull. AcodSciGeoergia SSR 113,0984
[6] Z.Zerekidze.Constaction of statistical
structures. Theory of probability and
application v. XV, 3, 1970, p. 573-578.
[7] Z.Zerekidze. Hibbest space of measures.
Ukx Mat. Journ 38.2 Kiev 1986 p. 148-
154
[8] A.Kharazishvil. On the existence of
consistent estimators for stongly separable
family probability measures. The
probability theory and mathecitiralstatistic
,,Hesnierebs” Tbilisi 1989 p. 100-105
[9] Z. Zerakidze, about the conditions
equivalence of Gaussian measures
corresponding to homogeneous fields.
Works of Tbilisi state University v –II,
(1969) 215-220.
[10] C. Krasnitskii. About the conditions
equivalence and orthogonality of
Gaussian measures corresponding to
homogeneous fields. Theory of
probability and application v. XVIII, 3,
1973, 615-623.

The Consistent Criteria of Hypotheses for Gaussian Homogeneous Fields Statistical Structures

More Related Content

Similar to The Consistent Criteria of Hypotheses for Gaussian Homogeneous Fields Statistical Structures (20)

Recently uploaded (20)

The Consistent Criteria of Hypotheses for Gaussian Homogeneous Fields Statistical Structures