Friday, 30 December 2022

SRFL News: Distance of Word

SRFL News: Distance of Word:   Complex Manifold Deformation Theory ​   ​   Conjecture A   1 Distance of Word   ​   TANAKA Akio          ​   Conjecture   Word has distanc...

SRFL News: Generating Function

SRFL News: Generating Function:   Symplectic Language Theory   TANAKA Akio          Note5   Generating Function ​ ​ ​ [Gromov-Witten invariant] ​ 3-dimensional Calabi-Yau m...

SRFL News: Symmetry of Language

SRFL News: Symmetry of Language:   Time of WANG Guowei Symmetry of Language TANAKA Akio September 15, 2011 [Preparation] 3 dimensional manifold  M Level of affine Lie ring  ...

Thursday, 29 December 2022

SRFL News: Structure of Word

SRFL News: Structure of Word:   Notes for KARCEVSKIJ Sergej Structure of Word TANAKA Akio September 11, 2011 [Preparation] <Therem> Differential manifold  M M '...

Monday, 19 December 2022

SRFL News: At least three elements for language universals 2015

SRFL News: At least three elements for language universals 2015:   At least three elements for language universals TANAKA Akio Supposition 1: Three elements for language universals For language universals,...

Sunday, 18 December 2022

SRFL News: Farewell to Language Universals Revised 2021

SRFL News: Farewell to Language Universals Revised 2021:     Farewell to Language Universals 18/03/2020 19:40 ; First uploaded date to Sekinan View I ever thought of language from the vie point of ...

SRFL News: Letter to O. again Bridge from Old China's Linguistics (Xiaoxue) to Linguistic Circle of Prague

SRFL News: Letter to O. again Bridge from Old China's Linguis...: Letter to O. again Bridge from Old China's Linguistics (Xiaoxue) to Linguistic Circle of Prague 20/05/2019 18:54 Letter to O. again Brid...

SRFL News: Publishing Revolution

SRFL News: Publishing Revolution:   Publishing Revolution 20/05/2019 19:22 Publishing Revolution TANAKA Akio   I read Pamela Pfiffner's  INSIDE THE PUBLISHING REVOLUTION ...

SRFL News: Hurrying up to library 2015. Note added 2016

SRFL News: Hurrying up to library 2015. Note added 2016:   Hurrying up to library Note added 20/05/2019 19:03   TANAKA Akio When I was a high school student, my dream of near future was to go to li...

Tuesday, 13 December 2022

SRFL News: For WITTGENSTEIN Ludwig Revised with Symplectic La...

SRFL News: For WITTGENSTEIN Ludwig Revised with Symplectic La...:   For WITTGENSTEIN Ludwig Revised with Symplectic Language Theory and Floer Homology Language   For WITTGENSTEIN Ludwig  0  Revised Position...

Tuesday, 6 December 2022

Stochastic Meaning Theory 5 Language as the Brownian motion

 



Language as the Brownian motion


[A]
1
Abstractive space     Ω
σ additive family that consists of subset of Ω     F
Measure that is defined over F     P
P satisfies P (Ω ) = 1. 
P      probability measure over ( Ω, F )
Ω      sample space     
Ω, F , )     Probability space
Element of Ω     sample ω
Element of F     event A
Probability that event A occurs     probability P ( )
Real number valued Borel measurable function over Ω     random variable X = X ω )
Random variable is integrable.
Mean (Expectation) of X     E[X] = Ω ω ) P  )
2
Measurable space     ( S)
X : Ω, F )  S)
X is measurable.
X      S-value random variable.
Random variable     X1, …, Xd
X : = (X1, …, Xd )     Rd-value random variable
3
Rd-value random variable     X
E[X i 2] < 
E[(X - E[X])2]     variance
4
S-value random variable.     X
PX : = P ( X A ), AS     distribution
5
Real number space     R
Borel set family over R    R )
Probability measure over ( RR ) )     μ
6
Rd-value random variable     X
ψX (ξ ) : = E[eiξ・X], ξ∈Rd     characteristic function
7
Lebesgue measure     dx
Mean     mR
Variance     v >0
Measure over R     μ dx ) = -(x-m)2 / 2v dx /     Gauss distribution ( normal distribution)
8
(2p – 1 ) !! : = (2p – 1 ) (2p – 3 ) … 31
E[X2p] = (2p – 1 ) !! p     moment of X
9
Event     ABF
When (AB) = P(A) P(B), A and B are independent each other.
10
Integrable and independent random variable     XY
Product XY     integrable
E[XY] = E[X]E[Y]
11
Time     t
[0, ∞)
Family of Rd-value random variable     X = ( Xt ) t  0     d-dimensional stochastic process
ωΩ
When X(ω) is continuous as function of t.d-dimensional stochastic process is called to be continuous.
12
σ additive family     Ft
F F
 s  t
F s  Ft
(Ft  ) = (Ft  ) t  0   increase information system
13
d-dimensional stochastic process     X = ( Xt ) t  0    
t ≥ 0
XΩ  Rd  is F– measurable.
X = ( Xt ) t  0 is (F) – adapted.
14
Mapping ( tω([0, ∞)×ΩB([0, ∞)]×F Xt ( ωRdRd ) )
When the mapping is measurable, X = ( Xt ) t  0  is called to be measurable.
15
X = ( Xt )
FtFt0,X : = σ XS st )
16
Probability space      ( Ω, F , )    
Stochastic process defined over  ( Ω, F , )      (Bt)≥ = (Bt(ω)) ≥ 0
(Bt)≥ that satisfies the next, it is called Brownian motion.
(i) P ( B0 = 0  ) = 1
(ii) For ωΩBt (ω) is continuous on t.
(iii) For 0 = t0<t1<…<tnnN, {Bti-Bti-1} satisfies the next.
a) {Bti-Bti-1} are independent each other.
b) {Bti-Bti-1} are followed by mean 0 and variance ti-ti-1 of Gauss distribution.
17
(Existence theorem)
Over adequate probability space, there exists Brownian motion.
18
Ω = W0
F = B ( W0 )
Brownian motion has the next.
(i)Bt ) = Wt
(ii) = ( wt ) 0 0
Measure over ( W0, B ( W) )      P
P is called Wiener measure.
19
d-dimensional Brownian motion     B = ( Bt ) t  0
d×d orthogonal matrix     A
ABt     d-dimensional Brownian motion
Sphere     S : = δ B (0, r),  B (0, r) = {|x r }
Hitting time     σS (ω) : = inf{t >0; BS }
Hitting place    BσS (ω)
Distribution of BσS (ω)      uniform stochastic measure
20
d-dimensional Brownian motion     B = ( Bt ) t  0
xRd
Brownian motion from x     ( x + Bt ) t  0
 d = B ( W d )
Space     (dd )
Distribution over  (dd )     Px
Mean on Px     Ex [  ]
Probability space     (dd , P)
Stochastic process over (dd , P)    Bt ( w ), wBt ( w ) = wt
Sub σ additive family of d     Ft=σ (Bs ; s) , Ft Ft Nt≥0 ; N : = {Nd ; Px (N) = 0, Rd }
Ft* = Ft+ : = s>Fs
Shift operator over d     θs : → s0 ; (θs (w) ) t : = wt+s
Bt   θ = Bt+s
21
(Markov property)
xRd
s0
Y (w) : d –measurable bounded function over d
Ex[Yθ1A] = Ex[EBs(w)[Y]θ1A] , AF s
By conditional mean
Ex[YθFs] (w) = EBs(w)[Y
Px-a.s.w
22
(Blumenthal’s 0-1 law)
When AF0 ( = F0* ), Px (A) = 0 or 1
23
Random variant of 1-dimensional Brownian motion starting from the origin     B
σ (0,: = inf {t >0; Bt(0,) }
= {σ(0,= 0 }
F0*
(σ(0,= 0 ) = 0 or 1
t0
P (σ(0,= 0 ) = 1
From symmetry of Brownian motion Bt = -Bt

[B]
Language that has Brownian motion     LB
Lhas actual language and imaginary language.

[References]


To be continued
Tokyo August 12, 2008

Monday, 5 December 2022

Stochastic Meaning Theory 3 Place of Meaning

 




Place of Meaning
For Aurora Theory especially for Dictron and Aurora <Language is aurora dancing above us.>


1
Sample space     Ω
Element of Ω     ω
ω is called sample point.
Subset     CΩ
C is called event.
C = Ω is all event.
C = ø is null event.
1-1
Valued space     X
Index space     I
Space     Ω I
Element     ω = {ai ; iIaiX}
1-2
Ω is finite.     |Ω| =m <
All the subsets of Ω     F
F is all of event C.
F consists of 2m number events.  
Family of subsets of Ω    G
that satisfies the next is called additive family.
(i)  ΩG
(ii)  C⇒ CCG
(iii)  C1C2, …, Ck i =1G
Complement of C     CC
1-3
Family of subsets of Ω   F
that satisfies the next is called perfect additive family.
(i) F is additive family.
(ii) C1C2, …, CkF   i =1F  
1-4
Perfect additive family     F
Measurable space     (Ω, F)
1-5
Ω is finite.
Arbitrary real function     f (ω)
f is called random variable.
1-6
Arbitrary sub-perfect additive family     FF
Arbitrary ab     a b
When ab satisfy the next, it is called what random variable ε f (ω) is F0- measurable.
{ω | f (ω)b}F0
1-7
Function defined over F     P
P that satisfies the next is called probability.
(i) For arbitrary CFP ( C )  0
(ii) P (Ω) = 1
(iii) i = 1, 2, …   When Ciand cicø, P (   i=1Ci ) = ∑ i=1(Ci ).
(C) is called probability of event C.
1-8
(Ω, F, P) is called probability space.

2
2-1
Probability space     (Ω, F, P)
Event     AF, BF
P (B)>0
A’s conditional probability on event B is defined by the next.
P ( A | B ) = 
When event A and B satisfy the next, they are called independent.
P(B) = P(A)P(B)
2-2
Sub-perfect additive family       F1F2
Arbitrary C1F1, C2F2
When Cand C2 satisfy the next, Fand Fare called independent.
P(C1C2) = P(C1)P(C2)
Perfect additive family     F
Finite family of F’s sub-perfect additive family. F1F2, …, Fn
When C1 ,C2, …, Cn satisfy the next, F(1i n) is called independent.
P(C1C2Cn) = P(C1)P(C2)…P(Cn)
2-3
Family of n-number random variable     η1 =f1(ω), …, ηn = fn(ω)
Element of Borel sets’ family     C1, …, Cn
When η1, …, ηsatisfies the next, η1, …, ηis called independent random variable on C1, …, Cn.
P{ η1 =f1(ω)C1, …, ηn = fn(ω)Cn } = n=1 P{ fi(ω)C}
When η1, …, ηhas density function p1(x), …, pn(x), η1, …, ηsatisfies the next.
P{ a1η1b1anηnbn } = n=1bkak pk(x)dx
<Theorem>
Independent random variable     η12…, ηn   
1n
Eηi < 
There exists E(η1η2・・・ ηn ) and  η12…, ηn  = Eη1 …,Eηn is formed.  

3
3-1
Matrix     P = [pij]  (i, j = 1, 2,…, n)
P that satisfies the next is called stochastic matrix.
(i) pij0
(ii) nj =1 pij = 1  (i, j = 1, 2,…, n)
3-2
Probability space     (Ω, F, P)
Sample point     ω
Ω = {ωi}
Cω := {ω}
Probability of ω    p (ω) = P(Cω) = P ({ω})
The set of numbers that satisfies the next is called probability distribution.
(i) (ω)0
(ii) ω(ω) = 1
3-3
Space of sample point ω = (ω0, ω1, …, ωn)      Ω
State space     X
≤ i ≤ n
ωi = {x(1)x(2), …, x(r)}
Initial distribution      
Probability matrix     P(1), P(2), …, P(n)
Probability distribution over Ω     P
X ,  and P(1), P(2), …, P(n) that satisfies the next is called Markov chain.
(ω) = μω0 . μω0ω1(1) …μωn-1ωn(n)
Markov chain that does not depend on k(1kn) is called invariant Markov chain..
3-4
Invariant Markov chain     P
Conditional probability     P(ωs+l = (x(jω=x(i))
P(ωx(i))>0
P(ωs+l = (x(jω=x(i)) = p (s)ij
p (s)ij is called class transitive probability.
3-5
Matrix    P
P has a certain s0.
For arbitrary ij p(s0)ij>0, P is called ergodic.
3-6
<Ergodic theorem>
Ergodic transitive matrix     P
When Markov chain that has P is given, there exists only one probability distribution π = (π1, …, πr)that satisfies the next.
(i) πP = π
(ii) limsp(s)ij = πj

4
4-1
Point     x = (x1, …, xd)  -<xi <
Integer     1id   
Lattice     Zd
Random walk over Zd     Markov chain at state space Zd
Random distribution over Zd     = {pz | zZd}
p that satisfies the next is called to be uniform in space.
Pxy = Py-x
4-2
Locus of random walk     ω = (ω0ω1, …, ωk)
Random walk that starts from the origin     ω0 = 0, -ωi-1 >0
All ωs that first return to the origin toward which ω happens to be at th      Ω(k)
k>0
ωΩ(k)
(ω) = -ω0・・・ k-ωk-1
k ωΩ(k) p (ω)
f 0 := 0
Random walk that satisfies the next is called to be recurrent.
ωΩ(kk = 1
Random walk that satisfies the next is called to be transient.
ωΩ(kk < 1
4-3
Arbitrary bounded sequence     {an}
Generating function of {an}     kanzn
4-4
Generating function    F(z) = kk zk     P(z) = k0 pk zk
pk = ki = 0fi .pk-i
p0 = 1
F(z) = 1 – 1/ P(z)
From Abel’s theorem,
k = 1 k = 1- lim z1(1/ P(z) )
When k = 0 pk  , lim z1(1/ P(z) ) = 1/ k = 0p= 0
Random walk that is only k = 0 pk ∞ is recurrent.
4-5
e : = zZzpz
Random walk that satisfies the next is called simple random walk.
(i)Unit coordinate vector     e1e2, …, ed
(ii-1)When y = ±es (1sd) , py-x = 1/2d.
(ii-2)When y ≠±es (1sd) , py-x = 0.
<Polya’s theorem>
When d = 1, 2 , simple random walk is recurrent.
When 3, simple random walk is transient.
4-6
Unit vector     νn ωn / ||ωn||
Unit vector is distributed on unit sphere by being uniform in space.
4-7
From 4-1
Word : = x = (x1, …, xd)  -<xi <
From 4-5
Language space : = 3 and transient
From 4-6
Sentence : = νn

[References]
<On vector, sphere and Language>
<More details on Aurora Theory group>

Tokyo July 11, 2008