随机过程:复合泊松过程

  • Post author:
  • Post category:其他







E

(

N

t

)

=

E

(

(

λ

t

)

k

k

!

e

λ

t

)

=

λ

t

λ

E(N_t)=E(\frac{(\lambda t)^k}{k!}e^{-\lambda t})=\lambda t(因为泊松过程的期望均值都为\lambda)






E


(



N










t


















)




=








E


(














k


!
















(


λ


t



)










k





























e














λ


t










)




=








λ


t












































λ










复合泊松过程de定义级性质





Y

t

=

n

=

1

N

t

ξ

n

,

N

t

ξ

n

t

0

Y_t=\sum_{n=1}^{N_t} \xi_n ,N_t为泊松过程,\xi_n独立同分布,则对于任意t\geq 0







Y










t




















=

















n


=


1




















N










t





































ξ










n


















,





N










t





































ξ










n



















































t













0








  • {

    Y

    t

    }

    \{Y_t是独立增量过程\}






    {




    Y










    t







































    }







  • E

    {

    Y

    t

    }

    =

    λ

    t

    E

    (

    ξ

    )

    E\{Y_t\}=\lambda tE(\xi)






    E


    {




    Y










    t


















    }




    =








    λ


    t


    E


    (


    ξ


    )





    ,



    V

    {

    Y

    t

    }

    =

    λ

    t

    E

    (

    ξ

    2

    )

    V\{Y_t\}=\lambda tE(\xi^2)






    V


    {




    Y










    t


















    }




    =








    λ


    t


    E


    (



    ξ










    2









    )








  • E

    (

    Y

    t

    )

    =

    E

    (

    E

    (

    Y

    t

    N

    t

    )

    )

    ,

    E

    (

    Y

    t

    N

    t

    =

    n

    )

    =

    n

    E

    (

    ξ

    )

    E

    (

    Y

    t

    N

    t

    )

    =

    N

    t

    E

    (

    ξ

    )

    E

    {

    N

    t

    E

    (

    ξ

    )

    }

    =

    E

    (

    ξ

    )

    E

    {

    N

    t

    }

    =

    E

    (

    ξ

    )

    λ

    t

    V

    (

    Y

    t

    )

    =

    E

    (

    V

    (

    Y

    t

    N

    t

    )

    )

    +

    V

    (

    E

    (

    Y

    t

    N

    t

    )

    )

    V

    (

    Y

    t

    N

    t

    =

    n

    )

    =

    V

    (

    n

    =

    1

    N

    ξ

    n

    )

    =

    N

    V

    (

    ξ

    )

    V

    (

    Y

    t

    )

    =

    E

    (

    V

    (

    Y

    t

    N

    t

    )

    )

    +

    V

    (

    E

    (

    Y

    t

    N

    t

    )

    )

    =

    E

    (

    N

    t

    V

    (

    ξ

    )

    )

    +

    V

    (

    N

    t

    E

    (

    ξ

    )

    )

    =

    V

    (

    ξ

    )

    E

    (

    N

    t

    )

    +

    E

    2

    (

    ξ

    )

    V

    (

    N

    t

    )

    =

    [

    V

    (

    ξ

    )

    +

    E

    2

    (

    ξ

    )

    ]

    λ

    t

    =

    E

    (

    ξ

    2

    )

    λ

    t

    {\color{red}均值}:E(Y_t)=E(E(Y_t|N_t)), \\ 其中E(Y_t|N_t=n)=nE(\xi) \\ \Rightarrow E(Y_t|N_t)=N_tE(\xi)\\ E\{N_tE(\xi) \}=E(\xi)E\{N_t \}=E(\xi)*\lambda t \\ {\color{red}方差}:V(Y_t)=E(V(Y_t|N_t))+V(E(Y_t|N_t))\\ V(Y_t|N_t=n)=V(\sum_{n=1}^{N} \xi_n)=N*V( \xi) \\ V(Y_t)=E(V(Y_t|N_t))+V(E(Y_t|N_t))\\ =E(N_t*V( \xi))+V(N_tE(\xi))\\ =V( \xi)*E(N_t)+E\color{red}^2\color{black}(\xi)*V(N_t)\\ =[V( \xi)+E\color{red}^2\color{black}(\xi)]*\lambda t\\ =E(\xi^2)*\lambda t\\

















    E


    (



    Y










    t


















    )




    =








    E


    (


    E


    (



    Y










    t






















    N










    t


















    )


    )


    ,
















    E


    (



    Y










    t






















    N










    t




















    =








    n


    )




    =








    n


    E


    (


    ξ


    )



















    E


    (



    Y










    t






















    N










    t


















    )




    =









    N










    t


















    E


    (


    ξ


    )








    E


    {




    N










    t


















    E


    (


    ξ


    )


    }




    =








    E


    (


    ξ


    )


    E


    {




    N










    t


















    }




    =








    E


    (


    ξ


    )













    λ


    t



















    V


    (



    Y










    t


















    )




    =








    E


    (


    V


    (



    Y










    t






















    N










    t


















    )


    )




    +








    V


    (


    E


    (



    Y










    t






















    N










    t


















    )


    )








    V


    (



    Y










    t






















    N










    t




















    =








    n


    )




    =








    V


    (











    n


    =


    1



















    N





















    ξ










    n


















    )




    =








    N













    V


    (


    ξ


    )








    V


    (



    Y










    t


















    )




    =








    E


    (


    V


    (



    Y










    t






















    N










    t


















    )


    )




    +








    V


    (


    E


    (



    Y










    t






















    N










    t


















    )


    )










    =








    E


    (



    N










    t





























    V


    (


    ξ


    )


    )




    +








    V


    (



    N










    t


















    E


    (


    ξ


    )


    )










    =








    V


    (


    ξ


    )













    E


    (



    N










    t


















    )




    +








    E













    2









    (


    ξ


    )













    V


    (



    N










    t


















    )










    =








    [


    V


    (


    ξ


    )




    +








    E













    2









    (


    ξ


    )


    ]













    λ


    t










    =








    E


    (



    ξ










    2









    )













    λ


    t










  • {

    Y

    t

    }

    \{Y_t\}的特征函数






    {




    Y










    t


















    }























  • φ

    Y

    (

    u

    )

    =

    E

    (

    e

    i

    u

    Y

    )

    =

    n

    =

    0

    +

    E

    [

    e

    i

    u

    Y

    t

    N

    t

    =

    n

    ]

    P

    (

    N

    t

    =

    n

    )

    =

    n

    =

    0

    +

    E

    [

    e

    i

    u

    n

    =

    1

    n

    ξ

    n

    ]

    P

    (

    N

    t

    =

    n

    )

    =

    n

    =

    0

    +

    [

    E

    [

    e

    i

    u

    ξ

    n

    ]

    ]

    n

    P

    (

    N

    t

    =

    n

    )

    =

    n

    =

    0

    +

    [

    φ

    ξ

    (

    u

    )

    ]

    n

    P

    (

    N

    t

    =

    n

    )

    =

    n

    =

    0

    +

    [

    φ

    ξ

    (

    u

    )

    ]

    n

    (

    λ

    t

    )

    n

    n

    !

    e

    λ

    t

    =

    e

    λ

    t

    n

    =

    0

    +

    [

    φ

    ξ

    (

    u

    )

    ]

    n

    (

    λ

    t

    )

    n

    n

    !

    =

    e

    λ

    t

    n

    =

    0

    +

    (

    λ

    t

    φ

    ξ

    (

    u

    )

    )

    n

    n

    !

    =

    e

    λ

    t

    e

    (

    λ

    t

    φ

    ξ

    (

    u

    )

    )

    φ_Y(u)=E(e^{iuY})=\sum_{n=0}^{+\infty}E[e^{iuY_t}|N_t=n]P(N_t=n)\\ =\sum_{n=0}^{+\infty}E[e^{iu\sum_{n=1}^{n} \xi_n}]P(N_t=n)\\ =\sum_{n=0}^{+\infty}[E[e^{iu \xi_n}]]^{n}P(N_t=n)\\ =\sum_{n=0}^{+\infty}[φ_\xi(u)]^{n}P(N_t=n) \\ =\sum_{n=0}^{+\infty}[φ_\xi(u)]^{n}\frac{(\lambda t)^n}{n!}e^{-\lambda t}\\ =e^{-\lambda t}\sum_{n=0}^{+\infty}[φ_\xi(u)]^{n}\frac{(\lambda t)^n}{n!}\\ =e^{-\lambda t}\sum_{n=0}^{+\infty}\frac{(\lambda t*φ_\xi(u))^n}{n!}\\ =e^{-\lambda t}e^{(\lambda t*φ_\xi(u))}







    φ










    Y


















    (


    u


    )




    =








    E


    (



    e











    i


    u


    Y










    )




    =

















    n


    =


    0



















    +























    E


    [



    e











    i


    u



    Y










    t






























    N










    t




















    =








    n


    ]


    P


    (



    N










    t




















    =








    n


    )










    =

















    n


    =


    0



















    +























    E


    [



    e











    i


    u

















    n


    =


    1










    n






















    ξ










    n


























    ]


    P


    (



    N










    t




















    =








    n


    )










    =

















    n


    =


    0



















    +





















    [


    E


    [



    e











    i


    u



    ξ










    n


























    ]



    ]











    n










    P


    (



    N










    t




















    =








    n


    )










    =

















    n


    =


    0



















    +





















    [



    φ










    ξ


















    (


    u


    )



    ]











    n










    P


    (



    N










    t




















    =








    n


    )










    =

















    n


    =


    0



















    +





















    [



    φ










    ξ


















    (


    u


    )



    ]











    n





















    n


    !














    (


    λ


    t



    )










    n




























    e














    λ


    t


















    =









    e














    λ


    t





















    n


    =


    0



















    +





















    [



    φ










    ξ


















    (


    u


    )



    ]











    n





















    n


    !














    (


    λ


    t



    )










    n



































    =









    e














    λ


    t





















    n


    =


    0



















    +


































    n


    !














    (


    λ


    t










    φ










    ξ


















    (


    u


    )



    )










    n



































    =









    e














    λ


    t











    e











    (


    λ


    t






    φ










    ξ


















    (


    u


    )


    )



















    E

    [

    e

    i

    u

    Y

    t

    N

    t

    =

    n

    ]

    =

    E

    [

    e

    i

    u

    n

    =

    1

    n

    ξ

    n

    N

    t

    =

    n

    ]

    =

    [

    E

    [

    e

    i

    u

    ξ

    n

    ]

    ]

    n

    =

    [

    φ

    ξ

    (

    u

    )

    ]

    n

    E

    [

    e

    i

    u

    Y

    t

    N

    t

    ]

    =

    [

    φ

    ξ

    (

    u

    )

    ]

    N

    t

    =

    E

    (

    [

    φ

    ξ

    (

    u

    )

    ]

    N

    t

    )

    =

    n

    =

    0

    +

    [

    φ

    ξ

    (

    u

    )

    ]

    n

    (

    λ

    t

    )

    n

    n

    !

    e

    λ

    t

    =

    e

    λ

    t

    n

    =

    0

    +

    [

    φ

    ξ

    (

    u

    )

    ]

    n

    (

    λ

    t

    )

    n

    n

    !

    =

    e

    λ

    t

    n

    =

    0

    +

    (

    λ

    t

    φ

    ξ

    (

    u

    )

    )

    n

    n

    !

    =

    e

    λ

    t

    e

    (

    λ

    t

    φ

    ξ

    (

    u

    )

    )

    或先求E[e^{iuY_t}|N_t=n]\\ =E[e^{iu\sum_{n=1}^{n} \xi_n}|N_t=n]\\ =[E[e^{iu \xi_n}]]^{n}\\ =[φ_\xi(u)]^{n},所以 E[e^{iuY_t}|N_t]=[φ_\xi(u)]^{N_t}\\ 原式 =E([φ_\xi(u)]^{N_t})=\sum_{n=0}^{+\infty}[φ_\xi(u)]^{n}\frac{(\lambda t)^n}{n!}e^{-\lambda t}\\ =e^{-\lambda t}\sum_{n=0}^{+\infty}[φ_\xi(u)]^{n}\frac{(\lambda t)^n}{n!}\\ =e^{-\lambda t}\sum_{n=0}^{+\infty}\frac{(\lambda t*φ_\xi(u))^n}{n!}\\ =e^{-\lambda t}e^{(\lambda t*φ_\xi(u))}















    E


    [



    e











    i


    u



    Y










    t






























    N










    t




















    =








    n


    ]










    =








    E


    [



    e











    i


    u

















    n


    =


    1










    n






















    ξ










    n






























    N










    t




















    =








    n


    ]










    =








    [


    E


    [



    e











    i


    u



    ξ










    n


























    ]



    ]











    n


















    =








    [



    φ










    ξ


















    (


    u


    )



    ]











    n



















    E


    [



    e











    i


    u



    Y










    t






























    N










    t


















    ]




    =








    [



    φ










    ξ


















    (


    u


    )



    ]












    N










    t








































    =








    E


    (


    [



    φ










    ξ


















    (


    u


    )



    ]












    N










    t


























    )




    =

















    n


    =


    0



















    +





















    [



    φ










    ξ


















    (


    u


    )



    ]











    n





















    n


    !














    (


    λ


    t



    )










    n




























    e














    λ


    t


















    =









    e














    λ


    t





















    n


    =


    0



















    +





















    [



    φ










    ξ


















    (


    u


    )



    ]











    n





















    n


    !














    (


    λ


    t



    )










    n



































    =









    e














    λ


    t





















    n


    =


    0



















    +


































    n


    !














    (


    λ


    t










    φ










    ξ


















    (


    u


    )



    )










    n



































    =









    e














    λ


    t











    e











    (


    λ


    t






    φ










    ξ


















    (


    u


    )


    )














其他相关习题或证明:



订阅报纸




λ

=

6

,

2

3

6

1

2

,

1

3

,

1

6

,

[

0

,

t

]

Y

t

=

n

=

i

N

t

ξ

n

E

(

Y

t

)

=

λ

t

E

(

ξ

n

)

=

18

t

,

V

(

Y

t

)

=

λ

t

E

(

ξ

2

)

=

6

t

(

2

2

1

2

+

3

3

1

3

+

6

6

1

6

)

某书亭订阅报纸的强度为\lambda=6,某位顾客订阅2年,3年,6年的概率为\frac{1}{2},\frac{1}{3},\frac{1}{6},求[0,t]订阅的总年数Y_t=\sum_{n=i}^{N_t} \xi_n的均值和方差\\ 由复合泊松过程性质得到E(Y_t)=\lambda tE(\xi_n)=18t,V(Y_t)=\lambda tE(\xi^2)=6t*(2*2*\frac{1}{2}+3*3*\frac{1}{3}+6*6*\frac{1}{6})







































λ




=








6


,






















2








3








6





























2
















1





















,
















3
















1





















,
















6
















1





















,







[


0


,




t


]





















Y










t




















=





















n


=


i











N










t






































ξ










n











































































E


(



Y










t


















)




=








λ


t


E


(



ξ










n


















)




=








1


8


t


,




V


(



Y










t


















)




=








λ


t


E


(



ξ










2









)




=








6


t













(


2













2

























2
















1























+








3













3

























3
















1























+








6













6

























6
















1





















)






等火车时间




t

E

(

i

=

1

N

t

(

t

S

i

)

)

火车在t时刻启程,求总等待时间的均值E(\sum_{i=1}^{N_t}(t-S_i))















t












































E


(















i


=


1











N










t



































(


t














S










i


















)


)










E

(

i

=

1

N

t

(

t

S

i

)

)

=

E

(

E

[

i

=

1

N

t

(

t

S

i

)

N

t

]

)

1.

E

[

i

=

1

N

t

(

t

S

i

)

N

t

=

n

]

E

[

i

=

1

N

t

(

t

S

i

)

N

t

=

n

]

=

E

[

i

=

1

N

t

t

N

t

=

n

]

E

[

i

=

1

N

t

S

i

N

t

=

n

]

=

n

t

E

[

i

=

1

N

t

S

i

N

t

=

n

]

E

[

i

=

1

N

t

S

i

N

t

=

n

]

=

E

[

i

=

1

n

U

i

]

=

n

t

2

E

[

i

=

1

N

t

(

t

S

i

)

N

t

=

n

]

=

n

t

2

E

[

i

=

1

N

t

(

t

S

i

)

N

t

]

=

N

t

t

2

2.

E

(

N

t

t

2

)

=

E

(

N

t

)

t

2

=

t

2

n

=

0

+

n

(

λ

t

)

n

n

!

e

λ

t

=

λ

t

2

2

E(\sum_{i=1}^{N_t}(t-S_i))=E(E[\sum_{i=1}^{N_t}(t-S_i)|N_t])\\ 1.先求E[\sum_{i=1}^{N_t}(t-S_i)|N_t=n]\\ E[\sum_{i=1}^{N_t}(t-S_i)|N_t=n]=E[\sum_{i=1}^{N_t}t|N_t=n]-E[\sum_{i=1}^{N_t}S_i|N_t=n]\\ =nt-E[\sum_{i=1}^{N_t}S_i|N_t=n]\\ 其中E[\sum_{i=1}^{N_t}S_i|N_t=n]由泊松到达时间的条件性质=E[\sum_{i=1}^{n}U_i]=\frac{nt}{2}\\ 所以E[\sum_{i=1}^{N_t}(t-S_i)|N_t=n]=\frac{nt}{2},E[\sum_{i=1}^{N_t}(t-S_i)|N_t]=\frac{N_tt}{2}\\ 2.E(\frac{N_tt}{2})=E(N_t)\frac{t}{2}=\frac{t}{2}\sum_{n=0}^{+\infty}n\frac{(\lambda t)^n}{n!}e^{-\lambda t}\\=\frac{\lambda t^2}{2}






E


(











i


=


1




















N










t


































(


t














S










i


















)


)




=








E


(


E


[











i


=


1




















N










t


































(


t














S










i


















)






N










t


















]


)








1


.








E


[











i


=


1




















N










t


































(


t














S










i


















)






N










t




















=








n


]








E


[











i


=


1




















N










t


































(


t














S










i


















)






N










t




















=








n


]




=








E


[











i


=


1




















N










t




































t






N










t




















=








n


]













E


[











i


=


1




















N










t





































S










i






















N










t




















=








n


]










=








n


t













E


[











i


=


1




















N










t





































S










i






















N










t




















=








n


]














E


[











i


=


1




















N










t





































S










i






















N










t




















=








n


]








































=








E


[











i


=


1



















n





















U










i


















]




=



















2














n


t
































E


[











i


=


1




















N










t


































(


t














S










i


















)






N










t




















=








n


]




=



















2














n


t























E


[











i


=


1




















N










t


































(


t














S










i


















)






N










t


















]




=



















2















N










t


















t


























2


.


E


(













2















N










t


















t




















)




=








E


(



N










t


















)













2














t






















=



















2














t































n


=


0



















+























n













n


!














(


λ


t



)










n




























e














λ


t


















=



















2














λ



t










2
































版权声明:本文为ResumeProject原创文章,遵循 CC 4.0 BY-SA 版权协议,转载请附上原文出处链接和本声明。