Advertisement
If you have a new account but are having problems posting or verifying your account, please email us on hello@boards.ie for help. Thanks :)
Hello all! Please ensure that you are posting a new thread or question in the appropriate forum. The Feedback forum is overwhelmed with questions that are having to be moved elsewhere. If you need help to verify your account contact hello@boards.ie
Hi there,
There is an issue with role permissions that is being worked on at the moment.
If you are having trouble with access or permissions on regional forums please post here to get access: https://www.boards.ie/discussion/2058365403/you-do-not-have-permission-for-that#latest

Tempered Stable Distributions - Computing R(dx)

  • 27-05-2011 10:40pm
    #1
    Closed Accounts Posts: 5


    Im trying to compute R(dx) from a paper by Jan Rosinski, which can be found here, (also here using slightly different notation).

    In the paper on page 3 we have the following theorem

    Theorem 2.3. The Levy measure M of a tempered alpha stable distribution can be written in the form

    [latex]M(A) = \int_{R^d}\int_0^{\infty} I_A(tx)t^{-\alpha-1}e^{-t}dtR(dx)[/latex]

    where I_A(tx) I assume is the indicator funtion, i.e. tx is defined on the interval A

    now I'm using definition of the Gamma function kernal to say

    [latex]\int_0^{\infty} t^{-\alpha-1}e^{-t}dt = \Gamma(-\alpha)[/latex]


    I know the Levy measure M, and so putting that in for M, I then had

    [latex]2^{\alpha}\delta\frac{\alpha}{\Gamma(1-\alpha)}x^{-1-\alpha}e^{-0.5\gamma^{1/\alpha}x} = \Gamma(-\alpha)\int_{R^d}R(dx)[/latex]

    So I need to work out what the function R(dx) is, which is my problem. I was thinking I could just differentiate both sides which would get the integral out of the right hand side and then I could easily rearrange to find R(dx). However I am not sure about this method as I have this integral over the range R^d.

    How do I work out what R(dx) is?

    I was told in another forum by someone...

    "From the statement of the theorem, it seems that equation (2.6) holds for a given Borel measure satisfying the finiteness criterion (2.8). Is this characterization of the measure not enough for your computations?"

    However I fail to see how this helps. Please any help in how I compute R(dx) would be much appreciated.


Comments

  • Registered Users, Registered Users 2 Posts: 2,481 ✭✭✭Fremen


    First observation is that you dropped your indicator function going from line 1 to line 3...


  • Closed Accounts Posts: 5 cernlife


    Thanks for the reply, a little more info on my problem is need I think.

    A tempered stable distribution is when a stable distribution is tempered by an exponential function of the form [latex]e^{-\theta{x}}[/latex]. In my particular case we are using a tempered stable law defined by Barndorff-Nielsen in the paper "modified stable processes" found here, http://economics.ouls.ox.ac.uk/12286/1/nmsprocnew1.pdf.

    In Barndorff's paper, [latex]\theta = (1/2)\gamma^{1/\alpha}[/latex], hence the tempering function is defined as [latex]e^{-(1/2)\gamma^{1/\alpha}{x}[/latex]. In Rosinski's paper he states that tempering of the stable density [latex]f \mapsto f_{\theta}[/latex] leads to tempering of the corresponding Levy measure [latex]M \mapsto M_{\theta}[/latex], where [latex]M_{\theta}(dx) = e^{-\theta{x}}M(dx)[/latex].

    Rosinski then goes on to say the Levy measure of a stable law in polar coordinates is of the form

    [latex]M_0(dr, du) = r^{-\alpha-1}dr\sigma(du)[/latex] (2.1)

    and then says the Levy measure of a tempered stable density can be written as

    [latex]M(dr, du) = r^{-\alpha-1}q(r,u)dr\sigma(du)[/latex] (2.2)

    he then says, the tempering function q in (2.2) can be represented as

    [latex]q(r,u) = \int_0^{\infty}e^{-rs}Q(ds|u)[/latex] (2.3)

    now I'm assuming in my particular case, [latex]q(r,u) = e^{-(1/2)\gamma^{1/\alpha}{r}[/latex] although by now I'm getting lost and unsure of what Im saying.

    Rosinski's paper also defines a measure R by

    [latex]R(A) = \int_{R^d} I_A(x/||x||^2)||x||^{\alpha}Q(dx)[/latex] (2.5)

    and has

    [latex]Q(A) = \int_{R^d} I_A(x/||x||^2)||x||^{\alpha}R(dx)[/latex] (2.6)

    now I know that for my particular tempered stable density the levy measure M is given by

    [latex]2^{\alpha}\delta\frac{\alpha}{ \Gamma(1-\alpha)}x^{-1-\alpha}e^{-(1/2)\gamma^{1/\alpha}x}dx[/latex]

    So im even more confused than ever now, how can I work out what [latex]Q[/latex] is? what is [latex]a[/latex]? how do I use this to work out what R(dx) is? I feel that Im just totally missing the point!

    I really fails to understand (2.5) and (2.6) as I fail to understand what [latex]||x||^{\alpha}[/latex] means, moreover i dont understand what [latex]I_A(x/||x||^2)[/latex] means.


  • Registered Users, Registered Users 2 Posts: 2,481 ✭✭✭Fremen


    I'm not really too comfortable with seeing a levy measure in polar coordinates - I'm not used to it and it's throwing my intuition off completely.

    Have you had a look at Cont and Tankov's book on Levy processes? They have a section on exponential tilting. It may be some help.


Advertisement