Multivariate stable distribution

The multivariate stable distribution is a multivariate probability distribution that is a multivariate generalisation of the univariate stable distribution. The multivariate stable distribution defines linear relations between stable distribution marginals.[clarification needed] In the same way as for the univariate case, the distribution is defined in terms of its characteristic function.

multivariate stable
Probability density function

Heatmap showing a Multivariate (bivariate) stable distribution with α = 1.1
Parameters exponent
- shift/location vector
- a spectral finite measure on the sphere
Support
PDF (no analytic expression)
CDF (no analytic expression)
Variance Infinite when
CF see text

The multivariate stable distribution can also be thought as an extension of the multivariate normal distribution. It has parameter, α, which is defined over the range 0 < α ≤ 2, and where the case α = 2 is equivalent to the multivariate normal distribution. It has an additional skew parameter that allows for non-symmetric distributions, where the multivariate normal distribution is symmetric.

Definition

edit

Let   be the unit sphere in  . A random vector,  , has a multivariate stable distribution - denoted as   -, if the joint characteristic function of   is[1]

 

where 0 < α < 2, and for  

 

This is essentially the result of Feldheim,[2] that any stable random vector can be characterized by a spectral measure   (a finite measure on  ) and a shift vector  .

Parametrization using projections

edit

Another way to describe a stable random vector is in terms of projections. For any vector  , the projection   is univariate  stable with some skewness  , scale   and some shift  . The notation   is used if X is stable with   for every  . This is called the projection parameterization.

The spectral measure determines the projection parameter functions by:

 
 
 

Special cases

edit

There are special cases where the multivariate characteristic function takes a simpler form. Define the characteristic function of a stable marginal as

 

Isotropic multivariate stable distribution

edit

The characteristic function is   The spectral measure is continuous and uniform, leading to radial/isotropic symmetry.[3] For the multinormal case  , this corresponds to independent components, but so is not the case when  . Isotropy is a special case of ellipticity (see the next paragraph) – just take   to be a multiple of the identity matrix.

Elliptically contoured multivariate stable distribution

edit

The elliptically contoured multivariate stable distribution is a special symmetric case of the multivariate stable distribution. If X is α-stable and elliptically contoured, then it has joint characteristic function   for some shift vector   (equal to the mean when it exists) and some positive definite matrix   (akin to a correlation matrix, although the usual definition of correlation fails to be meaningful). Note the relation to characteristic function of the multivariate normal distribution:   obtained when α = 2.

Independent components

edit

The marginals are independent with  , then the characteristic function is

 

Observe that when α = 2 this reduces again to the multivariate normal; note that the iid case and the isotropic case do not coincide when α < 2. Independent components is a special case of discrete spectral measure (see next paragraph), with the spectral measure supported by the standard unit vectors.

 
Heatmap showing a multivariate (bivariate) independent stable distribution with α = 1
 
Heatmap showing a multivariate (bivariate) independent stable distribution with α = 2

Discrete

edit

If the spectral measure is discrete with mass   at   the characteristic function is

 

Linear properties

edit

If   is d-dimensional, A is an m x d matrix, and   then AX + b is m-dimensional  -stable with scale function   skewness function   and location function  

Inference in the independent component model

edit

Recently[4] it was shown how to compute inference in closed-form in a linear model (or equivalently a factor analysis model), involving independent component models.

More specifically, let   be a set of i.i.d. unobserved univariate drawn from a stable distribution. Given a known linear relation matrix A of size  , the observation   are assumed to be distributed as a convolution of the hidden factors  .  . The inference task is to compute the most probable  , given the linear relation matrix A and the observations  . This task can be computed in closed-form in O(n3).

An application for this construction is multiuser detection with stable, non-Gaussian noise.

See also

edit

Resources

edit

Notes

edit
  1. ^ J. Nolan, Multivariate stable densities and distribution functions: general and elliptical case, BundesBank Conference, Eltville, Germany, 11 November 2005. See also http://academic2.american.edu/~jpnolan/stable/stable.html
  2. ^ Feldheim, E. (1937). Etude de la stabilité des lois de probabilité . Ph. D. thesis, Faculté des Sciences de Paris, Paris, France.
  3. ^ User manual for STABLE 5.1 Matlab version, Robust Analysis Inc., http://www.RobustAnalysis.com
  4. ^ D. Bickson and C. Guestrin. Inference in linear models with multivariate heavy-tails. In Neural Information Processing Systems (NIPS) 2010, Vancouver, Canada, Dec. 2010. https://www.cs.cmu.edu/~bickson/stable/