Monday, 9 May 2011

pr.probability - Covariance of points distributed in a n-ball

The covariance is a multiple of the identity by simple symmetry considerations. For the constant, you just need, again by symmetry, and integration in spherical coordinates,
mathbbEX21=frac1nmathbbE|X|2=fraccnintR0rn1r2dr=fraccn(n+2)Rn+2,


where R is the radius of your ball and c is a constant depending on n and R. To identify c,
1=cintR0rn1dr=fraccnRn,

so c=n/Rn and your covariance is frac1n+2R2 times the identity matrix. Hopefully I've included enough detail that if I've made an algebra mistake it will be easy for someone else to correct it, but I think I recognize that as the right answer.

No comments:

Post a Comment