+ Reply to Thread
Results 1 to 3 of 3

Thread: Conditional distribution for covariance matrix with constraint

  1. #1
    Points: 2,053, Level: 27
    Level completed: 36%, Points required for next Level: 97

    Posts
    6
    Thanks
    2
    Thanked 0 Times in 0 Posts

    Conditional distribution for covariance matrix with constraint




    Hi,

    I'm trying to work out how to draw from the conditional distribution for a covariance matrix given some correlated multivariate Gaussian data.

    I know the univariate distributions are all standard normal (i.e. zero mean, unit variance), so the only unknowns are the off-diagonal elements of the covariance matrix.

    If it wasn't for this constraint, I could just draw from the Wishart distribution, but I'm not sure what to do with this constraint in place.

    Thanks for any pointers or suggestions.

  2. #2
    TS Contributor
    Points: 22,410, Level: 93
    Level completed: 6%, Points required for next Level: 940

    Posts
    3,020
    Thanks
    12
    Thanked 565 Times in 537 Posts

    Re: Conditional distribution for covariance matrix with constraint

    Do you mean you already know that the data are from a multivariate normal, with the mean vector being a zero vector and the covariance matrix has a diagonal of one?

    And is it your only constraint? Or you mean other constraint? And you want to ask the distribution of the scatter matrix under this constraint, compare to the usual setting where you got the wishart distribution?

  3. #3
    Points: 2,053, Level: 27
    Level completed: 36%, Points required for next Level: 97

    Posts
    6
    Thanks
    2
    Thanked 0 Times in 0 Posts

    Re: Conditional distribution for covariance matrix with constraint


    Quote Originally Posted by BGM View Post
    Do you mean you already know that the data are from a multivariate normal, with the mean vector being a zero vector and the covariance matrix has a diagonal of one?
    Yes that's right.


    Quote Originally Posted by BGM View Post
    And is it your only constraint? Or you mean other constraint? And you want to ask the distribution of the scatter matrix under this constraint, compare to the usual setting where you got the wishart distribution?
    Yes, when I say constraint I mean the fact that the covariance matrix has ones on the diagonal. Sorry it wasn't clear.

+ Reply to Thread

           




Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts






Advertise on Talk Stats