Sign in
Microsoft.com
 
Microsoft
 
 
 

locked
Complex Gaussians. Arbitrary exponential models RRS feed

 > 

  • Question

  • Question
    Sign in to vote
    0
    Sign in to vote

    Hello, 

    I have gone through the doc and haven't seen any examples\mentions of modeling complex variables. I am wondering whether you are just not interested in those, or there exists an issue (possibly theoretical?) that does not allow seemless support for complex variables. Or maybe there is a workaround that allows representing them with the available set of distributions (e.g. like circularly-symmetric as a multivariate gaussian)?

    Also, if one wants to model some exponentially distributed variables (I am particularly interested in chi-squared), if I understand correctly it should be possible (at least theoretically) to apply variational bayes. Does this also mean that it will be possilbe to extend the VBM engine to support those?

    Thank you!


    Sunday, March 10, 2013 5:08 PM

Answers

  • Question
    Sign in to vote
    0
    Sign in to vote

    To answer the first question, we haven't been interested in complex variables yet.  You should be able to work with pairs of real-valued variables for many things.  Is there something that you want to do with complex numbers that you can't do with a pair of real-valued variables?

    For the second question, the exponential distribution and the Chi-squared distribution are both special cases of the Gamma distribution, so these are already supported using Variable.Gamma.

    • Marked as answer by psycharo Friday, April 5, 2013 5:32 PM
    Monday, March 18, 2013 1:04 PM
    Owner