locked
Question for adding a Wishart prior on a multivariate Gaussian distribution (Migrated from community.research.microsoft.com) RRS feed

  • Question

  • shengbo posted on 03-15-2011 10:50 AM

    Hi Team, could I ask a question about adding a Wishart prior on the precision matrix for a 2-dimensional Gaussian distribution of the latent variable. I refer to the mixture of Gaussian example that comes with Infer.net, and try to do as below: 

    -------------

    Variable<Vector> meanEpsilon = Variable.VectorGaussianFromMeanAndPrecision(Vector.Zero(2), PositiveDefiniteMatrix.Identity(2)).Named("meanEpsilon");

    Variable<PositiveDefiniteMatrix> precEpsilon = Variable.WishartFromShapeAndScale(1.0, PositiveDefiniteMatrix.Identity(2)).Named("precEpsilon");

    Variable<Vector> epsilon = Variable.VectorGaussianFromMeanAndPrecision(meanEpsilon, precEpsilon).Named("epsilon");

    I got a problem for the above code. Basically, it says that "precEpsilon" is a wishart rather that a "PositiveDefiniteMatrix". Could you please help me out? 

    --------------

    Thanks and regards,

    Shengbo

    Friday, June 3, 2011 6:34 PM

Answers

  • shengbo replied on 03-15-2011 11:20 AM

    Hi Minka, 

    Thanks a lot for letting me know. 

    Cheers,
    Shengbo 

    Friday, June 3, 2011 6:34 PM

All replies

  • minka replied on 03-15-2011 11:09 AM

    This isn't yet supported for EP.  You have to switch to VMP to do that.

    Friday, June 3, 2011 6:34 PM
  • shengbo replied on 03-15-2011 11:20 AM

    Hi Minka, 

    Thanks a lot for letting me know. 

    Cheers,
    Shengbo 

    Friday, June 3, 2011 6:34 PM