locked
matrix normal and kronecker product RRS feed

  • Question

  • Does Infer.NET support matrix normal distribution and Kronecker product ? I cannot find them. If not then, I will have to resort to reshaping and vectorizing.

    Also, is there an example usage of Variable.JaggedSubarray(.) ?
    • Edited by wittawatj Saturday, October 18, 2014 2:55 PM
    Friday, October 17, 2014 7:46 PM

Answers

  • No, there is no support for this yet. If you explain what you need, we can consider adding it. 

    There are no examples of JaggedSubarray but it's not a function that you should really need to use.

    • Marked as answer by wittawatj Saturday, October 25, 2014 12:28 PM
    Monday, October 20, 2014 11:34 AM
    Owner

All replies

  • No, there is no support for this yet. If you explain what you need, we can consider adding it. 

    There are no examples of JaggedSubarray but it's not a function that you should really need to use.

    • Marked as answer by wittawatj Saturday, October 25, 2014 12:28 PM
    Monday, October 20, 2014 11:34 AM
    Owner
  • For example, I have a random matrix C which follows matrix normal distribution. Of course, I can flatten C and treat it as VectorGaussian. But the complication comes when  I want to extract each column and do something with it. To extract columns of C (which is now a long vector), I may need like Subarray, Subvector or JaggedSubarray. In principle, it should work. I just find it long and tedious to specify my model in this way.
    Saturday, October 25, 2014 12:28 PM