Probability vectors about transformvariables.jl HOT 7 CLOSED

Actually, it might be working close to as-is, just need to check the Jacobian.

from transformvariables.jl.

@cscherrer: can you be more specific about

use this library to work with Dirichlet distributions

ie what you want to do, specifically?

If you want to transform to a simplex, that functionality is missing, but I could add it if that is what you need.

from transformvariables.jl.

Yes, I need to transform to a simplex. My approach was to adapt your l2 remainder for UnitVector to instead use l1 and positives only. Is that how you would go about it, it do you see a better way?

from transformvariables.jl.

I would just code up whatever is used in the Stan docs.

from transformvariables.jl.

Thanks for this handy package!

I also have a use case for a simplex transformation. Would you recommend to use the gist from @cscherrer or rather to implement something based on the Stan documentation?

from transformvariables.jl.

I am glad to hear you find this package useful. Since the code moved on a bit since that gist, I would recommend using the Stan docs directly. It should be reasonably easy, UnitVector should serve as a useful template. I am happy to help out, just ask here, or in a PR you open for this.

from transformvariables.jl.

Closed by #56, thanks @scheidan 👍

from transformvariables.jl.