Open vboussange opened 1 year ago
Hi!
Sorry, this went completely under my radar.
It's difficult for me to say what's "right" here as I don't know the transformation you're trying to implement.
But the transformation you've implemented is not bijective (abs
maps both -1
and 1
to the same number).
Hey guys, I really like that you provide this package independently from Turing.jl. I am actually making intensive use of Bijectors.jl for a project that I have called PiecewiseInference.jl.
I need to define few custom bijectors, and I would like to make sure that I am doing things right. Here is one that transforms a random variable with values in $\mathbb{R}$ to a random variable with values in $[C, \infty)$. Here is my code for it.
Is everything alright in this piece of code?
Thanks a lot for your help!