TransformVariables.jl
TransformVariables.jl copied to clipboard
tpapp
Transformations to contrained variables from ℝⁿ.
It would be useful, if we can optionally define a scale to `ShiftedExp`. For example, if we have two parameter `a` and `b` that are both positive, but have very...
It would be useful if `transform` would have an option, so that the result remains a flat vector: ```Julia transform(t, x, keep_flat=true) ``` A good use case is converting MCMC...
Attempt to implement #95 I think it works in principle. However, it fails some tests due to numerical inaccuracies. Unfortunately, I don't see how to improve this.
Dear Tamas Papp, Thanks for your efforts on TransformVariables.jl. Do you happen do know a source (paper/textbook/blog) which explains the transformation which underlies the UnitVector type? best Jon Eriksen
The `NaN`s will make it immediately obvious when some memory hasn't been properly set. I had a case where zero was the correct value so most often the random memory...
This PR implements the proposed changes to `UnitVector` in #66 (i.e. Stan's approach). If `x == zeros(n)`, the transformation is technically undefined. Stan handles this by always initializing with jitter....
Hi, have you looked into ordered transformations, as in [Stan's approach](https://mc-stan.org/docs/2_27/reference-manual/ordered-vector.html)? I was experimenting with this, and so far this seems to work ok: ```julia const TV = TransformVariables struct...
Would there be any interest in a "special array" transformation to create loading matrices for factor analysis? i.e., when `n` observed variables are represented by `m
I am working on an algorithm where I want to define a transformation for a vector of length `p`. Importantly, `p` may be zero. This transformation is part of a...
The current `UnitVector` transformation currently transforms an `n-1` dimensional vector to an `n` dimensional unit vector. However, upon transformation, all resulting unit vectors lie on the "hemi-hypersphere", where the final...