I'm thinking about implementing NUTS sampler in dynesty instead of hslice (or in addition)
https://arxiv.org/abs/1111.4246
as the existing hslice code is very slow and a bit obscure.
As far as I can see the mass matrix can be set based on the bounding ellipsoids and the step size can be adjusted in the same way the scaling is done to 'scale' of rwalk/rslice.
This issue is just to start the discussion/not to forget.
I'm thinking about implementing NUTS sampler in dynesty instead of hslice (or in addition) https://arxiv.org/abs/1111.4246 as the existing hslice code is very slow and a bit obscure.
As far as I can see the mass matrix can be set based on the bounding ellipsoids and the step size can be adjusted in the same way the scaling is done to 'scale' of rwalk/rslice.
This issue is just to start the discussion/not to forget.