tensorflow / tensorflow/probability
Other than standard normal distributions for momentum in HMC/NUTS?
Nobody has claimed this yet.
- Dominant language
- Jupyter Notebook
- Stars
- 4.4k
- Forks
- 1.1k
- PR merge metrics
- No merged PRs in 30d
Description
Thanks for the great probabilistic modelling tool!
I am wondering whether there is any simple way to adjust the proposal distribution for momentum in HMC/NUTS samplers? So that I will be able to adapt the Euclidean metric to the shape of the posterior, so that the sampling efficiency is increased. E.g. something similar to how the adaptation is implemented in Stan
https://mc-stan.org/docs/2_23/reference-manual/hmc-algorithm-parameters.html
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start by reviewing the HMC/NUTS sampler behavior described in the issue and compare it with the linked Stan HMC algorithm parameters documentation. Done would be a clearly defined way to adjust the momentum proposal and adapt the Euclidean metric, together with an agreed validation path.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- tensorflow
- Domain
- machine-learning
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100