dfm / dfm/tinygp

Conditional sampling with vmap

Open
#157 1 comment 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Python
Stars
349
Forks
35
Avg merge
2d 2h
Merged PRs (30d)
1

Description

Hi!

As discussed in #39 one has to use vmap with numpyro plate. But how to obtain a conditional samples (e.g. on a finer X grid) in that case? In this tutorial it is done as follows (but w/o vmaping):

numpyro.deterministic("pred", gp.condition(y, true_t).gp.loc)

However, I can't get GaussianProcess instance out of the vmap-ed function build_gp (as in #39).

Thanks in advance!
Ilya

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start with the sampling-with-numpyro section of tutorials/modeling.html and compare it with the build_gp pattern discussed in issue #39. Trace how vmap and numpyro plate affect access to the GaussianProcess instance, then clarify what supported conditional-sampling behavior should look like and how it can be demonstrated.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
machine-learning
Issue type
Feature
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.