Reuse jaxified logp when sampling via jax

[nataziel]

reuse jaxified logp when sampling via jax

Description

#7610 added logic to handle passing a pre-jaxified logp function into the blackjax/numpyro samplers, but missed actually passing the jaxified logp that is computed in sample_jax_nuts

Checklist

• Checked that the pre-commit linting/style checks pass
• Included tests that prove the fix is effective or that the new feature works
• Added necessary documentation (docstrings and/or example notebooks)
• If you are a pro: each commit corresponds to a [relevant logical change]

Type of change

• New feature / enhancement
• Bug fix
• Documentation
• Maintenance
• Other (please specify):

---

📚 Documentation preview 📚: https://pymc--7681.org.readthedocs.build/en/7681/

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 92.64%. Comparing base (358b825) to head (e63a8a2).
Report is 7 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #7681      +/-   ##
==========================================
- Coverage   92.70%   92.64%   -0.06%     
==========================================
  Files         107      107              
  Lines       18391    18324      -67     
==========================================
- Hits        17050    16977      -73     
- Misses       1341     1347       +6     

Files with missing lines	Coverage Δ
pymc/sampling/jax.py	94.11% <ø> (-0.91%)	⬇️

... and 1 file with indirect coverage changes

[nataziel]

@ricardoV94 not sure if you've seen this, but it's a super tiny change that we should have included with #7610 that I just missed

[twiecki]

Claude Code:
Thanks for this PR! I see this is a small but important maintenance fix that adds the parameter to the sampler function call, which was computed earlier but not passed through. This ensures we reuse the already jaxified logp function when sampling via JAX instead of recomputing it unnecessarily.\n\nThis change looks good and should improve efficiency by avoiding duplicate computation of the logp function. Appears to be a small oversight in PR #7610 that this properly fixes.

[ricardoV94]

Claude Code: Thanks for this PR! I see this is a small but important maintenance fix that adds the parameter to the sampler function call, which was computed earlier but not passed through. This ensures we reuse the already jaxified logp function when sampling via JAX instead of recomputing it unnecessarily.\n\nThis change looks good and should improve efficiency by avoiding duplicate computation of the logp function. Appears to be a small oversight in PR #7610 that this properly fixes.

Useless

[ricardoV94]

Thanks @nataziel