QuantEcon / QuantEcon/lecture-python-advanced.myst
[MCMC] Add diagnostics and a potential sequal
Nobody has claimed this yet.
- Dominant language
- TeX
- Stars
- 37
- Forks
- 26
- Avg merge
- 13d 22h
- Merged PRs (30d)
- 2
Description
These are summerized from a discussion with @jstac:
Some minor suggestions that we could add to the lecture on how to diagnose MCMC and where the Metropolis–Hastings algorithm may not perform well:
- Add trace plots for the JAX sampler under different proposal scales.
- Add autocorrelation plots for the same chains.
- Add a 2D correlated posterior example with contours and the chain trajectory overlaid.
- Use the 2D example to illustrate where random-walk Metropolis struggles in finite samples.
In the sequel, we could:
- Study how these issues motivate MALA and HMC/NUTS.
- Connect these algorithms to what is used in
numpyro.
Contributor guide
No contributing guide indexed for this repository
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 locating the existing MCMC lecture and its JAX sampler examples. Review how proposal scales and correlated posteriors are currently presented, then determine how the requested trace plots, autocorrelation plots, contours, and chain trajectories fit together. Done means the diagnostic material is added and the possible MALA, HMC/NUTS, and numpyro sequel is clearly scoped.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- documentation
- Issue type
- Documentation
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Quiet
- Clarity
- Mostly clear
- Newbie friendliness
- 48/100