multimodal hdi feature

[renecotyfanboy]

Hi @Samreay
I am drafting a PR to implement multimodal resilient estimator for the posterior parameters.

In the same way as in arviz, I am trying to add HDI as a summarizer. Below are examples for the same chain marked as unimodal and multimodal

I still should tinker a bit around to propose a proper implementation, as most of the current implementation is written by Codex.

The current way to use it is simply to mark a chain as multimodal when building it and to use the HDI stat to summarize each parameter :

chain_multimodal = Chain(
    samples=df,
    name="posterior-multimodal",
    statistics=SummaryStatistic.HDI,
    multimodal=True # Here
)

chain_unimodal = Chain(
    samples=df,
    name="posterior-unimodal",
    statistics=SummaryStatistic.HDI,
    multimodal=False # False by default 
)

[Samreay]

Sounds good to me! No initial thoughts on the implementation, tests, or doco updates apart from - when you're happy with the implementation - we should make sure usage.md is updated too, and potentially add a new example image in addition to docs/resources/stats.png like you've put above :)

[Copilot]

Pull request overview

This PR implements multimodal HDI (Highest Density Interval) estimation for posterior parameters, similar to the functionality provided by ArviZ. The feature allows ChainConsumer to identify and visualize multiple disjoint density bands in multimodal distributions.

• Adds HDI as a new summary statistic option
• Implements multimodal interval detection using density thresholding
• Updates visualization to display multiple intervals with appropriate formatting

Reviewed changes

Copilot reviewed 9 out of 9 changed files in this pull request and generated 9 comments.

Show a summary per file

File	Description
src/chainconsumer/statistics.py	Adds HDI enum value to SummaryStatistic
src/chainconsumer/chain.py	Adds multimodal configuration field and validation logic
src/chainconsumer/analysis.py	Implements HDI computation, interval detection, and multimodal bound conversion
src/chainconsumer/plotter.py	Updates bar plotting to handle multiple intervals and format multimodal titles
tests/test_analysis.py	Adds comprehensive tests for HDI functionality and ArviZ parity
tests/test_plotter.py	Adds tests for multimodal visualization
tests/test_translators.py	Updates MCMC configuration for sequential chain method
docs/examples/plot_7_multimodal_chains.py	Adds example demonstrating multimodal chain usage
docs/examples/plot_5_emcee_arviz_numpyro.py	Minor title correction

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

Pull request overview

Copilot reviewed 9 out of 9 changed files in this pull request and generated 5 comments.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[renecotyfanboy]

Hi @Samreay! I think this is good enough to be reviewed now. Sorry for the spam, I have been messing around with the new copilot reviews

[Samreay]

Only one real comment! Oh and also I think usage.md should have the "Statistics" heading updated in text, even if we dont have an updated stats.png to add to the mix.

[Samreay]

Looks great!

[Samreay]

All good on your end to merge, too? Feel free to merge, if so :)

[renecotyfanboy]

Done! TY for your reviews @Samreay!