Skip to content

feat(PValues): add IsPValue for composite null hypotheses - #71

Open
rajarshi-mukherjee24 wants to merge 1 commit into
stat-lib:mainfrom
rajarshi-mukherjee24:pvalue-composite-null
Open

rajarshi-mukherjee24 wants to merge 1 commit into
stat-lib:mainfrom
rajarshi-mukherjee24:pvalue-composite-null

Conversation

@rajarshi-mukherjee24

Copy link
Copy Markdown
Collaborator

Summary

Adds Statlib/PValues/PValue.lean, introducing ProbabilityTheory.IsPValue:
a structure characterising a valid p-value for a composite null hypothesis S.

A Pval : 𝓧 → ℝ≥0∞ is an IsPValue for S : Set (ProbabilityMeasure 𝓧) when it is:

  • measurable,
  • bounded: ∀ ω, Pval ω ≤ 1,
  • uniformly valid: for every μ ∈ S and every α, μ {ω | Pval ω ≤ α} ≤ α.

The new module is registered in the root Statlib.lean.

Notes

  • Scope is intentionally limited to the IsPValue definition for the composite null.
  • Values live in ℝ≥0∞, consistent with the measure-theoretic conventions used elsewhere in the library.

Build

Local verification was blocked by a toolchain/Mathlib cache mismatch that forces a full Mathlib rebuild on this machine; relying on CI for the authoritative build. An earlier local build did produce the module's .olean; the only change since was whitespace cleanup.

🤖 Generated with Claude Code

Introduce `ProbabilityTheory.IsPValue`, a structure characterising a
valid p-value for a composite null hypothesis `S`: a measurable
function into `ℝ≥0∞`, bounded by `1`, whose level sets have measure at
most their threshold under every probability measure in `S`.

Register the new module in the root `Statlib.lean`.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

This branch has not been deployed

No deployments
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant