Skip to content

[Merged by Bors] - feat: better uniqueness results for the Haar measure#9909

Closed
sgouezel wants to merge 52 commits intomasterfrom
SG_haar_unique3
Closed

[Merged by Bors] - feat: better uniqueness results for the Haar measure#9909
sgouezel wants to merge 52 commits intomasterfrom
SG_haar_unique3

Conversation

@sgouezel
Copy link
Copy Markdown
Contributor

@sgouezel sgouezel commented Jan 22, 2024

We show that two Haar measures give the same measure to sets with compact closure, and to open sets, without any regularity assumptions. This is based on McQuillan's answer on https://mathoverflow.net/questions/456670.


I will of course split this into several PRs. Here are the first ones:

Open in Gitpod

@ghost ghost removed the merge-conflict The PR has a merge conflict with master, and needs manual merging. (this label is managed by a bot) label Feb 10, 2024
@[to_additive "Given two left-invariant measures which are finite on compacts,
`addHaarScalarFactor μ' μ` is a scalar such that `∫ f dμ' = (addHaarScalarFactor μ' μ) ∫ f dμ` for
any compactly supported continuous function `f`."]
noncomputable def haarScalarFactor (μ' μ : Measure G) [IsFiniteMeasureOnCompacts μ]
Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

why noncomputable? It should be possible to prove that the factor does not depend on the choice of the chosen functions…

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

noncomputable means that I'm not providing an algorithm by which Lean's kernel could evaluate the number. Since real numbers are already noncomputable by definition, there is no way this function could be computable.

lemma exists_integral_isMulLeftInvariant_eq_smul_of_hasCompactSupport
(μ' μ : Measure G) [IsFiniteMeasureOnCompacts μ] [IsFiniteMeasureOnCompacts μ']
[IsMulLeftInvariant μ] [IsMulLeftInvariant μ'] [IsOpenPosMeasure μ] :
lemma exists_integral_isMulLeftInvariant_eq_smul_of_hasCompactSupport (μ' μ : Measure G)
Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

There is something bizarre here (for me, at least) that mu and mu' have different kind of hypotheses.
Can one compare two left invariant measures without knowing that one of them is positive on open subsets and the other one is finite on compacts?

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

If mu is the zero measure, you can not write mu' = c mu for some c. On the other hand, it's not a problem if mu' is the zero measure. That's why there is an assumption that mu is positive on open sets while there is no such assumption for mu', here and below.

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I have expanded the docstring of haarScalarFactor to insist on this point.


Assume for simplicity that all measures are normalized, so that the scalar factors are all `1`.
First, from the fact that `μ` and `μ'` integrate in the same way compactly supported functions,
they give the same measure to compact "zero sets", i.e., sets of the form `f⁻¹ {1}`
Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

maybe change "zero sets" to "level sets" (and 'f-1{1}' to 'f-1{a} for a neq 0').

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Zero set is unfortunately a standard terminology in this corner of the litterature. That's why I am putting it in the docstring, to make sure that people looking for this can locate the relevant lemmas.


/-- Two left invariant measures give the same mass to level sets of continuous compactly supported
functions, up to the scalar `haarScalarFactor μ' μ`.
Superseded by `measure_isMulInvariant_eq_smul_of_isCompact_closure`, which works for any set with
Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

If it's superseded, should it be kept? or should the proof be just an application of the more general result?

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

All the lemmas whose docstring mentions "superseded" are intermediate steps in the proof of the main theorem below, measure_isMulInvariant_eq_smul_of_isCompact_closure. So I can't remove them. The role of this docstring is to explain that the user shouldn't use them because there is a better version available below.

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I have expanded the docstrings to clarify that these are intermediate lemmas.

@sgouezel sgouezel added awaiting-author A reviewer has asked the author a question or requested changes. and removed awaiting-review labels Feb 12, 2024
@sgouezel sgouezel added awaiting-review and removed awaiting-author A reviewer has asked the author a question or requested changes. labels Feb 13, 2024
Copy link
Copy Markdown
Member

@jcommelin jcommelin left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This all looks very very solid! Great!

bors merge

@ghost ghost added ready-to-merge This PR has been sent to bors. and removed awaiting-review labels Feb 13, 2024
mathlib-bors bot pushed a commit that referenced this pull request Feb 13, 2024
We show that two Haar measures give the same measure to sets with compact closure, and to open sets, without any regularity assumptions. This is based on McQuillan's answer on https://mathoverflow.net/questions/456670.
@mathlib-bors
Copy link
Copy Markdown
Contributor

mathlib-bors bot commented Feb 13, 2024

Build failed (retrying...):

mathlib-bors bot pushed a commit that referenced this pull request Feb 13, 2024
We show that two Haar measures give the same measure to sets with compact closure, and to open sets, without any regularity assumptions. This is based on McQuillan's answer on https://mathoverflow.net/questions/456670.
@mathlib-bors
Copy link
Copy Markdown
Contributor

mathlib-bors bot commented Feb 13, 2024

Build failed (retrying...):

mathlib-bors bot pushed a commit that referenced this pull request Feb 14, 2024
We show that two Haar measures give the same measure to sets with compact closure, and to open sets, without any regularity assumptions. This is based on McQuillan's answer on https://mathoverflow.net/questions/456670.
@mathlib-bors
Copy link
Copy Markdown
Contributor

mathlib-bors bot commented Feb 14, 2024

Pull request successfully merged into master.

Build succeeded:

@mathlib-bors mathlib-bors bot changed the title feat: better uniqueness results for the Haar measure [Merged by Bors] - feat: better uniqueness results for the Haar measure Feb 14, 2024
@mathlib-bors mathlib-bors bot closed this Feb 14, 2024
@mathlib-bors mathlib-bors bot deleted the SG_haar_unique3 branch February 14, 2024 01:35
riccardobrasca pushed a commit that referenced this pull request Feb 18, 2024
We show that two Haar measures give the same measure to sets with compact closure, and to open sets, without any regularity assumptions. This is based on McQuillan's answer on https://mathoverflow.net/questions/456670.
dagurtomas pushed a commit that referenced this pull request Mar 22, 2024
We show that two Haar measures give the same measure to sets with compact closure, and to open sets, without any regularity assumptions. This is based on McQuillan's answer on https://mathoverflow.net/questions/456670.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

ready-to-merge This PR has been sent to bors. t-measure-probability Measure theory / Probability theory

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants