Skip to content

[Merged by Bors] - feat: everywhere positive sets for a measure#10139

Closed
sgouezel wants to merge 4 commits intomasterfrom
SG_everywherePos
Closed

[Merged by Bors] - feat: everywhere positive sets for a measure#10139
sgouezel wants to merge 4 commits intomasterfrom
SG_everywherePos

Conversation

@sgouezel
Copy link
Copy Markdown
Contributor

A set s is everywhere positive with respect to a measure mu if, for every point in s, all its neighborhoods inside s have positive measure. We create a new file about this notion.

The main result of the file is Halmos theorem (1950): the Haar measure on a locally compact group is completion-regular, i.e., finite-measure sets can be approximated from inside by level sets of compactly supported continuous functions. (its proof uses subtle properties of everywhere positive sets, although this is not apparent in the statement).


Open in Gitpod

Copy link
Copy Markdown
Contributor

@ocfnash ocfnash left a comment

Choose a reason for hiding this comment

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

Thanks!

bors d+

/-- If two measures coincide locally, then a set which is everywhere positive for the former is
also everywhere positive for the latter. -/
lemma IsEverywherePos.of_forall_exists_nhds_eq (hs : IsEverywherePos μ s)
(h : ∀ x ∈ s, ∃ t ∈ 𝓝 x, ∀ u ⊆ t, ν u = μ u) : IsEverywherePos ν s := by
Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

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

Does this concept of two measures coinciding locally occur enough that we should have a definition + API for it?

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 don't think so, because we have just one application for now. I will reconsider when we have a second application :-)

Copy link
Copy Markdown
Collaborator

@AntoineChambert-Loir AntoineChambert-Loir Feb 10, 2024

Choose a reason for hiding this comment

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

Deleted comment that measures form a sheaf — of course, that requires some countability assumption unless one mostly cares about measures of sigma-compact sets. On the other hand, I wonder whether the sheaf-theoretical version of measures would be. What are the objects that are locally defined by measures? What does it mean that two measures coincide locally?

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.

It really depends on the sigma-algebra you consider. On the sigma algebra of Baire sets (generated by compact G-delta sets), measures form indeed a sheaf. On the Borel sigma-algebra, they don't (an example I like is: let \mu be zero on sets which are contained in a countable union of balls of radius 1, and +\infty otherwise. On a non-separable Hilbert space, this is a measure which is locally zero, with trivial support, but still nonzero).

@mathlib-bors
Copy link
Copy Markdown
Contributor

mathlib-bors bot commented Feb 9, 2024

✌️ sgouezel can now approve this pull request. To approve and merge a pull request, simply reply with bors r+. More detailed instructions are available here.

@ghost ghost added delegated This pull request has been delegated to the PR author (or occasionally another non-maintainer). and removed awaiting-review labels Feb 9, 2024
Co-authored-by: Oliver Nash <github@olivernash.org>
@sgouezel
Copy link
Copy Markdown
Contributor Author

sgouezel commented Feb 9, 2024

bors r+
Thanks for the review!

@github-actions github-actions bot added the ready-to-merge This PR has been sent to bors. label Feb 9, 2024
mathlib-bors bot pushed a commit that referenced this pull request Feb 9, 2024
A set `s`  is everywhere positive with respect to a measure `mu` if, for every point in `s`, all its neighborhoods inside `s` have positive measure. We create a new file about this notion. 

The main result of the file is Halmos theorem (1950): the Haar measure on a locally compact group is completion-regular, i.e., finite-measure sets can be approximated from inside by level sets of compactly supported continuous functions. (its proof uses subtle properties of everywhere positive sets, although this is not apparent in the statement).
@mathlib-bors
Copy link
Copy Markdown
Contributor

mathlib-bors bot commented Feb 9, 2024

Build failed (retrying...):

mathlib-bors bot pushed a commit that referenced this pull request Feb 9, 2024
A set `s`  is everywhere positive with respect to a measure `mu` if, for every point in `s`, all its neighborhoods inside `s` have positive measure. We create a new file about this notion. 

The main result of the file is Halmos theorem (1950): the Haar measure on a locally compact group is completion-regular, i.e., finite-measure sets can be approximated from inside by level sets of compactly supported continuous functions. (its proof uses subtle properties of everywhere positive sets, although this is not apparent in the statement).
@mathlib-bors
Copy link
Copy Markdown
Contributor

mathlib-bors bot commented Feb 9, 2024

Build failed (retrying...):

mathlib-bors bot pushed a commit that referenced this pull request Feb 9, 2024
A set `s`  is everywhere positive with respect to a measure `mu` if, for every point in `s`, all its neighborhoods inside `s` have positive measure. We create a new file about this notion. 

The main result of the file is Halmos theorem (1950): the Haar measure on a locally compact group is completion-regular, i.e., finite-measure sets can be approximated from inside by level sets of compactly supported continuous functions. (its proof uses subtle properties of everywhere positive sets, although this is not apparent in the statement).
@mathlib-bors
Copy link
Copy Markdown
Contributor

mathlib-bors bot commented Feb 9, 2024

Build failed:

@sgouezel
Copy link
Copy Markdown
Contributor Author

bors r+

mathlib-bors bot pushed a commit that referenced this pull request Feb 10, 2024
A set `s`  is everywhere positive with respect to a measure `mu` if, for every point in `s`, all its neighborhoods inside `s` have positive measure. We create a new file about this notion. 

The main result of the file is Halmos theorem (1950): the Haar measure on a locally compact group is completion-regular, i.e., finite-measure sets can be approximated from inside by level sets of compactly supported continuous functions. (its proof uses subtle properties of everywhere positive sets, although this is not apparent in the statement).
@mathlib-bors
Copy link
Copy Markdown
Contributor

mathlib-bors bot commented Feb 10, 2024

Pull request successfully merged into master.

Build succeeded:

@mathlib-bors mathlib-bors bot changed the title feat: everywhere positive sets for a measure [Merged by Bors] - feat: everywhere positive sets for a measure Feb 10, 2024
@mathlib-bors mathlib-bors bot closed this Feb 10, 2024
@mathlib-bors mathlib-bors bot deleted the SG_everywherePos branch February 10, 2024 07:02

variable {α : Type*} [TopologicalSpace α] [MeasurableSpace α]

/-- A set `s` is *everywhere positive* (also called *self-supporting*) with respect to a
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.

Isn't there a notion of support for a measure? This seems to be equivalent to the fact that the support of the restriction of the measure to s is equal to s.

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.

(By reading the file further down, I understand that you're defining that, without defining the support of the measure in general)

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.

Yes, I could have phrased it in terms of support, by requiring that the support of the restriction of the measure to s contains s.

rw [hu]
exact hs.diff u_open.measurableSet

protected lemma _root_.IsClosed.everywherePosSubset (hs : IsClosed s) :
Copy link
Copy Markdown
Collaborator

@AntoineChambert-Loir AntoineChambert-Loir Feb 10, 2024

Choose a reason for hiding this comment

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

The support is closed, that's a general property, so the support will be closed in s, for the induced topology, whatever s is closed or not.

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.

Yes, that's what the previous lemma says.


open Pointwise

/-- If a compact closed set is everywhere positive with respect to a left-invariant measure on a
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.

Probably, this does not require that the mesure is left invariant, but much less : for example, if translating the measure multiplies it by a >0 function…

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.

Indeed, the proof just requires that the measure is quasi-invariant by left-multiplication (i.e., left-multiplication respects zero measure sets), which by the way only depends on the measure class and not the measure itself. If I'm not mistaken, in each measure class which is quasi-invariant by left multiplication, there is a measure which is exactly invariant (maybe one needs some sigma-finiteness here to apply Radon-Nikodym), so the current statement does not really lose much generality -- we'll generalize if/when we need it!

atarnoam pushed a commit that referenced this pull request Feb 10, 2024
A set `s`  is everywhere positive with respect to a measure `mu` if, for every point in `s`, all its neighborhoods inside `s` have positive measure. We create a new file about this notion. 

The main result of the file is Halmos theorem (1950): the Haar measure on a locally compact group is completion-regular, i.e., finite-measure sets can be approximated from inside by level sets of compactly supported continuous functions. (its proof uses subtle properties of everywhere positive sets, although this is not apparent in the statement).
dagurtomas pushed a commit that referenced this pull request Mar 22, 2024
A set `s`  is everywhere positive with respect to a measure `mu` if, for every point in `s`, all its neighborhoods inside `s` have positive measure. We create a new file about this notion. 

The main result of the file is Halmos theorem (1950): the Haar measure on a locally compact group is completion-regular, i.e., finite-measure sets can be approximated from inside by level sets of compactly supported continuous functions. (its proof uses subtle properties of everywhere positive sets, although this is not apparent in the statement).
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

delegated This pull request has been delegated to the PR author (or occasionally another non-maintainer). 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