Zulip Chat Archive

Stream: maths

Topic: Subtype of a measure space


Yaël Dillies (Jul 13 2022 at 16:40):

Is this true? Context is #2819

import measure_theory.measure.measure_space_def

variables {V : Type*} [measure_space V] {p : V  Prop} {s : set V}

instance subtype.measure_space : measure_space (subtype p) :=
{ volume := measure.comap subtype.val volume,
  ..subtype.measurable_space }

lemma subtype.volume_univ (hs : null_measurable_set s) : volume (univ : set s) = volume s := sorry

Yaël Dillies (Jul 13 2022 at 16:40):

Note that

lemma subtype.volume_univ' (hs : measurable_set s) : volume (univ : set s) = volume s := sorry

is indeed true.

Jason KY. (Jul 13 2022 at 17:01):

This is not true. If s is not measurable then the volume is just 0 (while you can certainly have a nonmeasurable set of measure > 0).

lemma subtype.volume_univ (hs : ¬ measurable_set s) : volume (univ : set s) = 0 :=
begin
  change measure.comap subtype.val volume _ = 0,
  rw [measure.comap, dif_neg],
  { refl },
  { rw [not_and_distrib, not_forall],
    refine or.inr univ, not_imp.2 measurable_set.univ, _⟩⟩,
    simp [hs] },
end

Yaël Dillies (Jul 13 2022 at 17:03):

Hmm, this is annoying because that means #2819 can't use docs#measure_theory.is_fundamental_domain.

Sebastien Gouezel (Jul 13 2022 at 17:06):

This probably means that we should fix the definition of measure.comap, in the same way that we fixed measure.map to allow for almost everywhere measurable functions.

Yaël Dillies (Jul 13 2022 at 17:33):

Would anyone do that for me? :smiley: I am very unaware of the measure theory library.

Sebastien Gouezel (Jul 13 2022 at 17:47):

Sorry, no available free time currently for me (in addition to the heavy duties of old people that have been mentioned in the previous thread, you should add that kids take a significant amount of time in July and August :-)

Rémy Degenne (Jul 13 2022 at 17:55):

I will have a look tomorrow. But if I don't manage to do it in the next two days then I also won't have time before a week or two.

Rémy Degenne (Jul 14 2022 at 10:31):

@Yaël Dillies see branch#RD_comap . PR coming soon.

Rémy Degenne (Jul 14 2022 at 11:22):

#15343

Yaël Dillies (Jul 27 2022 at 19:53):

Still on this, is this now true?

import measure_theory.measure.measure_space_def

variables {V : Type*} [measure_space V] {p : V  Prop} {s : set V}

instance subtype.measure_space : measure_space (subtype p) :=
{ volume := measure.comap subtype.val volume,
  ..subtype.measurable_space }

lemma null_measurable_set.subtype_image {t : set s} (hs : null_measurable_set s) :
  null_measurable_set t  null_measurable_set ((coe : s  V) '' t) := sorry

Rémy Degenne (Jul 28 2022 at 08:30):

Yes this is true. I just pushed a (rather horrible) proof to branch#RD_comap

Alex J. Best (Sep 07 2022 at 17:09):

@Rémy Degenne Did you PR your proof of subtype.volume_univ somewhere? It would be great to have this :smile:

Rémy Degenne (Sep 07 2022 at 20:06):

No I didn't. I was not clear at the time but when I wrote that I pushed stuff to a branch I meant "feel free to take from that branch anything you could need and PR it". I'll try to clean and PR that code next week-end

Alex J. Best (Sep 09 2022 at 12:19):

Ok I did a little golfing, I'd still like to simplify the long proofs a bit (involving lemma todo for instance).


Last updated: Dec 20 2023 at 11:08 UTC