-
Notifications
You must be signed in to change notification settings - Fork 297
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
feat(geometry/manifold): The preimage straightening theorem #10683
base: master
Are you sure you want to change the base?
Conversation
Can you expand in the PR description on what the statement of the theorem should be? For the manifold with boundary case, it is good to keep the following example in mind. Let |
Done! Sure but it should still be a manifold with boundary right? |
No, it wouldn't be a manifold with boundary (any neighborhood of |
Ok sure I understand this example now! But I am confused because this theorem should hold for orbifolds and sometimes you even find a proof directly for manifolds with boundary (like here: http://www.math.toronto.edu/mgualt/courses/MAT1300F-2012/docs/1300-2012-notes-5.pdf)... I do not have the time now to think about what is going wrong but I'll think more about this tonight! |
Oh ok I think I see the problem now. The theorem holds if the value is regular both for |
I am not sure about the title of this PR, but this is just the name we sometimes give to this theorem in Italy.
Theorem:
Let
f:M → N
be map which is smooth atp
, andp
be a regular point off
. Consider chartsφ : U ⊆ M → E
andψ : V ⊆ N → F
aboutp
andf(p)
respectively. Letg = ψ ∘ f ∘ φ⁻¹
. LetK=ker (d g_{φ (p)})
. Then there exists a local homeomorphismϕ: W → F × K ≅ E
whereW ⊆ U ⊆ M
is an open containingp
, such thatψ ∘ f ∘ ϕ⁻¹ : (v,k) : F × K ↦ v : F
.This PR is incomplete but the only two missing lemmas should be straightforward once some API for boundaryless manifolds has been written. In any case, I did not want to write the API yet because I think this theorem holds for manifolds with corners meant to be manifolds locally modeled on a quadrant of the euclidean space but does not hold for the general model with corners we have in Mathlib now. To my knowledge, this could be one of the first times that we try to formalize something which is true for manifolds with corners but that does not extend easily to them, so it should be a good stress test for the current implementation. I might be wrong because I am not familiar with strict differentiability but I am sure that the regular value theorem is true for manifolds with corners in the usual sense but not in this case and this PR is meant to be the first step for that proof, so the above problem holds even if I am wrong.
What should we do? Restrict the current definition of model with corners or give up on this theorem in that case? I believe this is a useful theorem for manifolds with boundary.