Zulip Chat Archive
Stream: Is there code for X?
Topic: Joint smoothness of dslope
Geoffrey Irving (Sep 23 2025 at 21:41):
I believe we have
/-- `dslope f` is jointly smooth in both variables -/
lemma ContDiffAt.dslope {𝕜 E : Type} [NontriviallyNormedField 𝕜]
[NormedAddCommGroup E] [NormedSpace 𝕜 E] {f : 𝕜 → E} {p : 𝕜 × 𝕜} {n m : ℕ}
(d1 : ContDiffAt 𝕜 n f p.1) (d2 : ContDiffAt 𝕜 n f p.2) (le : m + 1 ≤ n) :
ContDiffAt 𝕜 m (fun p : 𝕜 × 𝕜 ↦ dslope f p.1 p.2) p := sorry
but I don't see a way to prove it that isn't a horrible mess: the lemmas we have about dslope don't go up very far, and the explicit derivatives are annoying. Is there a clean route via existing Mathlib?
(Aside: possibly one needs [CharZero 𝕜] for this, I haven't got that far into the details.)
Last updated: Dec 20 2025 at 21:32 UTC