Zulip Chat Archive

Stream: mathlib4

Topic: Stabilizer of a finset


Yaël Dillies (Jan 16 2024 at 10:31):

In LeanCamCombi, @Mantas Baksys and I have a proof of Kneser's addition theorem.

Yaël Dillies (Jan 16 2024 at 10:32):

The statement is about pointwise addition of sets. If A,BA, B are nonempty sets in an abelian group and HH is the stabilizer of A+BA + B (which is finite since the stabilizer of a nonempty finite set XX is contained in XXX - X, hence is finite itself), then it says A+H+B+HA+B+H|A + H| + |B + H| ≤ |A + B| + |H|

Yaël Dillies (Jan 16 2024 at 10:33):

This is useful (and in fact very powerful) because it gives a lower bound on A+B|A + B| in terms of H|H|.

Yaël Dillies (Jan 16 2024 at 10:36):

Now the question is how this should be stated in Lean. Currently, I use docs#Finset.card for all the cardinalities in the statement and therefore must define HH as a finset (called Finset.addStab in LeanCamCombi). Since the stabilizer of is the full group (which might be infinite), I define ∅.addStab = ∅ and as a bonus my statement is now true without the non-emptiness assumption on AA and BB.

Yaël Dillies (Jan 16 2024 at 10:37):

However, PFR has recently had some success using docs#Nat.card on subgroups, so I would like your opinion on whether you think Finset.addStab is worth having to keep the entire Kneser statement in the finset world.


Last updated: May 02 2025 at 03:31 UTC