feat: root space decomposition of ideals of Lie algebras#35326
feat: root space decomposition of ideals of Lie algebras#35326jano-wol wants to merge 12 commits intoleanprover-community:masterfrom
Conversation
PR summary eb5df47f87Import changes for modified filesNo significant changes to the import graph Import changes for all files
|
| /-- In a Lie algebra with non-degenerate Killing form, root spaces are one-dimensional, so a Lie | ||
| ideal either contains a root space entirely or intersects it trivially. -/ | ||
| lemma rootSpace_le_or_disjoint (I : LieIdeal K L) (α : Weight K H L) (hα : α.IsNonZero) : | ||
| (rootSpace H α).toSubmodule ≤ I.toSubmodule ∨ | ||
| I.toSubmodule ⊓ (rootSpace H α).toSubmodule = ⊥ := by | ||
| by_cases h : I.toSubmodule ⊓ (rootSpace H α).toSubmodule = ⊥ | ||
| · exact .inr h | ||
| · exact .inl (inf_eq_right.mp (Submodule.eq_of_le_of_finrank_le inf_le_right | ||
| ((finrank_rootSpace_eq_one α hα).symm ▸ Submodule.one_le_finrank_iff.mpr h))) |
There was a problem hiding this comment.
I'm not convinced by this lemma. It's really a statement about atoms of lattices and submodules of rank 1. I think the result about atoms is already covered since we have things like:
example {α : Type*} [CompleteLattice α] (a b : α) (ha : IsAtom a) :
a ≤ b ∨ Disjoint a b := by
rw [← ha.not_le_iff_disjoint]
tautobut after a quick look I seem not to be able to find the result that a submodule of finrank 1 is an atom (over a field). So that might be worth adding if you need it to use the lemmas about IsAtom as above.
There was a problem hiding this comment.
I decommissioned rootSpace_le_or_disjoint and added a lemma on finrank 1 -> IsAtom. Please check, I hope it is good now!
Root space decomposition of ideals of Lie algebras
PR proves that in an IsKilling Lie algebra, every Lie ideal decomposes as its intersection with the Cartan subalgebra plus a sum of root spaces. This result is needed for two larger proofs: