Documentation

LeanPool.Ado.Algebra.Lie.HighestWeight.CompleteReducibility

Weyl's complete reducibility theorem #

Let L be a finite-dimensional Lie algebra with nondegenerate Killing form over a field of characteristic zero. Every Lie submodule of a finite-dimensional L-module is a direct summand (Ado.exists_isCompl_of_isKilling), so the lattice of Lie submodules is complemented and a finite-dimensional module is a direct sum of irreducibles.

All the formal content of the proof is already in TauCeti/Algebra/Lie/CompleteReducibility.lean, which derives complete reducibility from the single representation-theoretic input Ado.HasInvariantOutsideIrreducible: whenever L carries a finite-dimensional module M into a proper irreducible submodule N and acts nontrivially somewhere on M, the module M has a nonzero invariant vector outside N. Over an algebraically closed field this file supplies that input from the Casimir element of U(L), which is what makes the argument work for a general semisimple L rather than only for sl₂ (TauCeti/Algebra/Lie/Sl2/CompleteReducibility.lean, the rank-one instance of the same reduction). Over an arbitrary field of characteristic zero it is then obtained by descent from an algebraic closure.

The Casimir step #

Two facts about the Casimir element Ω ∈ U(L) do the work.

Being injective on N and having its image inside the proper submodule N, the Casimir operator of M is not surjective, hence — M being finite-dimensional — not injective; a nonzero kernel vector lies outside N, and its brackets lie in N ∩ ker Ω = 0, so it is invariant.

The remaining case is the one where L acts trivially on N. Then the action of L on M composes to zero, and a Lie algebra with nondegenerate Killing form is perfect (Ado.derivedSeries_one_eq_top_of_isKilling), so the action is already zero, contradicting the hypothesis that L acts nontrivially somewhere on M.

Descent from an algebraic closure #

Over a field K of characteristic zero, let A = AlgebraicClosure K. The Killing form of A ⊗[K] L is the extension of that of L, so it is again nondegenerate (Ado.isKilling_baseChange_iff). If L carries M into a proper Lie submodule N, then A ⊗[K] L carries A ⊗[K] M into the proper Lie submodule N.baseChange A, and the algebraically closed case gives an invariant vector of A ⊗[K] M outside N.baseChange A. The invariants of A ⊗[K] M are the extension of the invariants of M (LieModule.maxTrivSubmodule_baseChange), so the invariants of M are not all contained in N.

Main results #

References #

The Casimir element is injective on a nontrivial irreducible #

The Casimir element is injective on a nontrivial finite-dimensional irreducible module. The module is generated by a highest weight vector, whose weight is dominant integral and, the action being nontrivial, nonzero; the Casimir element then acts by a nonzero scalar.

The Casimir step over an algebraically closed field #

Complete reducibility over a field of characteristic zero #

theorem Ado.exists_invariant_notMem_of_isKilling {K : Type u} {L : Type v} [Field K] [CharZero K] [LieRing L] [LieAlgebra K L] [LieAlgebra.IsKilling K L] [FiniteDimensional K L] {M : Type w} [AddCommGroup M] [Module K M] [LieRingModule L M] [LieModule K L M] [FiniteDimensional K M] (N : LieSubmodule K L M) (hN : N ≠ ⊤) (htriv : ∀ (x : L) (m : M), ⁅x, m⁆ ∈ N) :
∃ v ∉ N, ∀ (x : L), ⁅x, v⁆ = 0

An invariant vector outside a proper submodule. If L carries a finite-dimensional module M into a proper Lie submodule N, then M has a nonzero invariant vector outside N.

The input of the formal complete-reducibility argument holds in characteristic zero: if L carries a finite-dimensional module into a proper irreducible Lie submodule N and acts nontrivially somewhere, the module has an invariant vector outside N.

theorem Ado.exists_isCompl_of_isKilling {K : Type u} {L : Type v} [Field K] [CharZero K] [LieRing L] [LieAlgebra K L] [LieAlgebra.IsKilling K L] [FiniteDimensional K L] {M : Type w} [AddCommGroup M] [Module K M] [LieRingModule L M] [LieModule K L M] [FiniteDimensional K M] (N : LieSubmodule K L M) :
∃ (N' : LieSubmodule K L M), IsCompl N N'

Weyl's complete reducibility theorem. Over a field of characteristic zero, every Lie submodule of a finite-dimensional module over a Lie algebra with nondegenerate Killing form has a complement, so the module is a direct sum of irreducibles.

The sl₂ case, Ado.exists_isCompl_of_toLieSubalgebra_eq_top, is the rank-one instance of the same argument, run with the concrete Casimir operator of an sl₂ triple.

The lattice form of Weyl's theorem: the Lie submodules of a finite-dimensional module over a Lie algebra with nondegenerate Killing form form a complemented lattice.

theorem Ado.exists_isInternal_isIrreducible_of_isKilling (K : Type u) (L : Type v) [Field K] [CharZero K] [LieRing L] [LieAlgebra K L] [LieAlgebra.IsKilling K L] [FiniteDimensional K L] (M : Type w) [AddCommGroup M] [Module K M] [LieRingModule L M] [LieModule K L M] [FiniteDimensional K M] :
∃ (k : ℕ) (N : Fin k → LieSubmodule K L M), (DirectSum.IsInternal fun (i : Fin k) => ↑(N i)) ∧ ∀ (i : Fin k), LieModule.IsIrreducible K L ↥(N i)

Weyl's theorem as a decomposition. A finite-dimensional module over a Lie algebra with nondegenerate Killing form, over a field of characteristic zero, is the internal direct sum of a finite family of irreducible Lie submodules.

theorem LieSubmodule.exists_isCompl_of_le_ker {K : Type u} {L : Type v} [Field K] [CharZero K] [LieRing L] [LieAlgebra K L] [FiniteDimensional K L] {M : Type w} [AddCommGroup M] [Module K M] [LieRingModule L M] [LieModule K L M] [FiniteDimensional K M] (N : LieSubmodule K L M) {I : LieIdeal K L} [LieAlgebra.IsKilling K (L ⧸ I)] (hI : I ≤ LieModule.ker K L M) :
∃ (N' : LieSubmodule K L M), IsCompl N N'

Weyl's theorem for an action through a quotient. If an ideal I of L acts trivially on a finite-dimensional module M and the quotient L ⧸ I has nondegenerate Killing form, then every Lie submodule of M has a complement. This applies when L itself is not semisimple, for instance to a module of L on which its solvable radical acts trivially.