Documentation

TauCeti.Algebra.Lie.GeneralLinear.Existence

Every finite-dimensional gl n-module has a dominant highest weight vector #

Let K be a field of characteristic zero and let M be a nonzero finite-dimensional module over gl n K = Matrix n n K. This file proves that M carries a highest weight vector for the matrix unit positive system, and that the weight of any highest weight vector of a finite-dimensional module is dominant integral. Read for an irreducible M, the two statements say that every finite-dimensional irreducible gl n-module has a dominant highest weight, which is the existence half of the classification of the irreducibles of gl n.

The argument #

Existence is Lie's theorem followed by a maximality argument, and both halves are needed. Lie's theorem is applied to the abelian diagonal Cartan subalgebra rather than to the Borel subalgebra of upper triangular matrices, which is the other classical route: a common eigenvector of the Borel is a highest weight vector outright, but that route first needs the Borel to be solvable and its derived subalgebra to contain the raising matrix units, neither of which is available.

The diagonal Cartan subalgebra is abelian, hence solvable, so over an algebraically closed field Lie's theorem (LieModule.exists_nontrivial_weightSpace_of_isSolvable) produces a nonzero simultaneous eigenvector of the diagonal matrix units. Algebraic closure is not a convenience here: over ℝ a simultaneous eigenvector need not exist at all. Neither may one replace this step by the argument used for a semisimple Lie algebra, where the generalized weight spaces are refined to honest eigenspaces once and for all: gl 1 acting on K² by a nilpotent Jordan block is a finite-dimensional module on which the Cartan subalgebra does not act semisimply, so a highest weight vector has to be produced one vector at a time rather than as a whole weight space.

Such an eigenvector need not be annihilated by the raising matrix units, and the second half of the argument moves it up until it is. Bracketing with Eₚq sends an eigenvector of weight μ to an eigenvector of weight μ + εₚ - ε_q, so, writing ht μ for the value of μ on the fixed diagonal matrix ∑ i, -i · Eᵢᵢ, one raising step with p < q increases ht by the positive natural number q - p. All the vectors so produced are eigenvectors of a single operator, the action of that fixed diagonal matrix, with pairwise distinct eigenvalues, so there are only finitely many of them: the set of naturals m for which an eigenvector of ht-value ht μ₀ + m exists is finite. A vector realizing its greatest element is moved by no raising operator, which is exactly a highest weight vector.

Dominance is the rank-one reduction. For p < q the matrix units Eₚq, E_qp and their commutator Eₚₚ - E_qq form an sl₂ triple, and a highest weight vector is a primitive vector for it with eigenvalue μ p - μ q; Mathlib's IsSl2Triple.HasPrimitiveVectorWith.exists_nat then makes that eigenvalue a natural number. Only finite-dimensionality of M is used, so dominance needs neither irreducibility nor an algebraically closed field.

Main results #

References #

This is the first target of the "classification and the named carrier" item of Layer 9 of TauCetiRoadmap/RepresentationTheory/LieHighestWeight/README.md: "Every finite-dimensional gl_n-irreducible has a dominant highest weight".

Simultaneous eigenvectors of the diagonal #

theorem TauCeti.exists_ne_zero_forall_lie_single_self_eq_smul {K : Type u} [Field K] {n : Type v} [Fintype n] [DecidableEq n] {M : Type w} [AddCommGroup M] [Module K M] [LieRingModule (Matrix n n K) M] [LieModule K (Matrix n n K) M] [CharZero K] [IsAlgClosed K] [FiniteDimensional K M] [Nontrivial M] :
∃ (mu : n → K) (v : M), v ≠ 0 ∧ ∀ (i : n), ⁅Matrix.single i i 1, v⁆ = mu i • v

Lie's theorem for the diagonal Cartan subalgebra: over an algebraically closed field of characteristic zero a nonzero finite-dimensional gl n-module has a nonzero simultaneous eigenvector of the diagonal matrix units.

The eigenvector is not asserted to be annihilated by the raising matrix units; that is what TauCeti.exists_isGlHighestWeightVector adds.

theorem TauCeti.lie_single_self_lie_single_eq_smul {K : Type u} [Field K] {n : Type v} [Fintype n] [DecidableEq n] {M : Type w} [AddCommGroup M] [Module K M] [LieRingModule (Matrix n n K) M] [LieModule K (Matrix n n K) M] {mu : n → K} {v : M} (hv : ∀ (i : n), ⁅Matrix.single i i 1, v⁆ = mu i • v) (p q i : n) :

Raising a weight vector. Bracketing a simultaneous eigenvector of the diagonal matrix units of weight μ with the matrix unit Eₚq gives a simultaneous eigenvector of weight μ + εₚ - ε_q, whenever the result is nonzero.

Existence of a highest weight vector #

theorem TauCeti.exists_isGlHighestWeightVector {K : Type u} [Field K] {N : ℕ} {M : Type w} [AddCommGroup M] [Module K M] [LieRingModule (Matrix (Fin N) (Fin N) K) M] [LieModule K (Matrix (Fin N) (Fin N) K) M] [CharZero K] [IsAlgClosed K] [FiniteDimensional K M] [Nontrivial M] :
∃ (mu : Fin N → K) (v : M), IsGlHighestWeightVector mu v

Existence of a highest weight vector for gl n. Over an algebraically closed field of characteristic zero every nonzero finite-dimensional gl N-module has a highest weight vector for the matrix unit positive system.

Neither irreducibility nor a choice of generator is assumed; for irreducible M this is the existence half of the classification of the finite-dimensional irreducibles of gl N.

Dominance of the highest weight #

theorem TauCeti.isSl2Triple_matrix_single {K : Type u} [Field K] [CharZero K] {N : ℕ} (p q : Fin N) (hpq : p ≠ q) :

The sl₂ triple of a pair of indices in gl n. For p ≠ q the matrix units Eₚq and E_qp together with their commutator Eₚₚ - E_qq satisfy the sl₂ relations; this is the rank-one subalgebra of gl n along which dominance is read off.

It is the standard triple TauCeti.isSl2Triple_single of sl n K, pushed forward along the inclusion of sl n K in gl n K.

theorem TauCeti.IsGlHighestWeightVector.isGlDominantIntegral {K : Type u} [Field K] [CharZero K] {N : ℕ} {M : Type w} [AddCommGroup M] [Module K M] [LieRingModule (Matrix (Fin N) (Fin N) K) M] [LieModule K (Matrix (Fin N) (Fin N) K) M] {mu : Fin N → K} {v : M} [FiniteDimensional K M] (hv : IsGlHighestWeightVector mu v) :

The weight of a highest weight vector is dominant integral. Restricting to the sl₂ triple of a pair p < q makes the highest weight vector a primitive vector of eigenvalue μ p - μ q, and the eigenvalue of a primitive vector in a finite-dimensional module is a natural number.

Neither irreducibility of M nor algebraic closure of K is used.

Every nonzero finite-dimensional gl N-module has a dominant highest weight vector. Over an algebraically closed field of characteristic zero, combining the existence of a highest weight vector with the dominance of its weight.

For an irreducible M this is the statement that a finite-dimensional irreducible gl N-module has a dominant highest weight, the first half of the classification of the irreducibles of gl N.