Documentation

TauCeti.Algebra.Lie.HighestWeight.Character

The formal character of a highest weight module against the Weyl denominator #

Let L be a finite-dimensional Lie algebra with non-degenerate Killing form over an algebraically closed field of characteristic zero, let H be a Cartan subalgebra and b a base of its root system. The Weyl character formula is the identity ch L(λ) · Δ = N(λ) in the integral group algebra ℤ[Module.Dual K H], between the formal character of the irreducible module of highest weight λ, the Weyl denominator Δ = ∏_{α>0}(1 - e^{-α}) and the Weyl numerator N(λ) = ∑_w sgn(w) e^{w ⬝ λ}. This file establishes the three properties of the left-hand side that the formula is read off from.

The first is universal: for any finite-dimensional module M the product ch M · Δ is alternating for the dot action (TauCeti.isDotAlternating_formalCharacter_mul_weylDenominator), because the character is invariant for the linear action (TauCeti.isWeylInvariant_formalCharacter, which is the Weyl invariance of weight multiplicities, proved in TauCeti/Algebra/Lie/Weights/WeylInvariance.lean directly from the rank-one theory) and the denominator is alternating for the dot action.

The other two locate the product in the weight order, and they are what makes the alternating element ch M · Δ a specific one. If M is generated by a highest weight vector of weight lam, then ch M · Δ is supported in lam - Q⁺ (TauCeti.sub_mem_posRootCone_of_coeff_formalCharacter_mul_weylDenominator_ne_zero) and its coefficient at lam itself is 1 (coeff_formalCharacter_mul_weylDenominator_eq_one_of_isHighestWeightVector_of_lieSpan_eq_top): the top weight space of a highest weight module is a line and the constant term of Δ is 1, and the positive root cone is pointed, so lam admits only the one decomposition.

Together these are the input to the alternation step of the character formula: ch M · Δ is an alternating element, supported in lam - Q⁺, whose coefficient at lam is 1. Being alternating and supported in lam - Q⁺ already pins the element down completely once its coefficients at the dominant integral weights are known, and the last result below says so. Identifying it with the Weyl numerator N(lam) therefore comes down to a single further statement, which is not proved here: that no dominant integral weight other than lam carries a nonzero coefficient.

The determination is read off from TauCeti.IsDotAlternating.eq_of_forall_coeff_dominantIntegral_eq rather than from the chamber statement TauCeti.IsDotAlternating.eq_of_coeff_openDotDominantChamber_eq, whose fundamental domain is cut out by inequalities and so needs a linear order on the coefficient ring compatible with its ring structure: the ring here is an algebraically closed field, which carries no such order.

Main results #

References #

This builds towards the Weyl character formula of Layer 6 ("the Weyl character, dimension, and Kostant formulas") of TauCetiRoadmap/RepresentationTheory/LieHighestWeight/README.md, whose proof route is fixed there as "defining Verma characters, proving the Weyl denominator identity combinatorially, and concluding by Weyl alternation".

The character of any finite-dimensional module #

The formal character of a finite-dimensional module is Weyl-invariant: the multiplicity of a weight is unchanged by the linear action of the Weyl group. This is TauCeti.formalCharacter_coeff_weylGroup_smul, packaged as the predicate.

The formal character of a finite-dimensional module times the Weyl denominator is alternating for the dot action.

This holds for every finite-dimensional module, with no highest weight hypothesis, and it is the hypothesis TauCeti.IsDotAlternating.eq_weylNumerator consumes, which is what reduces the character formula to a statement about the coefficients of the product on a fundamental domain of the dot action.

The character of a highest weight module #

The formal character of a highest weight module is supported in lam - Q⁺. This is TauCeti.sub_mem_posRootCone_of_genWeightSpace_ne_bot_of_isHighestWeightVector_of_lieSpan_eq_top, read on the coefficients of the character.

The formal character of a highest weight module has coefficient 1 at its highest weight: the top weight space is the line spanned by the generator, by TauCeti.genWeightSpace_eq_span_singleton_of_isHighestWeightVector_of_lieSpan_eq_top.

The formal character of a highest weight module times the Weyl denominator is supported in lam - Q⁺. Both factors are supported downwards from their own top weight, and the positive root cone is closed under addition.

The formal character of a highest weight module times the Weyl denominator has coefficient 1 at the highest weight.

The only way to write lam as a weight of M plus a weight of the denominator is lam + 0: the two summands are lam minus an element of the positive root cone and minus an element of that cone, and the cone is pointed. The surviving term is the product of the two top coefficients, both of which are 1.

Determination by the dominant integral coefficients #

Every simple coroot takes an integer value on a weight below lam. The difference lam - chi lies in Q⁺, where the coroot functionals take integer values (TauCeti.exists_intCast_eq_coroot'_of_mem_posRootCone), and lam itself takes natural values on the simple coroots, being the weight of a highest weight vector in a finite-dimensional module.

This is the integrality that TauCeti.IsDotAlternating.eq_of_forall_coeff_dominantIntegral_eq consumes, and it applies to ch M · Δ through TauCeti.sub_mem_posRootCone_of_coeff_formalCharacter_mul_weylDenominator_ne_zero.

The product of the formal character of a highest weight module with the Weyl denominator is determined by its coefficients at the dominant integral weights.

This is what reduces the Weyl character formula ch L(lam) · Δ = N(lam) to a statement about the dominant integral weights alone: the two sides are alternating for the dot action and supported in lam - Q⁺, so TauCeti.IsDotAlternating.eq_of_forall_coeff_dominantIntegral_eq identifies them as soon as their dominant integral coefficients agree. The two hypotheses on the comparison element g are exactly the two properties of ch M · Δ proved above, so the statement is symmetric in the two sides; beyond the dot alternation hgalt, no Weyl-orbit and no Weyl-group finiteness hypothesis is asked of g. Its integrality on the simple coroots is not asked either: it already follows from hgcone, by TauCeti.exists_intCast_eq_coroot'_of_sub_mem_posRootCone.