A Kostant root subgroup is a closed copy of the additive group #
Let a Kostant integral form act on a rational representation, preserving an integral lattice M
with finite basis b. A nilpotent root vector eᵢ then gives the scheme morphism
xᵢ : 𝔾ₐ → GLₙ of RootSubgroup.Scheme.Basic. A pinning of a Chevalley--Demazure group needs
more than this morphism: the root subgroup has to be a closed subgroup scheme, and it has to be
a faithful copy of 𝔾ₐ, so that xᵢ(t) determines t.
Both follow from one extra hypothesis on ρ, M and b, a root step: a pair of basis indices
r, s together with a scalar c : ℤ such that
ρ(eᵢ) (b s) = c • b r and ρ(eᵢ) (ρ(eᵢ) (b s)) = 0,
with c a unit. The coordinate expansion below is unconditional; the faithfulness and
closed-immersion results carry these three assumptions as hypotheses. These assumptions are not
derived here from the Kostant form, from the representation, or from the lattice.
The second equation truncates the divided-power exponential in that matrix column, so the
(r, s) entry of xᵢ(t) is exactly c t rather than a polynomial of higher degree. Consequently
the coordinate Hopf-algebra morphism O(GLₙ) → O(𝔾ₐ) hits the polynomial generator, hence is
surjective, and xᵢ is a closed immersion.
A root step is not a normalization that could be arranged by rescaling the basis: it says that the
column of eᵢ at b s is a single basis vector with unit coefficient. For the adjoint
representation on a Chevalley lattice of a simply laced type the intended witness is s the index
of a root vector e_β with β + αᵢ a root and β + 2αᵢ not a root; supplying such a witness is
left to the caller in the general construction.
Main declarations #
TauCeti.UniversalEnvelopingAlgebra.kostantRootGeneratorIntMatrix: the represented generator in an integral lattice basis, shared with the matrix construction ofTauCeti.Algebra.Lie.Symplectic.StandardCarrier.AlternatingForm.TauCeti.UniversalEnvelopingAlgebra.repr_kostantRootSubgroupPoints_baseChange: the coordinates of a root-subgroup point on a base-changed basis vector.TauCeti.UniversalEnvelopingAlgebra.repr_kostantRootSubgroupPoints_of_isRootStep: a root step makes one coordinate equal to the parameter, scaled by the unitc.TauCeti.UniversalEnvelopingAlgebra.kostantRootSubgroupMatrix_apply_of_isRootStep: the same statement read as a matrix entry.TauCeti.UniversalEnvelopingAlgebra.kostantRootSubgroupPoints_apply_baseChange_basis_of_action: the class-two exponential formula for an operator with one nonzero basis column.TauCeti.UniversalEnvelopingAlgebra.kostantRootSubgroupMatrix_eq_transvectionUnit_of_action: a class-two root operator with one nonzero basis column exponentiates to a transvection.TauCeti.UniversalEnvelopingAlgebra.kostantRootSubgroupMatrix_eq_sum: a nilpotent root operator exponentiates to the finite sum of its integral divided-power matrices.TauCeti.UniversalEnvelopingAlgebra.kostantRootSubgroupMatrix_eq_one_add_smuland, in the same namespace,map_genericMatrix_kostantRootSubgroupCoordinateMap_eq_one_add_smul: without the single-column hypothesis a square-zero root operator still exponentiates to1 + t X, on a point and on the generic matrix respectively.TauCeti.UniversalEnvelopingAlgebra.map_genericMatrix_eq_kostantRootSubgroupMatrix: the universal additive-group point identifies the image of the generic matrix with the represented root-subgroup matrix.TauCeti.UniversalEnvelopingAlgebra.kostantRootSubgroupPoints_injective: the root subgroup is faithfully parametrized by𝔾ₐ.TauCeti.UniversalEnvelopingAlgebra.kostantRootSubgroupCoordinateMap_surjective: the coordinate Hopf-algebra morphism of the root subgroup is surjective.TauCeti.UniversalEnvelopingAlgebra.isClosedImmersion_kostantRootSubgroup: the root subgroup𝔾ₐ → GLₙis a closed immersion.TauCeti.UniversalEnvelopingAlgebra.mono_kostantRootSubgroup: it is a monomorphism.TauCeti.UniversalEnvelopingAlgebra.kostantRootSubgroupClosedSubgroup: the resulting closed subgroup scheme ofGLₙ.TauCeti.UniversalEnvelopingAlgebra.coe_kostantRootSubgroupClosedSubgroup: that closed subgroup is the subobject represented by the root-subgroup morphism.
References #
- J. E. Humphreys, Introduction to Lie Algebras and Representation Theory, §§26--27.
- R. W. Carter, Simple Groups of Lie Type, §4.4.
- J. C. Jantzen, Representations of Algebraic Groups, II.1.
The integral matrix of a represented root generator in an invariant lattice basis.
Equations
- TauCeti.UniversalEnvelopingAlgebra.kostantRootGeneratorIntMatrix e h ρ M hM i b = b.toMatrix fun (s : η) => ⟨(ρ ((UniversalEnvelopingAlgebra.ι ℚ) (e i))) ↑(b s), ⋯⟩
Instances For
A represented root generator acts on each lattice basis vector by its integral matrix column.
If a root generator takes the lattice basis vector b a to c • b a', then the a-th
column of its integral matrix is supported at a', with entry c there.
The coordinates of a root-subgroup point on a base-changed basis vector: the r-th
coordinate of xᵢ(t) (1 ⊗ b s) is the divided-power polynomial in t whose coefficients are the
r-th coordinates of the integral divided powers of b s.
The pinning coordinate of a root subgroup. At a root step, the r-th coordinate of
xᵢ(t) (1 ⊗ b s) is the parameter itself, scaled by the unit c. Every higher divided power
vanishes on this basis vector, so no higher power of the parameter appears, and the zeroth one
contributes b s, which has no r-th coordinate because r ≠ s.
The root subgroup is a faithful copy of 𝔾ₐ. Distinct parameters give distinct
automorphisms of the base-changed lattice.
At a root step, the (r, s) entry of the root-subgroup matrix is the parameter, scaled by
the unit c.
A class-two root operator with one nonzero basis column acts by adding the parameter times that column after base change.
A class-two root operator that sends one basis vector to another and kills all remaining basis vectors exponentiates to the corresponding elementary transvection.
At a root step, the coordinate morphism of the root subgroup sends the (r, s) generic matrix
coordinate to the polynomial generator, scaled by the unit c.
The coordinate morphism of a root subgroup is surjective. Its image is a subalgebra of
the polynomial coordinate algebra of 𝔾ₐ containing the generator, hence everything.
A Kostant root subgroup is a closed immersion. The one-parameter subgroup
xᵢ : 𝔾ₐ → GLₙ identifies 𝔾ₐ with a closed subscheme of GLₙ over ℤ.
A root subgroup is a monomorphism of group schemes over ℤ.
The root subgroup as a closed subgroup scheme of GLₙ. This is the subgroup U_αᵢ that a
pinning carries: a closed subgroup scheme which the root-subgroup morphism identifies with 𝔾ₐ.
Equations
- One or more equations did not get rendered due to their size.
Instances For
The root subgroup presents its own closed subgroup. The subobject underlying
kostantRootSubgroupClosedSubgroup is the one represented by kostantRootSubgroup itself, so a
consumer never has to unfold the bundled definition; the inclusion arrow agrees with
kostantRootSubgroup up to CategoryTheory.Subobject.underlyingIso.
The zeroth divided power has the identity matrix in an integral lattice basis.
A divided power has the prescribed matrix in an integral lattice basis.
A nilpotent root-subgroup matrix is the finite divided-power sum. If X k is the
integral matrix of the kth divided power for every k < d, and the nilpotency class is at most
d, then evaluation at t is ∑ k < d, t ^ k • X k.
The matrix of a square-zero root subgroup is 1 + t X. When the root operator squares to
zero its divided-power exponential stops after the linear term, so the root-subgroup matrix at
parameter t is the identity plus t times the integral matrix X of the operator itself.
Unlike TauCeti.UniversalEnvelopingAlgebra.kostantRootSubgroupMatrix_eq_transvectionUnit_of_action
this does not assume the operator has a single nonzero basis column, so it also covers operators
with several nonzero columns, such as the nonfinal root generators of type C.
The generic matrix of a root subgroup is its matrix at the universal point of 𝔾ₐ. The
identity of the additive coordinate algebra is a point of 𝔾ₐ over that algebra, and the
root-subgroup matrix there is the image of the generic matrix of GL N under the root-subgroup
coordinate morphism. A matrix formula proved at every algebra-valued point is read on the
coordinate morphism through this equation, which is the only place the universal point is
handled.
The generic matrix of a square-zero root subgroup is 1 + t X. This is
TauCeti.UniversalEnvelopingAlgebra.kostantRootSubgroupMatrix_eq_one_add_smul read on the
coordinate morphism rather than on a point: the entries of the generic matrix of GL N are carried
to those of 1 + t X for the parameter t of the universal point of 𝔾ₐ. A consumer that has to
check a matrix equation on every algebra-valued point at once evaluates it here instead.