Toric

4 Scheme theory

4.1 Correspondence between affine group schemes and Hopf algebras

We want to show that affine group schemes correspond to Hopf algebras. We must decide what this means mathematically.

We choose to interpret this as lifting Spec to a fully faithful functor from Hopf algebras to group schemes, with essential image affine group schemes.

An alternative would have been to lift the Gamma-Spec adjunction to an adjunction between Hopf algebras and affine group schemes. This is unfortunately much harder to do over an arbitrary scheme, so we leave this as future work.

4.1.1 Spec of an algebra

Definition 4.1.1 Spec as a functor on algebras
#

Spec is a contravariant functor from the category of \(R\)-algebras to the category of schemes over \(\operatorname{Spec}_R\).

Proposition 4.1.2 Spec as a functor on algebras is fully faithful

Spec is a fully faithful contravariant functor from the category of \(R\)-algebras to the category of schemes over \(\operatorname{Spec}_R\), preserving all limits.

Proof

\(\operatorname{Spec}: \operatorname{Ring}\to \operatorname{Sch}\) is a fully faithful contravariant functor which preserves all limits, hence so is \(\operatorname{Spec}: \operatorname{Ring}_R \to \operatorname{Sch}_{\operatorname{Spec}R}\) by Proposition 1.1.2 (alternatively, by Proposition 1.1.1).

4.1.2 Spec of a bialgebra

Definition 4.1.3 Spec as a functor on bialgebras
#

Spec is a contravariant functor from the category of \(R\)-bialgebras to the category of monoid schemes over \(\operatorname{Spec}_R\).

Proposition 4.1.4 Spec as a functor on bialgebras is fully faithful

Spec is a fully faithful contravariant functor from the category of \(R\)-bialgebras to the category of monoid schemes over \(\operatorname{Spec}_R\).

Proof

\(\operatorname{Spec}: \operatorname{Ring}_R \to \operatorname{Sch}_{\operatorname{Spec}R}\) is a fully faithful contravariant functor preserving all limits according to Proposition 4.1.1, therefore \(\operatorname{Spec}: \operatorname{Bialg}_R \to \operatorname{GrpSch}_{\operatorname{Spec}R}\) too is fully faithful according to 1.2.1.

Proposition 4.1.5 Spec sends cocommutative bialgebras to commutative monoid schemes

If \(A\) is a cocommutative bialgebra over \(R\), then \(\operatorname{Spec}A\) is a commutative monoid scheme.

Proof

Diagrams are the same up to identifying \(\operatorname{Spec}(A \otimes A)\) with \(\operatorname{Spec}A \otimes \operatorname{Spec}A\).

4.1.3 Spec of a Hopf algebra

Definition 4.1.6 Spec as a functor on Hopf algebras
#

Spec is a contravariant functor from the category of \(R\)-Hopf algebras to the category of group schemes over \(\operatorname{Spec}_R\).

Proposition 4.1.7 Spec as a functor on Hopf algebras is fully faithful

Spec is a fully faithful contravariant functor from the category of \(R\)-Hopf algebras to the category of group schemes over \(\operatorname{Spec}_R\).

Proof

\(\operatorname{Spec}: \operatorname{Ring}_R \to \operatorname{Sch}_{\operatorname{Spec}R}\) is a fully faithful contravariant functor preserving all limits according to Proposition 4.1.1, therefore \(\operatorname{Spec}: \operatorname{Hopf}_R \to \operatorname{GrpSch}_{\operatorname{Spec}R}\) too is fully faithful according to 1.2.1.

4.1.4 Essential image of Spec on Hopf algebras

Proposition 4.1.8 Essential image of Spec on algebras
#

The essential image of \(\operatorname{Spec}: \operatorname{Ring}_R \to \operatorname{Sch}_{\operatorname{Spec}R}\) is precisely affine schemes over \(\operatorname{Spec}R\).

Proof

Direct consequence of Proposition 1.1.3.

Proposition 4.1.9 Essential image of Spec on bialgebras

The essential image of \(\operatorname{Spec}: \operatorname{Bialg}_R \to \operatorname{GrpSch}_{\operatorname{Spec}R}\) is precisely affine monoid schemes over \(\operatorname{Spec}R\).

Proof

Direct consequence of Propositions 1.2.3 and 4.1.8.

Proposition 4.1.10 Essential image of Spec on Hopf algebras

The essential image of \(\operatorname{Spec}: \operatorname{Hopf}_R \to \operatorname{GrpSch}_{\operatorname{Spec}R}\) is precisely affine group schemes over \(\operatorname{Spec}R\).

Proof

Direct consequence of Propositions 1.2.3 and 4.1.8.

4.2 Diagonalisable groups

Definition 4.2.1 The diagonalisable group scheme functor
#

Let \(G\) be a commutative group and \(S\) a base scheme. The diagonalisable group scheme \(D_S(G)\) is defined as the base-change of \(\operatorname{Spec}{\mathbb Z}[G]\) to \(S\). For a commutative ring \(R\), we write \(D_R(G) := D_{\operatorname{Spec}R}(G)\).

Definition 4.2.2 Diagonalisable group schemes

An algebraic group \(G\) over \(\operatorname{Spec}R\) is diagonalisable if it is isomorphic to \(D_R(G)\) for some commutative group \(G\).

Lemma 4.2.3 The diagonalisable group scheme torus over \(\operatorname{Spec}R\)

Let \(R\) be a commutative ring and \(M\) an abelian monoid. Then \(D_R(M)\) is isomorphic to \(\operatorname{Spec}R[M]\).

Proof

Ask any toddler on the street.

An algebraic group \(G\) over a field \(k\) is diagonalizable if and only if \(\Gamma (G)\) is spanned by its group-like elements.

Proof

See Theorem 12.8 in [ 2 ] .

Theorem 4.2.5
#

Let \(R\) be a domain. The functor \(D_R(G) := G \rightsquigarrow \operatorname{Spec}R[G]\) from the category of groups to the category of group schemes over \(\operatorname{Spec}R\) is fully faithful.

Proof

Compose Propositions 4.1.7 and 2.3.31.

Also see Theorem 12.9(a) in [ 2 ] . See SGA III Exposé VIII for a proof that works for \(R\) an arbitrary commutative ring.

Proposition 4.2.6 Morphisms between diagonalisable group schemes are affine

Let \(S\) be a scheme. Let \(M, N\) be commutative monoids and \(f : M \to N\) a monoid hom. Then the map \(D_S(f) : D_S(N) \to D_S(M)\) is affine.

Proof

\(\operatorname{Spec}f : \operatorname{Spec}{\mathbb Z}[N] \to \operatorname{Spec}{\mathbb Z}[M]\) is affine, since it’s a morphism of affine schemes. Therefore \(D_S(f)\) is affine, as affine morphisms are preserved under base change.

Proposition 4.2.7 Closed embeddings between diagonalisable group schemes

Let \(S\) be a scheme. Let \(M, N\) be commutative monoids and \(f : M \to N\) a surjective monoid hom. Then the map \(D_S(f) : D_S(N) \to D_S(M)\) is a closed embedding.

Proof

Since \(f\) is surjective, the corresponding map \(\hat f : {\mathbb Z}[M] \to {\mathbb Z}[N]\) is surjective too. Hence \(\operatorname{Spec}\hat f : \operatorname{Spec}{\mathbb Z}[N] \to \operatorname{Spec}{\mathbb Z}[M]\) is a closed embedding. Therefore \(D_S(f)\) is a closed embedding, as closed embeddings are preserved under base change.

Proposition 4.2.8 Faithfully flat morphisms between diagonalisable group schemes

Let \(S\) be a scheme. Let \(G, H\) be abelian groups and \(f : G \to H\) an injective group hom. Then the map \(D_S(f) : D_S(H) \to D_S(G)\) is faithfully flat.

Proof

Since \(f\) is injective, \({\mathbb Z}[H]\) is a free module over \({\mathbb Z}[G]\) by Proposition ??, hence the map \({\mathbb Z}[G] \to {\mathbb Z}[H]\) is faithfully flat and so is \(\operatorname{Spec}{\mathbb Z}[H] \to \operatorname{Spec}{\mathbb Z}[G]\). Therefore \(D_S(f)\) is faithfully flat, as faithfully flat morphisms are preserved under base change.

Proposition 4.2.9 Faithfully flat morphisms between diagonalisable group schemes

Let \(S\) be a scheme. Let \(G, H\) be abelian groups and \(f : G \to H\) an injective group hom. Then the map \(D_S(f) : D_S(H) \to D_S(G)\) is faithfully flat.

Proof

Since \(f\) is injective, \({\mathbb Z}[H]\) is a free module over \({\mathbb Z}[G]\) by Proposition 2.3.10, hence the map \({\mathbb Z}[G] \to {\mathbb Z}[H]\) is faithfully flat and so is \(\operatorname{Spec}{\mathbb Z}[H] \to \operatorname{Spec}{\mathbb Z}[G]\). Therefore \(D_S(f)\) is faithfully flat, as faithfully flat morphisms are preserved under base change.

Proposition 4.2.10 A subgroup of a diagonalisable group scheme is a diagonalisable group scheme

Let \(R\) be a domain. Let \(G\) be an abelian group. If \(H\) is a closed subgroup of \(D_R(G)\), then \(H\) is a diagonalisable group scheme.

Proof

\(H\) is a closed subscheme of an affine scheme, hence it is affine. By Proposition 4.1.10, write \(H = \operatorname{Spec}A\) where \(A\) is a \(R\)-Hopf algebra. The closed subgroup embedding \(H \hookrightarrow D_R(G)\) becomes a surjective bialgebra hom \(R[G] \to A\) by Propositions 4.2.3 and 4.1.7. By Proposition 2.3.25, \(A\) is a diagonalisable bialgebra and therefore \(H\) is a diagonalisable group scheme.

Proposition 4.2.11 Diagonalisable group scheme of a torsion group is disconnected

Let \(G\) be an abelian group with an element of torsion \(n\). Let \(R\) be a commutative ring with \(n\) invertible. Then \(D_R(G)\) is disconnected.

Proof

Say \(x \in G\) is such that \(x^n = 1\). Then

\[ e : R[G] := \frac1n \sum _{i = 0}^n x^i \]

is such that \(e ^2 = e\). We are done by Proposition 4.2.3.

4.3 The torus

Definition 4.3.1 The split torus

The split torus \({\mathbb G}_m^n\) over a scheme \(S\) is the pullback of \(\operatorname{Spec}{\mathbb Z}[x_1^{\pm 1}, \dotsc , x_n^{\pm 1}]\) along the unique map \(S \to \operatorname{Spec}{\mathbb Z}\).

Lemma 4.3.2 Diag is a group isomorphism on hom sets

Let \(R\) be a domain. The functor \(G \rightsquigarrow \operatorname{Spec}R[G]\) from the category of groups to the category of group schemes over \(\operatorname{Spec}R\) is a group isomorphism on hom sets.

Proof

Toddlers and streets by Lemmas 4.2.5 and 1.2.4.

Definition 4.3.3 Characters of a group scheme
#

For a group scheme \(G\) over \(S\), the character lattice of \(G\) is

\[ X(G) := \operatorname{Hom}_{\mathsf{GrpSch}_S}(G, {\mathbb G}_m). \]

An element of \(X(G)\) is called a character.

Definition 4.3.4 Cocharacters of a group scheme
#

For a group scheme \(G\) over \(S\), the cocharacter lattice of \(G\) is

\[ X^*(G) := \operatorname{Hom}_{\mathsf{GrpSch}_S}({\mathbb G}_m, G). \]

An element of \(X^*(G)\) is called a cocharacter or one-parameter subgroup.

Proposition 4.3.5 Character lattice of a diagonalisable group scheme

Let \(R\) be a domain and \(G\) be a commutative group. Then \(X(\operatorname{Spec}R[G]) = G\).

Proof

By Propositions 4.2.3 and 4.2.5 in turn, we have

\[ X(G) = \operatorname{Hom}_{\mathsf{GrpSch}}(G, {\mathbb G}_m) = \operatorname{Hom}(k[{\mathbb Z}], k[G]) = \operatorname{Hom}({\mathbb Z}, G) = G. \]
Proposition 4.3.6 Cocharacter lattice of a diagonalisable group scheme

Let \(R\) be a domain and \(G\) be a commutative group. Then \(X^*(\operatorname{Spec}R[G]) = \operatorname{Hom}(G, {\mathbb Z})\).

Proof

By Propositions 4.2.3 and 4.2.5 in turn, we have

\[ X^*(G) = \operatorname{Hom}_{\mathsf{GrpSch}}({\mathbb G}_m, G) = \operatorname{Hom}(k[G], k[{\mathbb Z}]) = \operatorname{Hom}(G, {\mathbb Z}). \]
Proposition 4.3.7 Character lattice of the torus

Let \(G\) be a torus of dimension \(n\) over a domain \(R\). Then \(X(G) = {\mathbb Z}^n\).

Proof

Immediate from Propositions 4.2.3 and 4.3.5.

Proposition 4.3.8 Cocharacter lattice of the torus

Let \(G\) be a torus of dimension \(n\) over a domain \(R\). Then \(X^*(G) = \operatorname{Hom}({\mathbb Z}^n, {\mathbb Z})\).

Proof

Immediate from Propositions 4.2.3 and 4.3.6.

Definition 4.3.9 The character-cocharacter pairing

Let \(R\) be a domain and \(G\) a group scheme over \(\operatorname{Spec}R\). Then there is a \({\mathbb Z}\)-valued perfect pairing between \(X(G)\) and \(X^*(G)\).

Proposition 4.3.10 The character-cocharacter pairing is perfect

The character-cocharacter pairing on a torus is perfect.

Proof

Transfer everything across the isos \(X({\mathbb G}_m^n) = {\mathbb Z}^n, X*({\mathbb G}_m^n) = \operatorname{Hom}({\mathbb Z}^n, {\mathbb Z})\).

Proposition 4.3.11 The image of a torus is a torus

Let \(R\) be a domain. Let \(T\) be a split torus over \(R\). Let \(G\) be a diagonalisable group scheme over \(R\) and let \(\phi : T \to G\) be a homomorphism. Then the (scheme theoretic) image of \(\phi \) is a split torus over \(R\) and the maps

\[ T \xrightarrow {\hat{\phi }} \operatorname {im}\phi \xrightarrow {\iota } G \]

are group homomorphisms, and \(\hat{\phi }\) is fpqc. Furthermore, if \(T = D_R(H), G = D_R(I), \phi = D_R(f)\) for \(H\) a finitely generated free abelian group, \(I\) an abelian group, \(f : I \to H\) a group hom, then \(\operatorname {im} \phi \cong D_R(\operatorname {im}(f))\).

Proof

By fullness of \(D_R\) (Proposition 4.2.5), it’s enough to handle the case where \(T = D_R(H), G = D_R(I), \phi = D_R(f)\) for \(H\) a finitely generated free abelian group, \(I\) an abelian group, \(f : I \to H\) a group hom.

Then \(\operatorname {im}(f)\) is a subgroup of the finitely-generated free abelian group \(H\), hence itself a finitely-generated free abelian group (since free is equivalent to torsion-free for finitely-generated abelian groups, and a subgroup of a torsion-free group is torsion-free).

Proposition 4.3.12 A subgroup of a torus is a torus

Let \(R\) be a commutative ring of characteristic zero. Let \(T\) be a split torus. If \(H \subseteq T\) is a connected closed subgroup, then \(H\) is a split torus.

Proof

By assumption, write \(T \cong D_k[G]\) for \(G\) a free abelian group. By Proposition 4.2.10, \(H\) is a diagonalisable group scheme, say \(H \cong D_k(I)\) for \(I\) an abelian group. Since \(H\) is a closed subscheme, the map \(G \to I\) is surjective, so \(I\) is a finitely-generated abelian group. Since \(H\) is connected, Proposition 4.2.11 says \(I\) is torsion-free, hence free. Thus \(H\) is a split torus.