Textbook Actions and Invariants of Algebraic Groups Second Edition Ferrer Santos Ebook All Chapter PDF

Actions and invariants of algebraic
groups Second Edition Ferrer Santos

Visit to download the full and correct content document:
https://textbookfull.com/product/actions-and-invariants-of-algebraic-groups-second-ed
ition-ferrer-santos/
More products digital (pdf, epub, mobi) instant
download maybe you interests ...
Actions and Invariants of Algebraic Groups, Second

Edition Walter Ricardo Ferrer Santos
https://textbookfull.com/product/actions-and-invariants-of-
algebraic-groups-second-edition-walter-ricardo-ferrer-santos/
Modernization and Political Actions in the Brazilian

Amazon The City of Barcarena Pará 1st Edition João
Santos Nahum (Auth.)
https://textbookfull.com/product/modernization-and-political-
actions-in-the-brazilian-amazon-the-city-of-barcarena-para-1st-
edition-joao-santos-nahum-auth/
Algebraic Groups The Theory of Group Schemes of Finite

Type over a Field 1st Edition J.S. Milne
https://textbookfull.com/product/algebraic-groups-the-theory-of-
group-schemes-of-finite-type-over-a-field-1st-edition-j-s-
milne-2/
Algebraic Groups The Theory of Group Schemes of Finite

Type over a Field 1st Edition J. S. Milne
https://textbookfull.com/product/algebraic-groups-the-theory-of-
group-schemes-of-finite-type-over-a-field-1st-edition-j-s-milne/
Algebraic Numbers and Algebraic Functions First Edition
Cohn
https://textbookfull.com/product/algebraic-numbers-and-algebraic-
functions-first-edition-cohn/
Theory of Groups and Symmetries Finite groups Lie

groups and Lie Algebras 1st Edition Alexey P. Isaev
https://textbookfull.com/product/theory-of-groups-and-symmetries-
finite-groups-lie-groups-and-lie-algebras-1st-edition-alexey-p-
isaev/
Longitudinal Multivariate Psychology Emilio Ferrer
https://textbookfull.com/product/longitudinal-multivariate-
psychology-emilio-ferrer/
Theory of Groups and Symmetries: Representations of

Groups and Lie Algebras, Applications 1st Edition
Alexey P. Isaev
https://textbookfull.com/product/theory-of-groups-and-symmetries-
representations-of-groups-and-lie-algebras-applications-1st-
edition-alexey-p-isaev/
An Introduction to Quantum and Vassiliev Knot

Invariants David M. Jackson
https://textbookfull.com/product/an-introduction-to-quantum-and-
vassiliev-knot-invariants-david-m-jackson/
Chapter 14
Quotient varieties: basic results
1 Introduction
In this chapter we touch upon some of the more geometric aspects of invariant theory.
The systematic study of what is now called geometric invariant theory was initiated by
D. Mumford (see [121] or its subsequent editions, and the surveys [126] and [134]). But
as Mumford himself mentions in [123] when he refers to some of his results concerning
quotients of projective varieties acted on by reductive groups: “The above [· · · ] is in fact a
natural extension of Hilbert’s own ideas about the ring of invariants, specially as developed
in his last big paper on the subject, Über die vollen Invariantensystemen [71].”
Next we describe briefly the contents of this chapter.
In Section 2 we generalize some of the results obtained in Chapter 7 concerning quotients
by finite groups to quotients by reductive groups: we prove that a reductive group acting
on an affine variety admits a categorical quotient — in fact a semigeometric quotient in the
terminology of [6].
In Section 3 we discuss — still in the situation of a reductive group acting on an affine
variety — the problem of the existence of a stable open subset of the original variety where
the quotient is geometric. In the case of finite groups the categorical quotient is always
geometric, but for general reductive groups this is not so: subtle difficulties appear when
trying to control the way the (closure of the) orbits are located inside the variety (see the
introduction to Chapter 7).
In Section 4 we describe in certain detail the geometric picture corresponding to the
classical case of the general linear group acting by conjugation on the space of all the
matrices, i.e., the problems related to the canonical forms of matrices.
In Section 5 we present a proof of a theorem of M. Rosenlicht ([156]) that guarantees
the generic existence of geometric quotients: we prove that if G is an affine algebraic group
acting regularly on an arbitrary variety X, then there exists a dense G-stable open subset
X0 ⊂ X such that the geometric quotient X0 //G exists.
In Section 61 we consider the geometric version of the construction of the induced
representation. Assume that G is an affine algebraic group, H a closed subgroup and X
an algebraic H-variety. We construct the homogeneous fiber bundle — the denomination
of this object is not uniform, for example, in [162] J.-P. Serre presents this construction
as a particular case of his construction of algebraic fiber spaces (see also Observation 6.2),
whereas A. Bialynicki-Birula in [5] introduces the terminology of induced action; in [144]
and [177] G ∗H X is called a homogeneous fiber bundle, and finally other authors call it
simply induced space. The homogeneous fiber bundle is the natural minimal extension of
the H-variety X to a G-variety that we denote as G ∗H X. The existence of such a quotient
1 The results presented in this section and the next are introduced in the second edition of the book.
333
334 Actions and Invariants of Algebraic Groups
variety is related to the local isotriviality of the canonical projection π : G → G/H — in

observations 6.2 (2) and 6.6 (1) below, we give some references for this concept. To keep
the elementary spirit of the book, we only deal with the case of exact subgroups and affine
varieties.
In Section 7 we obtain further information on the orbits of an action in the case that
the affinized quotient exists.
In Section 8 we return to the case of actions of finite groups, with the intention to
complete the picture concerning properties of the algebra of invariants and the geometry
of the quotient. A natural problem in constructive invariant theory is the following: if G is
an affine algebraic group and V a finite dimensional rational G-module, is there a natural
bound for the degrees of a complete system of invariants of k[V ]? We present a classical
theorem by E. Noether (see [135]) that guarantees that in the case of a finite group whose
order is prime with the characteristic of the field, this bound can taken to be the order of
G.
Another natural problem concerning invariants of groups is the following: if the group
acts linearly on a finite dimensional vector space, are there natural conditions that guarantee
that the algebra of invariants is generated by algebraically independent elements? If G is
a finite group generated by pseudo-reflections, then its algebra of invariants is indeed a
polynomial algebra. This theorem — together with its converse that we will not prove —
is called the Chevalley–Shephard–Todd theorem (see [25] and [168]). In geometric terms, it
guarantees that in the situation of a finite group G generated by reflections, the geometric
quotient An //G ∼= An as algebraic varieties.
Recall that — as it was mentioned in Observation 10.4.4, but without proof — an affine
algebraic group is reductive if and only if it is geometrically reductive. In this chapter we
will use this fact without any distinction talking simply about reductive groups.
2 Actions by reductive groups: the semigeometric quotient

As we mentioned before, in [121] D. Mumford developed his theory in less generality
than is customary today. He considered algebraic varieties acted upon by linearly reductive
groups; in particular he worked with reductive groups in characteristic zero. He also observed
that most of the theory would be valid in arbitrary characteristic if reductive groups were
geometrically reductive — the so-called Mumford conjecture. In [121, Chap. 1] he proves
the existence of the categorical quotient of an affine variety under the action of a linearly
reductive group. The generalization of this result to reductive groups is mainly due to
M. Nagata and appears, for example, in [132] and [134], as well as in the second and third
editions of [121].
In this section we follow the standard literature on the subject; see, for example, [125]
or [134].
If G is a reductive group acting on an affine variety X, then the categorical quotient
π : X → G\X exists, and verifies some additional geometrical properties, similar to the
conditions that characterize the geometric quotient — in this case, the fibers of π are not
necessarily closed orbits, but they will contain a unique closed orbit. These conditions are
the ones satisfied by the semigeometric quotient — this terminology was introduced in [6].
Quotient varieties: basic results 335
Before presenting the existence theorem for quotients by reductive groups (Theorem 2.9),
we collect some results on the semigeometric quotient.
Definition 2.1. Let G be an affine algebraic group acting regularly on a variety X.
A pair (Y, π), where π : X → Y is a morphism of algebraic varieties, is said to be a
semigeometric quotient for the action of G on X if the following conditions are satisfied:
(1) The map π is surjective and constant along the G-orbits.
G
(2) For any open subset V ⊂ Y the map πV] : OY (V ) → OX π −1 (V ) is an isomorphism

of k-algebras.
(3) If W ⊂ X is closed and G-stable, then π(W ) ⊂ Y is closed.
(4) If W, Z ⊂ X are disjoint closed G-stable subsets, then π(W ) ∩ π(Z) = ∅.
Observation 2.2. (1) It is clear that a geometric quotient π : X → Y is semigeometric.
Indeed, conditions (1) and (2) are automatically satisfied: see Definition 7.4.12. Since the
fibers of π consist of a unique orbit, it follows that any G-stable subset is saturated, and
hence (4) is satisfied. Condition (3) follows from Exercise 7.17.
(2) The semigeometric quotient is a categorical quotient; see Theorem 2.5 below. On the
other hand, the semigeometric quotient is not necessarily a geometric quotient, as Example
2.12 below shows.
(3) A categorical quotient is not necessarily semigeometric. Indeed, the action of Ga on A2
given by a · (x, y) = (x, y + ax) for a ∈ Ga and for (x, y) ∈ A2 (see Figure 7.1) is such that
O(x, y) is closed for all (x, y) ∈ A2 , with categorical quotient π : A2 → A1 , π(x, y) = x (see
Example 7.4.7 and Exercise 7.12).
In the case of a semigeometric quotient, besides having some control over the images
of the closed G-invariant subsets of X, we can obtain a certain degree of control over the
images by π of G-stable open sets.
Lemma 2.3. Let G be an affine algebraic group acting regularly on an algebraic variety
X. If (Y, π) is a semigeometric quotient for the action, then the image by π of a saturated
open subset is open.
Proof. If U ⊂ X is an open saturated subset, then X \ U is closed and G-stable,

and we deduce from Definition 2.1 that π(X \ U ) is closed in Y . As U is saturated, then
π(U ) = Y \ π(X \ U ), so that π(U ) is open in Y .
Next we show that any semigeometric quotient (Y, π) is a categorical quotient. In view
of condition (2) in the definition of semigeometric quotient, we are in a situation similar
to the one treated in Theorem 7.4.20, where it was proved that a geometric quotient is
categorical. However, in the present context we have to proceed with more care than for
the geometric quotient, because π is not necessarily an open map.
Lemma 2.4. Let G be an affine algebraic group, X a G-variety and assume that the
quotient π : X → G\X exists. Then for all V ⊂ Y open subset, the pair
semigeometric
V, π −1 is a semigeometric quotient.
π (V )
Proof. We prove that conditions (1) to (4) of Definition 2.1 are satisfied by the pair
π : π −1 (V ) → V .
Conditions (1) and (2) follow from the corresponding properties for X.
(4) Let C, D ⊂ π −1 (V ) be G-stable closed subsets such that there exists y ∈ π(C) ∩
π(D). Then the subsets π −1 (y) ∩ C, π −1 (y) ∩ D ⊂ X are closed and G-stable, and y ∈
π π −1 (y) ∩ C ∩ π π −1 (y) ∩ D . Since (G\X, π) is a semigeometric quotient, we conclude

that π −1 (y) ∩ C ∩ π −1 (y) ∩ D 6= ∅, and hence

π −1 (y) ∩ C ∩ D = π −1 (y) ∩ C ∩ π −1 (y) ∩ D 6= ∅.

Since C, D are closed in π −1 (V ) and π −1 (y) ⊂ π −1 (V ), it follows that C ∩D = π −1 (V )∩

C ∩ D 6= ∅.
The proof of (3) is very similar to the proof of (4) and it is left as an exercise for the
reader (see Exercise 5).
Next we prove that a semigeometric quotient is categorical.
Theorem 2.5. Let G be an affine algebraic group and X a algebraic G-variety. If (Y, π)
is a semigeometric quotient for the action of G on X, then it is a categorical quotient. In
particular, if V ⊂ Y is open, then V, π −1 is a categorical quotient for the action of G
π (V )
on π −1 (V ).
Proof. Let Z be an algebraic variety and f : X → Z a morphism constant along the

G-orbits. We need to complete the diagram below with a morphism fb : Y → Z.
X
f
/Z
>
π
fb
Y
If x1 , x2 ∈ X are such that π(x1 ) = π(x2 ), then

f (x1 ) = f (x2 ). Indeed, if f (x1 ) 6= f (x2 ),
then C = f −1 f (x1 )⊂ X and D = f −1
f (x2 ) ⊂ X are disjoint closed G-stable subsets.
Hence, π f −1 f (x1 ) ∩ π f −1 f (x2 ) = ∅, and this contradicts the fact that π(x1 ) =
π(x2 ). Since π is surjective, there exists a unique (set theoretical) map fb that completes
the above diagram.
We have to check that fb is a morphism. If U ⊂ Z is an open subset, then
fb−1 (U ) = Y \ π X \ π −1 fb−1 (U ) = Y \ π X \ f −1 (U ) .

Since f −1 (U ) is an open G-stable subset of X, we conclude that X \ f −1 (U ) is a closed

G-stable subset, so that fb−1 (U ) is open in Y , i.e., the map fb is continuous.
In order to finish the proof, we need to prove that if U ⊂ Z is an affine open subset, then
G
: fb−1 (U ) → U is a morphism. Consider f ] : OZ (U ) → OX f −1 (U ) ,

the map fb
fb−1 (U )
G G
the isomorphism π ] : OY fb−1 (U ) ∼ = OX π −1 fb−1 (U ) = OX f −1 (U ) , and the

morphism of k-algebras (π ] )−1 ◦ f ] : OZ (U ) → OY fb−1 (U ) . Since U is affine, (π ] )−1 ◦ f ]

induces a morphism gU : fb−1 (U ) → U . As

] ] ]
π ◦ gU = π ◦ (π ] )−1 ◦ f ] = f ] ,
f −1 (U ) f −1 (U )
we deduce that gU ◦ π =f . As π is surjective, fb = gU and hence fb is a

f −1 (U ) f −1 (U ) fb−1 (U )
morphism.
In order to prove the existence of semigeometric quotients of affine varieties by reductive
groups, first we need to complete the algebraic picture concerning the behavior of ideals in
a given commutative algebra and in its subalgebra of invariants (see Lemma 12.4.2).
Lemma 2.6. Let G be a reductive group and R a commutative rational G-module algebra.
G
(1) If I is an ideal of R, then I ⊂ IR ∩ G R ⊂ rad(I). Moreover. if G is linearly reductive,
then I = IR ∩ G R.
(2) If J1 and J2 are G-stable ideals of R then
J1 ∩ G R + J2 ∩ G R ⊂ (J1 + J2 ) ∩ G R ⊂ rad (J1 ∩ G R) + (J2 ∩ G R) .

Moreover, if G is linearly reductive, then
J1 ∩ G R + J2 ∩ G R = (J1 + J2 ) ∩ G R.

In particular, if J1 + J2 = R, then J1 ∩ G R + J2 ∩ G R = G R.

Proof. (1) We leave as an exercise for the reader the proof of this part of the lemma (see
Exercise 1).
(2) It is clear that J1 ∩ G R + J2 ∩ G R ⊂ (J1 + J2 ) ∩ G R. If r1 + r2 ∈ (J1 + J2 ) ∩ G R, with

r1 ∈ J1 and r2 ∈ J2 , consider r1 + J1 ∩ J2 ∈ J1 /(J1 ∩ J2 ) and r2 + J1 ∩ J2 ∈ J2 /(J1 ∩ J2 ). As

G
x·(r1 +r2 ) = r1 +r2 , it follows that x·r1 −r1 = r2 −x·r2 ∈ J1 ∩J2 , i.e., r1 ∈ J1 /(J1 ∩J2 ) ,
G
and similarly r2 ∈ J2 /(J1 ∩ J2 ) . If p is the characteristic exponent of k, using the
geometric reductivity of Gn we deduce the existence of n ≥ 0 and elements s1 ∈ G J1 and
G p pn n
s2 ∈ J2 such that s1 − r1 ∈ J1 ∩ J2 and s2 − r2 ∈ J1 ∩ J2 . Then (r1 + r2 )p = s1 + s2 + t
G n
with t ∈ (J1 ∩ J2 ). In other words, (r1 + r2 )p ∈ G J1 + G J2 = J1 ∩ G R + J2 ∩ G R. The rest
of the assertions are easy to prove.
Observation 2.7. (1) Notice that in the particular case that IR = R, conclusion (1)
of the above lemma implies that rad(I) = G R and then that I = G R. In this sense, this
result is a generalization of Lemma 12.4.2.
(2) If I is a radical ideal, then I = IR ∩ G R.
Observation 2.8. Let G be a reductive group acting regularly on an affine variety X.
G G
Then k[X] is finitely generated (see Theorem 13.3.6), and the inclusion k[X] ⊂ k[X]
G
induces a dominant morphism π : X → Y = Spm k[X] . Recall that the principal open
G
sets Yf : f ∈ k[X] form a basis of the topology of Y , and that if z ∈ X and Mz ⊂ k[X]
G
denotes the associated maximal ideal, then Mz ∩ k[X] is the maximal ideal associated with
π(z).
Observe that in the situation above, π −1 (Yf ) = Xf (see Exercise 3).
We prove now that if G is a reductive group acting on an affine algebraic variety X,

then the semigeometric quotient exists and is an affine variety.
Theorem 2.9. Let G be a reductive group acting regularly on an affine variety X.
G
Consider the affine variety Y = Spec k[X] and let π : X → Y be the morphism induced
G
by the inclusion k[X] ⊂ k[X]. Then π is the semigeometric quotient for the G-action. In
particular, π : X → Y is the categorical quotient.
Proof. We need to prove condition (1) to (4) of Definition 2.1.

G
(1) As f π(g · x) = f π(x) for all f ∈ k[X], it follows that π(g · x) = π(x) for all x ∈ X,
g ∈ G.
G
In order to verify that π is surjective, consider a maximal ideal My ⊂ k[X] and
its extension My k[X] ⊂ k[X]. It follows from Lemma 2.6 — or Lemma 12.4.2 — that
this ideal is proper in k[X]. Choose a maximal ideal My k[X] ⊂ Mz ⊂ k[X]; then My =
G G G
My k[X] ∩ k[X] ⊂ Mz ∩ k[X] and My = Mz ∩ k[X], i.e., π(y) = z.
G
(2) If V is an open subset of Y , then πV] : OY (V ) → OX π −1 (V ) . Clearly, in order to

G
verify that πV] is an isomorphism, we can assume that V = Yf for some f ∈ k[X]. In this
G G
case, πY] f : OY (Yf ) = k[X] f → k[Xf ] is the identity map.

(3) If W ⊂ X is closed and G-stable, let y ∈ Y \ π(W ). Clearly, the closed G-stable
subsets W andπ −1 (y) are disjoint. Hence, J1 + J2 = k[X], where J1 = I(W ) ⊂ k[X] and
J2 = I π −1 (y) ⊂ k[X]. Since J1 and J2 are G-stable, using Lemma 2.6 we find f ∈ G J1 ,
g ∈ G J2 such that 1 = f + g. Then f π(W ) = 0 and f (y) = 1 − g(y) = 1, and y 6∈ π(W ).

Hence, π(W ) is closed in Y .

(4) The proof of this part is left as an exercise for the reader (see Exercise 4).
Corollary 2.10. In the notations of Theorem 2.9, if V ⊂ Y is an open subset, then
is the categorical quotient for the action of G on π −1 (V ).

V, π −1
π (V )
Proof. This is a straightforward consequence of Lemma 2.4.

Observation 2.11. It is important to notice that the hypothesis concerning the reduc-
tivity of G is crucial in the above theorem, not only to guarantee the finite generation of
the invariants but to control the behavior of the ideals in k[X] in relation to the ideals in
G
k[X] (Lemma 2.6).
Indeed, even if the action of an algebraic group G on
X has finitely generated invariants
G
and hence one can define π : X → Y = Spm k[X] , the pair (Y, π) may be completely
unrelated with the categorical quotient. For example, if B is a Borel subgroup of a group
B B
G, then k[G] = k, and Spm k[G] = {p}, whereas in general G/B is a projective variety
of positive dimension.
Example 2.12. Consider the action of the reductive group Gm on A2 given by t·(x, y) =
(tx, t−1 y). There are three types of orbits:
(1)
closed orbits of dimension 1, which are parameterized by k∗ and are of the form Oa =
2
(x, y) ∈ k : xy = a , a 6= 0;

(2) two non closed orbits of dimension 1: Ox = (x, 0) : x 6= 0 and Oy = (0, y) : y 6= 0 ;

(3) one closed orbit of dimension 0: O0 = (0, 0) — the only fixed point.
Gm
Let p(X, Y ) ∈ k[X, Y ], i.e., p(t−1 X, tY ) = p(X, Y ) for all t ∈ Gm . If p is homogeneous
of degree n, then
X X
p(X, Y ) = ar,s X r Y s = ar,s ts−r X r Y s ,
r+s=n r+s=n
and ar,s = 0 unless r = s. Hence,

(
0 n odd
p(X, Y ) =
akk (XY )k n = 2k.
G
It follows that m k[X, Y ] = k[XY ] and the categorical quotient is given by the mor-
phism π : A2 → A1 , π(x, y) = xy.
Notice that the generic fiber of π is Oa = π −1 (a), a ∈ k∗ , which is a closed orbit of
maximal dimension. The other orbits appear agglomerated in the fiber π −1 (0) = Ox ∪ Oy ∪
O0 .
Observe that the principal open subset A2xy ⊂ A2 is G-stable and

[
A2xy = (x, y) ∈ k2 : xy 6= 0 =

Oa ,
a∈k∗
i.e., A2xy is the union of all the closed orbits of maximal dimension. If we restrict π to A2xy
we obtain that k∗ , π 2 is the geometric quotient. Indeed, in view of Lemma 2.4 we only

Axy
need to verify that π is an open map, and this follows from the fact that the fibers of
A2
xy
π are the orbits (see Exercise 7).

A2
xy
In particular, this is an example of a semigeometric quotient that is not geometric.

The main features of this example are typical of the general behavior of the quotient
by the action of a reductive group (see Theorem 3.13).
n n
Example 2.13.The action of GLn by matrix multiplication on A = k has two orbits:
n
the open orbit A \ (0, 0) and the fixed point (0, 0) . Hence, a regular function constant
GLn
along the orbits is constant on An , so that k[An ] = k, and the categorical quotient is
n
the map A → {∗}.
3 Actions by reductive groups: the geometric quotient

As we just observed in the examples of the last section, there are structural obstructions
— even in the case of a reductive group — to the existence of the geometric quotient. As
shown in Example 2.12, these problems are related with the existence of non closed orbits
(see Section 7.4). In a sense, Theorem 3.4 below shows that this is the only obstruction.
At this point, two approaches to the problem of the existence of a geometric quotient
for actions of reductive groups are possible. One is to restrict our attention to actions that
have always closed orbits — for example, homogeneous spaces or double cosets spaces —
in which case the categorical quotient will be the geometric quotient. The other approach
consists in looking for large open subsets inside of the categorical quotient, whose points
have fibers that are closed orbits. This seems to be the more interesting viewpoint and,
following [121], is the one we adopt in this section.
First we extract two consequences of Theorem 2.9.
Corollary 3.1. In the notations of Definition 2.1, if y ∈ Y , then π −1 (y) contains one
and only one closed orbit.
In particular, the above result is valid in the case that G is a reductive group and X an
affine G-variety.
Proof. Since π −1 (y) is closed and G-stable, any orbit of minimal dimension in π −1 (y)
is closed in X (see Corollary 4.4.23), and therefore each fiber contains at least one closed
orbit. If O1 and O2 are closed orbits contained in π −1 (y), then condition (4) of Definition
2.1 guarantees that O1 ∩ O2 6= ∅, i.e., O1 = O2 .
Example 3.2. In the case of Example 2.12, if y ∈ k∗ then π −1 (y) is a closed orbit, and
−1

π {0} contains the closed orbit (0, 0) .
Corollary 3.3. In the notations of Definition 2.1, x1 , x2 ∈ X are in the same fiber
of π if and only if O(x1 ) ∩ O(x1 ) 6= ∅.
affine G-variety.

Proof. For all x ∈ X, π O(x) = π(x) , and if O(x1 ) ∩ O(x1 ) = ∅, then

π(x1 ) ∩ π(x2 ) = π O(x1 ) ∩ π O(x1 ) = ∅.
Hence, π(x1 ) 6= π(x2 ). The converse is clear.

Theorem 3.4. In the notations of Definition 2.1, if V ⊂ Y is an open subset such that
the restricted action G × π −1 (V ) → π −1 (V ) has closed orbits in π −1 (V ), then V, π −1
π (V )
is a geometric quotient for the action of G on π −1 (V ) — therefore, if G acts on X with

closed orbits, then (Y, π) is a geometric quotient.
affine G-variety.
Proof. In view of Lemma 2.4, if V ⊂ Y is an open subset, then (V, ππ−1 (V ) ) is a

semigeometric quotient; hence, we may assume that X = V . The only conditions that still
need to be proved are that the map π is open and that each fiber is an orbit.
As we know that on each fiber there is only one closed orbit (see Corollary 3.1) and by
hypothesis all the orbits are closed, we conclude that on each fiber there is only one orbit.
Once this is established, we conclude from Lemma 2.3 that the image by π of a G-stable
open set is open. Then the map π is open (see Exercise 7).
As finite groups act with closed orbits, we obtain a new proof of Theorem 7.6.2:
G
Corollary 3.5. Let G be a finite group and X an affine G-variety. Then, k[X], π
is a geometric quotient.
If H ⊂ G is a reductive subgroup, then the orbits of the action by right translations

are closed, and hence we obtain a new proof of Theorem 12.4.4:
Corollary 3.6. Let H ⊂ G be a reductive subgroup. Then G/H is an affine variety
and π : G → G/H is the geometric quotient.
Observation 3.7. It is worth remarking that if (Y, π) is the categorical quotient of a
G-variety X, then in order to find a G-saturated open subset U ⊂ X such that the map
π(U ), π is the geometric quotient, two conditions should be satisfied: (a) all the orbits
U
of G on U are closed in U ; (b) U is the inverse image by π of an open subset of Y .
In some cases it may be impossible to find such a set U (see Example 2.13).
The proof of the next corollary is left as an exercise (see Exercise 8). It is a strengthening
of Theorem 7.4.22.
Corollary 3.8. In the notations of Definition 2.1, all the orbits of the action are
closed if and only if they have the same dimension.
affine G-variety.
Observation 3.9. One can easily construct examples of groups acting on affine varieties
with closed orbits with non equal dimensions (see Example 7.2.3).
In view of Theorem 3.4 and the observations that followed it, the relevance of the next
definitions for the problems related to the existence of geometric quotients should be clear
to the reader.
Definition 3.10. Let G be an affine algebraic group and X a G-variety. Define Ω(X) ⊂
X max as follows:
X max = x ∈ X : dim O(x) is maximal ,

Ω(X) = x ∈ X max : O(x) is closed in X .

Next we establish some properties of these subsets.

Lemma 3.11. In the notations
of Definition 3.10, X max
is an open G-stable subset of
max
X. If x ∈ X, then bd O(x) ∩ X = ∅, where bd O(x) denotes the boundary of O(x).

Moreover, if x ∈ X max , then bd O(x) = O(x) ∩ (X \ X max ). In particular, the orbits of
the points in X max are closed in X max .

Proof. It follows from Theorem 7.3.3 that for all n ≥ 0, x ∈ X : dim O(x) ≥ n
is open in X and hence X max is open. If x ∈ X, then bd O(x) is the union of orbits
of dimension smaller than the dimension of O(x) (see Theorem 7.3.3). Thus, bd O(x) ∩

X max = ∅ and if x ∈ X max , then bd O(x) ⊂ O(x) ∩ X \ X max .

Conversely, if x ∈ X max and y ∈ O(x) ∩ X \ X max , then the orbit of y has dimension
strictly smaller than dim O(x), and hence y 6∈ O(x).
The next lemma shows the importance of X max when dealing with geometric quotients.
Lemma 3.12. Let G be an affine algebraic group and X an irreducible G-variety. If
U ⊂ X is a G-stable subset of X such that there exists an orbit space for the action of G
on U , then U ⊂ X max .
Proof. If (Z, λ : U→ Z) is an orbit space, then Theorem2.6.4 guarantees that d : U →

N, u 7→ dim λ−1 λ(u) = dim O(u), is upper semicontinuous. As we proved in Theorem
7.3.3 d is also lower semicontinuous and then it is a continuous function. It follows that d
is locally constant in U , and hence if u ∈ U , then the subset

Vu = v ∈ U : dim O(v) = dim O(u)
is open in X. As X is irreducible, Vu ∩ X max 6= ∅, i.e., u ∈ Vu ⊂ X max and the result

follows.
If the semigeometric quotient exists, then useful information concerning Ω(X) can be
obtained: Ω(X) is the inverse image of an open set in the semigeometric quotient, and in
particular Ω(X) is open (see [126]). These conclusions will be valid in the case of a reductive
group acting on an affine variety.
Theorem 3.13. Let G be an affine algebraic group, X a G-variety such that there
exists a semigeometric quotient (Y, π). Then there exists an open subset Y 0 ⊂ Y such that
−1 0 0

π (Y ) = Ω(X) (see Definition 3.10) and consequently Y , π is a geometric quotient
Ω(X)
for the action of G on Ω(X).
affine G-variety.
Proof. If Y 0 = Y \ π X \ X max , since π X \ X max is closed — X max is open and

G-stable — then Y 0 is open.

Next we show that π−1 (Y 0 ) = Ω(X). First we prove that Ω(X) ⊂ π −1 (Y 0 ) — or

equivalently that π Ω(X) ∩ π X \ X max = ∅. Assume that there exist z ∈ Ω(X) and
w ∈ X \ X max such that π(z) = π(w). Then O(z) ⊂ X and X \ X max ⊂ X are G-stable,
closed and their images intersect. Thus, there exists w0 ∈ X \X max such that O(w0 ) = O(z).
This is clearly impossible as the dimensions of these orbits cannot coincide.
Conversely, if x ∈ X \X max is such that π(x) ∈ Y 0 , then π(x) ∈ π X \X max and this is

clearly impossible. Hence, π −1 (Y 0 ) ⊂ X max . Let x ∈ X max \ Ω(X) bemaxsuch that π(x) ∈ Y 0 .
In this case O(x) is not closed, so that
there exists z ∈ bd O(x) \ X (see Lemma 3.11).
max
Then π(x) = π(z) ∈ π X \ X , and this is a contradiction as we were assuming that
π(x) 6∈ π X \ X max .
Since the orbits in Ω(X) are closed, it follows from Theorem 3.4 that Y 0 , π

is a
Ω(X)
geometric quotient.
Observation 3.14. The reader should be aware that the set Ω(X) may be empty: in
the situation of Example 2.13 the orbit of maximal dimension is a proper open subset. More
generally, Ω(X) is empty whenever the action has a dense, and hence open, orbit.

In Example 2.12, X max = A2 \ (0, 0) and Ω(X) = A2xy , and if π : A2 → A1 is the
categorical quotient, then Ω(X) = π −1 (A1 − (0) ) and π −1 (A1 ) = A2 6= X max . In other

words, X max is not the inverse image of an open subset of the categorical quotient.
Observation 3.15. Concerning actions of reductive groups, in this section we con-
sidered mainly the case of affine varieties. In Chapter 16 we present some initial results
towards the study of geometric invariant theory, that deals mainly with the construction
of quotients for actions of reductive groups on projective varieties. The reader interested in
these results — that are crucial for the study of moduli — should read [121] and for a more
concrete presentation, [134].
Definition 3.16. Let G be an affine algebraic group acting on an algebraic variety X.

We define the socle of the action to be
[
Xsoc = O(x).
O(x)=O(x)
Observation 3.17. Observe that (Xsoc )soc = Xsoc .

Lemma 3.18. Let G be a reductive group acting on an affine variety X, and π : X →
G\X the categorical quotient. Then
Ω(X) = π −1 (G\X) \ π(X \ X max ) .

In particular, Ω(X) is a (possibly empty) G-stable open subset.
Proof. It is clear that Ω(X) ⊂ π −1 (G\X) \ π(X \ X max ) . Since X \ X max is a

G-stable closed subset, it follows that π(X \ X max ) is closed in G\X. Let x ∈ X \ Ω(X);
then O(x) contains a unique closed orbit O0 , with dim O0 < dim O. It follows that π(x) =
π(O0 ) ∈ π(X \ X max ); that is x ∈
/ π −1 (G\X) \ π(X \ X max ) .
Theorem 3.19 (V. Popov, [140]). Let G be reductive group acting on an affine algebraic
variety X. Then Ω(X) is an open subset of X (that may eventually be the empty set).
Proof. The result is an easy consequence of Theorem 3.13; see also [140, Theorem 4]
for an early proof.
Next we rephrase the contents of Lemma 3.18 and Theorem 3.19 with a view towards a
characterization of observable actions in terms of the non emptiness of the interior of Ω(X)
(see Theorem 15.4.2).
Lemma 3.20. Let G be a reductive group acting on an affine variety X, and π : X →
G\X the categorical quotient. Then the following are equivalent.
(1) π max
: X \ X max → G\X is dominant.
X\X
(2) π : X \ X max → G\X is surjective.

X\X max
(3) X max does not contain any closed orbit.

(4) Ω(X) = ∅.
We record the following result for future use in the study of observable actions; see
Chapter 15, Section 6.
Theorem 3.21. Let G be a reductive group acting on an affine variety X, and Z ⊂ X
a G-stable closed subset such that π : Z → G \X is dominant. Then π(Z) = G\X and
Z
the unique morphism ϕ : G\Z → G \X, defined by the following commutative diagram:

Z /X
ρ π

G\Z / G\X,
ϕ
is bijective and purely inseparable, i.e., ϕ is a bijective

morphism,
and ϕ∗ : k[X]G → k[Z]G
induces a purely inseparable field extension k[X]G ⊂ k[Z]G .
Moreover: (a) Xsoc ⊂ Z; (b) Xsoc is irreducible; (c) Ω(Xsoc ) 6= ∅.
In particular, if char k = 0, then π : Z → G \X is the categorical quotient and
Z
ϕ : G\Xsoc → G \X is an isomorphism.
Proof. From the fact that Z ⊂ X is closed, it follows that π(Z) ⊂ G\X is also closed
and hence that π(Z) = G\X. In particular, if O ⊂ X is a closed orbit, then π(O) ∈ π(Z) =
X, and hence there exists an orbit O0 ∈ Z such that π(O0 ) = π(O). Therefore, since π
separates closed orbits, it follows that O0 = O ⊂ Z and thus Xsoc ⊂ Z.
Since ϕ ◦ ρ is dominant, it follows that ϕ is also dominant. Since π and ρ separate closed
orbits, then ϕ is injective. Thus, we have a factorization as in the following commutative
diagram:
/ k[X] / / k[Z]
k[X]G v
O
ϕ∗ ?
)
k[Z]G
Since G is reductive, k[Z]G is finitely generated and there exists n ≥ 0 such that for
n
all f ∈ k[Z]G , there is a g ∈ k[X]G , with ϕ(g) = f p (see Theorem 10. 2.10). Hence
π : Z → G \X is a purely inseparable morphism. It follows that if char k = 0, then
Z
G\X ∼ = G \Z.
Since π : Xsoc → G\X is surjective, it follows that ϕ : G \Xsoc G → G\X is a
Xsoc
purely inseparable morphism.
Let Z ⊂ Xsoc be an irreducible component such that π is dominant. Then Xsoc ⊂ Z,

Z
and hence Xsoc is irreducible.
Consider Z = Xsoc \ (Xsoc )max . If Z dominates G\Xsoc , it follows that Xsoc =
(Xsoc )soc ⊂ Z, that is a contradiction. Hence, by Lemma 3.20, it follows that Xsoc =
6 ∅.
4 Canonical forms of matrices: a geometric perspective

In this section we consider the canonical form of a matrix from the viewpoint of geo-
metric invariant theory.
First we fix some notations that will be in force along this section.
If A ∈ Mn (k) denote as
n
X
χA (t) = det(t Id −A) = tn + (−1)i ci (A)tn−i
i=1
the characteristicpolynomial and call π : Mn (k) → An , the morphism given as π(A) =

c1 (A), . . . , cn (A) .
Call C : An → Mn (k), C(c1 , . . . , cn ) = C(c1 ,...,cn ) , the morphism defined by taking the
companion matrix (see Appendix, Definition 2.2).
Let GLn act by conjugation on Mn (k). It is clear that π is constant along the GLn -orbits
and that C is a section of π, i.e., π ◦ C = idAn .
Next we show that there is a close relationship between the algebraic properties of a
linear transformation and the geometric properties of its GLn -orbit.
Lemma 4.1. Let A = As + An be the Jordan decomposition of A ∈ Mn (k). Then, for
all a ∈ k∗ the matrices Aa = As + aAn are conjugate, As ∈ O(A) and π(A) = π(As ).
If A, B ∈ Mn (k), then O(As ) = O(Bs ) if and only if O(A) ∩ O(B) 6= ∅.
Proof. As an arbitrary matrix is similar to a matrix in Jordan form, it is enough to

prove the result assumingthat A is a Jordan
 block; in other words, we may suppose that
0 1 0 0
 
As = λ id and that An = 
 0 .

 1 
0 0
If g = diag(1, a, . . . , an−1 ) ∈ GLn , then g −1 As g = As and g −1 An g = aAn . Thus, for all

a ∈ k∗ , Aa = As +aAn ∈ O(A), and hence A0 = As ∈ O(A). The continuity of π guarantees
that π(A) = π(As ).
If As and Bs are in the same orbit, as both matrices are in the closure of the orbits of A
and B, respectively, it follows that O(A) ∩ O(B) 6= ∅. Conversely, if O(A) ∩ O(B) 6= ∅, then
π(A) = π(B) so that π(As ) = π(Bs ). As two diagonalizable linear transformations with the
same characteristic polynomial are similar, the proof is finished.
Observation 4.2. From Lemma 4.1 we deduce that the only closed orbits of the above
action are the orbits of semisimple matrices (see Exercise 11).
The situation considered above is related to the problem of the closedness of conjugacy
classes in arbitrary affine algebraic groups. This subject has been extensively studied, and
in a similar direction as the above result in [173] it is proved that for a reductive group the
only closed conjugacy classes are the ones corresponding to semisimple elements.
Theorem 4.3. Consider the action of GLn on Mn (k) given by conjugation. Then, the
pair (An , π) is the categorical quotient.
Proof. Let f : Mn (k) → Z be a morphism constant along the conjugacy classes, and
assume that A and B are matrices with the same characteristic polynomial, i.e., matrices
on the same fiber of π; then O(As ) = O(Bs ), and this implies that f (As ) = f (Bs ). Hence,
by continuity we deduce that f (A) = f (B), and then that there exists a set theoretical map
fˆ : An → Z such that fˆ ◦ π = f . Hence fˆ = fˆ ◦ π ◦ C = f ◦ C, which is a morphism.
Observation 4.4. As the group GLn is reductive, we conclude from the uniqueness of
the categorical quotient and theorems 2.9 and 2.5 that (An , π) is a semigeometric quotient.
Theorem 4.5. In the notations of Theorem 4.3, call
n n
X o
D = (c1 , . . . , cn ) ∈ An : the discriminant of tn + (−1)i ci tn−i is zero .
i=1
Then D ⊂ An is closed and if p ∈ An \ D, then the fiber π −1 (p) consists of only one
orbit that is closed.
Proof. As the characteristic polynomial of any matrix in π −1 (p) has different roots, it
coincides with the minimal polynomial and hence the matrix is semisimple; hence Lemma
4.1 guarantees the result.
max
Our next goal is to identify Mn (k) and Ω Mn (k) (see Definition 3.10).
If A ∈ Mn (k), it is clear that dim O(A) is maximal when dim GLnA = n2 − dim O(A)
is minimal. Since
GLnA = {g ∈ GLn : gA = Ag} ⊂ GLn ⊂ Mn (k),

it follows that dim GLnA = dim B ∈ Mn (k) : BA = AB .

In order to compute dim B ∈ Mn (k) : BA = AB , observe that if we consider kn
as a k[X]-module via the action f · v = f (A)(v) for v ∈ kn and f ∈ k[X], then {B ∈
Mn (k) : AB = BA} = Endk[X] (kn ). Using the well-known structure L theory for finitely
s
generated modules over principal ideal domains (see [69]) we write kn = i=1 k[X]/fi k[X],
where fi ∈ k[X] and (a) fi+1 divides fi for i = 1, . . . , s − 1; (b) f1 = mA ; (c) the degrees
di = deg(fi ) form a partition of n — observe that d1 ≥ d2 ≥ · · · ≥ ds .
Clearly, M
Endk[X] (kn ) =

Endk[X] k[X]/fi k[X], k[X]/fj k[X] ,
i,j
where the dimension of the (i, j)-summand

P is mi,j = min(di , dj ) (see Appendix, Observation
3.7), and consequently dim GLn,A = i,j mi,j .
In order to finishPour calculations we need to find the partitions p = p1 ≥ · · · ≥ ps of
n such that L(p) = i,j mi,j is minimal. If we call p0 the partition of n − p1 obtained by
eliminating p1 from the partition p, then L(p) = p1 +· · ·+ps +L(p0 ). Then L(p) ≥ n and it
will be equal to n if and only if L(p0 ) = 0, i.e., p2 = · · · = ps = 0. In other words we deduce
that the minimum is n and that it is obtained at the partition {n}. Hence, the minimal
dimension of GLnA corresponds to a decomposition of kn with only one summand, i.e., to
a matrix A that has characteristic polynomial coinciding with the minimal polynomial, and
hence has a cyclic vector.
The next theorem summarizes the results obtained in this section.

Theorem 4.6. The map π : Mn (k) → kn , π(A) = c1 (A), . . . , cn (A) , is the semige-
ometric, and hence categorical, quotient for the action by conjugation of GLn on Mn (k),
and
Mn (k)max = A ∈ Mn (k) : A has a cyclic vector ,

Ω Mn (k) = A ∈ Mn (k) : A has n different eigenvalues .
In particular — recall the definition of D in Theorem 4.5 — Ω Mn (k) = π −1 (An \ D)

and hence An \ D, π is a geometric quotient for the action of GLn on Ω Mn (k) .
Ω(Mn (k))

5 Rosenlicht’s theorem
In this section, generalizing somewhat the results obtained in Section 3 for reductive
groups, we prove that if G is an affine algebraic group and X is a G-variety, then there
exists a nonempty G-stable open subset of X such that the restricted action has a geometric
quotient. The proof we present below is an adaptation for the case of irreducible algebraic
varieties of Rosenlicht’s original proof ([153], [156]). For explicit constructions of this open
subset in particular situations see Example 7.4.21 and Exercise 7.18.
Lemma 5.1. Let G be an algebraic group acting on an irreducible algebraic variety
X. Then there exist a G-stable open set U ⊂ X, an algebraic variety Y and a morphism
π : U → Y such that: (1) π is surjective separable; (2) π is constant along the orbits of U ,
and its fibers are equidimensional; (3) k(Y ) ∼
G
= k(U ).
G G
Proof. Let {f1 , . . . , fr } ∈ k(X) be field generators of the extension k ⊂ k(X).
Then for all i = 1, . . . , r, the domain of definitionTrof fi , D(fi ) ⊂ X, is an open G-stable

subset. Consider the open G-stable subset U = i=1 D(fi ), and Y = Spm k[f1 , . . . , fr ] .
Then the inclusion k[f1 , . . . , fr ] ,→ OU (U ) induces a dominant G-morphism π : U → Y .
Using Chevalley’s theorem 2.6.4, we restrict Y and U in order to guarantee that π : U → Y
is surjective and with equidimensional fibers.
G
We deduce from Theorem 1.2.30 that the field extension k(X) ⊂ k(X) is separable.
Then, the morphism π is separable.
Observation 5.2. (1) Since the set of normal points of Y is open (see Theorem 2.4.11),
further restricting U and Y we can suppose that Y is a normal algebraic variety. Hence,
Chevalley’s theorem 2.6.4 guarantees that we can also assume that π is an open map (see
also Observation 2.6.5).
(2) Notice that we can further restrict the open set U in order to guarantee that Y is an
affine variety.
Theorem 5.3 (M. Rosenlicht). Let G be an algebraic group and X an irreducible G-
variety. Then there exists a G-stable open subset ∅ =
6 X0 ⊂ X such that the action of G
restricted to X0 has a geometric quotient.
Proof. First, we assume that the group G is connected. Let U ⊂ X and π : U → Y be

as in Lemma 5.1 and Observation 5.2. Theorem 7.3.3 guarantees that we can also assume
that G-orbits of U have the same dimension.
Next we prove that further restricting U and Y then π : U → Y satisfies the conditions
(1) and (2 0 ) of Lemma 7.4.14 that characterize the geometric quotient.
G −1
(2 0 ) If f ∈ k(U ) is a G-invariant rational function defined at x ∈ U and π ] (f ) — that
we identify with f — is not defined at y = π(x), then 1/f ∈ k(Y ) is defined and vanishes
at y. Hence, 1/f = π ] (1/f ) is defined and vanishes at x, and this contradicts the fact that
f is defined at x.
(1) We further restrict U and Y in order to guarantee that the fibers of π are the G-orbits:
Consider the morphism ϕ : G × U → U ×Y U , ϕ(g, x) = (x, g · x), where U ×Y U =
(x, x0 ) ∈ U × U : π(x) = π(x0 ) . If we denote as π1 , π2 : U ×Y U → U the projections in

the corresponding coordinates, then the diagram below is commutative

G×U
ϕ
%
U ×Y U
π2
/U
π1 π

U /Y
π
−1
and π1−1 (x) = {x} × π −1 (x) = {x} × O(x). Similarly, π2−1 (x) =

π(x) ,
π1
Im ϕ
−1
π −1 π(x) × {x}, π2 Im ϕ

(x) = O(x) × {x}.
In this situation, all we have to prove is that Im ϕ is dense, i.e., that ϕ is dominant.
Indeed, since dim ϕ−1 (x, g · x) = dim Gx , then, using the equidimensionality of the fibers
and of the orbits, we obtain that
dim Gx = dim(G × U ) − dim Im ϕ = dim G + dim U − dim(U ×Y U ) =
dim G − dim U + dim Y = dim G − dim π −1 (x),
and we conclude that dim O(x) = dim π −1 (x). Hence, as all the fibers have the same di-
mension, each fiber is a finite union of closed orbits. Moreover, since Im ϕ is dense and G is
connected, we can further restrict U in order to guarantee that each fiber of π is only one
orbit.
All that remains to be proven is that ϕ is dominant:
It is enough to prove that if V1 , V2 ⊂ U are affine open subsets, then ϕ :
ϕ−1 (V1 ×Y V2 )
−1
ϕ (V1 ×Y V2 ) → V1 ×Y V2 is dominant. In this case, V1 ×Y V2 is affine and the dominance
of ϕ is characterized by the injectivity of ϕ] : k[V1 ] ⊗k[Y ] k[V2 ] → OG×U ϕ−1 (V1 ×Y V2 ) .

Pl −1
Pl
Assume that ϕ] i=1 gi ⊗fi = 0 ∈ OG×U ϕ (V1 ×Y V2 ) ⊂ k(G×U ), with i=1 gi ⊗
G
fi ∈ k[V1 ] ⊗k[Y ] k[V2 ] and f1 , . . . , fl linearly independent over k(U ). Explicitly, we have
that if a ∈ G and x ∈ U , then
l
X X
ϕ] gi ⊗ fi (a, x) = gi (x)fi (ax) = 0.
i=1
Applying Exercise 20 we conclude that g1 = · · · = gl = 0, i.e., that the map ϕ] is

injective.
Assume now that G is not connected, and let U1 ⊂ X be an open G1 -stable subset
Sr a geometric quotient π : U1 → Y for the action of G1 . Consider
admitting Tr the decomposition
G = i=1 ai G1 of G in its connected components; then U = i=1 ai · U1 is a nonempty
G-stable open subset. Then π : U → Z = π(U ) ⊂ Y is a geometric quotient for the
U
G1 -action. Moreover, since G1 is normal in G, for all i = 1, . . . , l the map ϕi : U → Z,
ϕi (x) = π(ai x), is a morphism constant along the G1 -orbits. Hence, G/G1 acts on Z, by
(ai G1 ) · π(x) = π(ai x). Let U 0 ⊂ Z be an open G/G1 -stable subset such that the geometric
quotient π e : U 0 → Y exists (see Exercise 21). Then π e ◦ π −1 0 : π −1 (U 0 ) → Y is the
π (U )
geometric quotient for the action of G on π −1 (U 0 ).

We finish this section with a useful application of Rosenlicht’s theorem: orbits in general
position can be separated by a finite number of invariant rational functions; see [166] for
another proof.
Theorem 5.4. Let G be an affine algebraic group and X an irreducible G-variety. Then
G
there exist U ⊂ X, open G-stable subset, and F ⊂ k(X), finite set of invariant rational
functions, such that F separates the G-orbits in U .
Proof. Let U be an open subset where, in accordance with Theorem 5.3, the geometric
quotient π : U → Y exists. Assume moreover that Y is affine (see Observation 5.2). Then
G G
k(X) = k(U ) = π ] k(Y ) is a finitely generated field extension of k. Since any finite set of
generators F 0 separates the points of Y , it follows that F = π ] (F 0 ) separates the orbits of
G in U .
6 Induced actions and homogeneous fiber bundles

In this section we present the geometric counterpart of the concept of induced module.
We start by recalling Definition 7.7.3: if H ⊂ G is a closed inclusion of affine algebraic
groups and M ∈ H M is a rational H-module, the module induced from M by the inclusion
H ⊂ G is the rational G-module IndG H

H (M ) = k[G] ⊗ M , where in order to take the H-
fixed part we use the diagonal action. The action of G on IndG H (M ) is given as the restriction
of the action on k[G] ⊗ M given for z ∈ G, f ∈ k[G] and m ∈ M as z ? (f ⊗ m) = f · z −1 ⊗ m.
The induced module is characterized by the following universal property: if N is a G-
module and ϕ : N → M is a morphism of H-modules, then there exists a unique morphism
of G-modules ϕ e : N → IndG H (M ) such that ϕ = EM ◦ ϕ, e where EM : IndG H (M ) → M
is thePevaluation map
P — recall that E M is the morphism of H-modules given by the rule
EM ( fi ⊗ mi ) = fi (e)mi (see Definition 7.7.9). As we noted in Observation 7.7.2, the
induction functor is the right adjoint of the restriction function G M → H M, the map EM
being the counit. That is, the map ϕ 7→ ϕ e defines, for each N ∈ G M and M ∈ H M as
above, natural isomorphisms: HomH (ResH ∼ G
G (N ), M ) = HomG (N, IndH ).
A geometric version of the above construction is the following:
Definition 6.1. Let G be an affine algebraic group and H ⊂ G a closed subgroup
and X an algebraic variety equipped with a regular H-action. Consider the variety G × X
equipped with the regular action of H given as follows: h · (g, x) = (gh−1 , h · x) for h ∈ H,
g ∈ G, x ∈ X. The homogeneous fiber bundle (associated to the inclusion H ⊂ G) of fiber
X is — in case it exists — the geometric quotient p : G × X → G ∗H X with respect to the
above mentioned action. It is customary to write p(g, x) = [g, x]. The map p is called the
quotient morphism.
Observation 6.2. (1) If X is an affine variety and H is a reductive group, the con-
struction of the induced module and of the variety G ∗H X can be related by the duality
between affine algebras and affine varieties introduced in Chapter 1: consider the affine
G-module algebra IndG G
H k[X] and define G ∗H X = Spm IndH (k[X]) . Then G ∗H X is a
G-variety, and by duality it satisfies the universal property of the definition of homogeneous
fiber bundle.
(2) At variance with the situation of the induced representation, the homogeneous fiber
bundle G ∗H X does not exist in general. But with mild conditions on the geometry of
the algebraic variety X — e.g., if X is quasi–projective — its existence can be guaranteed,
following results by J.-P. Serre (see[162]) or, in accordance with work by A. Bialynicki–
Birula, in a more general context (see [5] for the precise formulation), see Theorem 6.5
below. The usual proofs of these existence results use tools that go beyond the scope of
the subjects treated in this book and it will not be proved here in its full generality; see
Theorem 6.12 below.
Before stating an existence theorem (Theorem 6.5, see also Theorem 6.12 below), we
present some initial properties of the homogeneous fiber bundle associated to X that can
be deduced from Definition 6.1.
Observation 6.3. (1) Being a geometric quotient, the fibers of the quotient morphism
p are the H-orbits.
(2) Moreover, if we let G act on G × X by g 0 · (g, x) = (g 0 g, x), this action factors to the
categorical quotient
— we write g 0 · [g, x] = [g 0 g, x]. Indeed g 0 · (gh−1 , h · x) = (g 0 gh−1 , h · x)
0 0
and g · h · (g, x) = h · g · (g, x) . With respect to these actions the morphism p is
G-equivariant. Notice also that [g, x] = g · [1, x].
(3) The map (g, x) 7→ gH : G × X → G/H satisfies that (gh−1 , h · x) 7→ gh−1 H = gH
and then it induces a morphism of algebraic varieties ρ : G ∗H X → G/H, ρ([g, x]) = gH.
This map is called the canonical projection and has all fibers isomorphic. Indeed, the G-
equivariance of ρ : G ∗H X → G/H implies that its fibers are of the form
ρ−1 ρ[g, x] = [g, y] : y ∈ X = g · [1, y] : y ∈ X .

Hence, all the fibers of ρ are isomorphic to ρ−1 (1H); this justifies the name homogeneous
fiber bundle.
(4) If we define the map ιX : X ,→ G ∗H X (called the canonical immersion) as
ιX (x) = p(1, x), then ιX is an H-equivariant injective morphism, with closed image. In-
deed, if ιX (x) = [1, x] = ιX (y) = [1, y], we easily deduce that x = y. Also ιX (h · y) =
[1, hy] = [h, y] = h · [1, y]. Moreover, we have that Im(ιX ) = p(H × X) , which is a closed
subset, since H × X is a H-stable closed subset of G × X and p is a geometric quotient.
(5) Assume that Y is a G-variety and f : X → Y a H-equivariant morphism; then there
exists a unique G-equivariant morphism fe : G ∗H X → Y such that the following diagram
is commutative
X
ιX
/ G ∗H X
f
{ fe
Y
The proof of this adjunction property goes as follows: given f as above, define F :
G × X → Y , F (g, x) = g · f (x). Clearly, F (gh−1 , h · x) = gh−1 · f (h · x) = (gh−1 h) · f (x) =

F (g, x). Then, it follows by the universal property of the quotient that there is a morphism

fe : G ∗H X → Y with the property that fe [g, x] = g · f (x). It follows from the definition
that fe is G-equivariant. As the quotient morphism is surjective, the uniqueness is clear.
Hence, the natural transformation ιX is the unit of the adjunction
VarG (G ∗H X, Y ) ∼
= VarH (X, Y ),
H
where the categories are that of varieties acted regularly by affine groups and its morphisms.
(6) If (g, x) ∈ G × X, then OH (g, x) = (gh−1 , hx) : h ∈ H . This set is the graph of the

morphism gH → X, gh 7→ h−1 · x and therefore is closed.

Example 6.4. Let G be an affine algebraic group, H ⊂ G a closed subgroup and
X = {p} a point. Then G ∗H {p} = H \\(G × {p}) ∼ = G/H — note that since H acts on G
by h · g = gh−1 , the quotient is G/H instead of H \G.
In particular, if H is a Borel subgroup with H ( G, then G/H is a projective variety
and it is not a point while IndG

H k[{p}] = k: as in the study of the existence of quotients,
G

the finite generation of IndH k[{p}] does not imply that homogeneous fiber bundle is affine
(see also Lemma 6.9).
Assuming some mild conditions on X — e.g., that any finite subset of points in X is
contained in an affine open subset — the existence of the quotient G ∗H X can be proved,
as well as more subtle properties of this construction.
Theorem 6.5. Let G be an algebraic group, H ⊂ G a closed subgroup and X an
algebraic H-variety with the property that any finite subset of points in X is contained
in an affine open subset. Consider the diagonal regular action of H on G × X, given by
h · (g, x) = (gh−1 , h · x). Then the geometric quotient H \\(G × X) exists. Moreover:
(1) The canonical projection ρ : G ∗H X → G/H is a G-equivariant fibration of fiber X, in
such a way that ρ : G ∗H X → G/H is a locally isotrivial fibration — for the definition of
a locally isotrivial fibration, see, for example, [5] or [162].
(2) The map ιX : X → G∗H X is a closed immersion, with Im(ιX ) = ρ−1 (1H). In particular,
ρ−1 (gH) ∼
= X for all g ∈ G.
Proof. For a proof, see [5, Theorem 3], [162, Proposition 5] and [144, Theorem 4.19]).
See also [5] for a generalization of this result to the category of algebraic spaces.
From now on, we assume that X satisfies the hypothesis of Theorem 6.5, namely that
any finite subset of points in X is contained in an affine open subset.
Observation 6.6. (1) From the perspective of torsors — see, for example, [21] for the
basic definitions and references on this topic — ρ : G ∗H X → G/H is the fiber bundle
of fiber X associated to the H-torsor (or principal H-bundle) G → G/H, that fits in the
following commutative diagram
G×X
p1
/G (14.1)
p π

G ∗H X / G/H
ρ
where both p and π are H-torsors.

(2) It follows from the local isotriviality of ρ : G∗H X → G/H that if K is a closed subgroup
such that H ⊂ K ⊂ G, then the inclusion K × X ,→ G × X induces a closed immersion

K ∗H X ,→ G ∗H X. In particular, the commutative diagram of item (2) corresponding to
the homogeneous fiber bundle K ∗H X is obtained by restriction of the diagram (14.1).
(3) As a consequence of the Leray–Grothendieck spectral sequence (see [61] and [115]), one
can prove that if M ∈ H M is a H-module, then ρ : G ∗H M → G/H is a G-linearized
vector bundle. That means that there exists an open covering of G/H by affine subsets
{Ui : i ∈ I}, such that ρ−1 (Ui ) ∼
= Ui × M for all i ∈ I, and the G-action is linear on the
fibers; see also Exercise 19.
Lemma 6.7. Let G be an affine algebraic group, H ⊂ G a closed subgroup and X an
algebraic H-variety — recall that we are assuming that X satisfies the hypothesis of Theorem
6.5. We use the notations established above and identify X with the fiber ρ−1 (1H) = [1, x] :

x ∈ X . Then:

(1) Any G-orbit OG [g, x] = G·[g, x] intersects X in a unique H-orbit, namely OH [1, x] =
H · [1, x].
(2) Let Z ⊂ G ∗H X be a G-stable closed subset. Then Z = G · (Z ∩ X). Conversely, if
Y ⊂ X is a H-stable closed subset, then G · Y ⊂ G ∗H X is a closed subset.
Hence, the above maps establish a one-to-one correspondence between the G-stable closed
subsets of the homogeneous fiber bundle G ∗H X and the H-stable closed subsets of X.
Proof. Assertion (1) is an easy calculation and is left as an exercise (see Exercise 17).
It is clear that Z ∩ X is H-stable. Since any G-orbit intersects X, it follows that
Z = G · (Z ∩ X). If Y ⊂ X is a H-stable closed subset then the orbit G · Y is the image by
the canonical projection of the H-saturated closed subset G × Y , and hence it is a closed
subset.
Observation 6.8. In the algebraic situation, if we apply the construction of the induced
G-module to an object that is already a G-module, the transfer principle guarantees that
the induced module is isomorphic with k[G]H ⊗ M . As it could be expected, there is also
a geometric counterpart of this result. If the H-variety X in the definition of homogeneous
fiber bundle is already a G-variety, then G ∗H X ∼ = G/H × X. In the case that X is affine
and H exact in G, this can be obtained by duality as a direct consequence of the transfer
principle and the tensor identity (see Observation 7.7.17 and 7.7.15). This result generalizes
to a situation where the homogeneous fiber bundle exists; this is the content of the next
lemma.
Lemma 6.9. Let G be an affine algebraic group and H ⊂ G a closed subgroup, and
X a G-variety, and endow G/H × X with the diagonal G-action. Then there exists a G-
equivariant isomorphism Γ : G ∗H X → G/H × X that extends the identity IdX : X ∼ =
ρ−1 (1H) → {1H} × X ∼ = X.
Proof. It is easy to see that the morphism G × X → G/H × X, (g, x) 7→ (gH, gx) is
constant in the H-orbits. Indeed, (gh−1 H, gh−1 hx) = (gH, x). Let Γ : G ∗H X → G/H × X
be the induced morphism. Then Γ is a G-equivariant isomorphism, with inverse Γ−1 :
G/H × X → G ∗H X, Γ−1 (gH, x) = [g, g −1 x] (see Exercise 18).
Next we prove the existence of the homogeneous fiber bundle in the situation that H is
an exact subgroup in G and the variety X is affine. In this case the existence theorem 6.5
becomes a particular
case of a more general result concerning properties of the H-invariants
for H, k[G] -module algebras and their geometric counterparts; see [52].
Theorem 6.10. Let G be an affine algebraic group and H ⊂ G an exact subgroup. Let
H
S be an algebra object in (k[G],H) M that is finitely generated over k; then S is finitely
generated over k.
Proof. If U = Ru (H), then H/U is reductive and it follows that G/U ∼ =

(G/H)/(H/U ) is an affine algebraic variety. Therefore, U is exact in G by Theorem 12.6.7.
It follows from Theorem 12.5.5 that there exists a total integral σ : k[U ] → k[G], and, hence,
since S is a k[G]-algebra, one can think of σ as a total integral σ : k[U ] → S. Applying now
Theorem 12.8.2 wededuce that the categorical quotient Spm(S)/U exists and is affine; in
particular, U S = k (S/U ) is a finitely generated algebra. Since H S = H/U (U S) and H/U
is reductive, the result about the finite generation of the invariants follows.
of Lemma 2.6 to the situation where H ⊂ G
The following lemma is a generalization
is an exact subgroup and S is a k[G], H -module algebra. Recall that in accordance with
Theorem 12.5.5, the exactness assumption guarantees the existence of a family of Reynolds
operators, i.e., operators pM : M → H M indexed in M ∈ (H,k[G]) M, that are projections in
H
M , natural in M , and for f ∈ H k[G], pM (f m) = f pM (m) — see Section 12.5.
Lemma 6.11. Let G be an algebraic group and H exact in G. Let S be an algebra object
in (k[G],H) M.
Then:
(1) The extension H S ⊂ S is pure, i.e., for an arbitrary H
S-module M , the canonical
morphism ιM : M → S ⊗H S M is injective.
H
(2) The extension S ⊂ S is cyclically pure, i.e., if I ⊆ HS is an ideal, then I = IS ∩ HS.
(3) If J1 , J2 ⊂ S are H-stable ideals, then:
J1 ∩ HS + J2 ∩ HS = (J1 + J2 ) ∩ HS.
(4) If J1 + J2 = S, then J1 ∩ HS + J2 ∩ HS = HS.
Proof. (1) The map γM : S ⊗H S M → M given by γM (m ⊗H S s) = pS (s)m, where pS

denotes as usual the Reynolds operator, is clearly a section of ιM .
(2) It is well-known that in this situation, (1) implies (2) (just take M = H S/I).
(3) It is clear that J1 ∩ HS + J2 ∩ HS ⊆ (J1 + J2 ) ∩ HS. If we take an element s1 + s2 ∈
(J1 + J2 ) ∩ HS, and consider s1 + J1 ∩ J2 ∈ J1 /(J1 ∩ J2 ) and s2 + J2 ∩ J2 ∈ J2 /(J1 ∩ J2 ), from
the equality x · (s1 + s2 ) = (s1 + s2 ) we deduce that x · s1 − s1 = s2 − x · s2 ∈ J1 ∩ J2 . In other
H H
words, we have that s1 + J1 ∩ J2 ∈ J1 /(J1 ∩ J2 ) and s2 + J1 ∩ J2 ∈ J2 /(J1 ∩ J2 ) .
Using the Reynolds operator we find `1 ∈ HJ1 and `2 ∈ HJ2 such that `1 − s1 ∈ J1 ∩ J2 and
`2 − s2 ∈ J1 ∩ J2 . To find `1 (the same for `2 ) consider the morphism J1 → J1 /(J1 ∩ J2 )
and observe that we can use the commutative diagram
J1 / J1 /(J1 ∩ J2 )
pJ pJ1 /(J1 ∩J2 )
1

/
H H

J1 J1 /(J1 ∩ J2 )
that has surjective vertical arrows to find `1 as required. Then we write s1 + s2 = `1 + `2 + t

with t ∈ H(J1 ∩ J2 ). Hence, s1 + s2 ∈ HJ1 + GJ2 .
(4) This condition follows immediately from the previous one.
Theorem 6.12. Let G be an affine algebraic group and H ⊂ G an exact subgroup. Let
X be an affine algebraic H-variety, and consider the diagonal action of H on G × X. Then
the geometric quotient H \\(G × X) exists and ιX : X → G ∗H X is a closed immersion.
k[G × X] = k[G] ⊗ k[X] is clearly an object in

Proof. The algebra of regular functions
H
(k[G],H) M, and, therefore, k[G] ⊗ k[X] is a finitely generated algebra. Call its maximal
spectrum G ∗H X. In order to prove that the (affinized) quotient is semigeometric we only
need to follow the steps of Theorem 2.9, using Lemma 6.11 as a substitute for the results
of Lemma 2.6 (see Exercise 16).
Of the four conditions to check (see Definition 2.1), concerning the semigeometric prop-
erty we only deal with the following:
(1) the canonical map p : G × X → G ∗H X is constant along the orbits and surjective;
(3) if W ⊂ G × X is closed and H-stable, then π(W ) ⊂ G ∗H X is closed.
The first assertion of (1) is clear. The surjectivity of π means that given a maximal ideal
M ⊂ H (k[G] ⊗ k[X]) there is a maximal ideal N ⊂ (k[G] ⊗ k[X]) such that N ∩ H (k[G] ⊗
k[X]) = M . This follows easily if we take N maximal such that M (k[G] ⊗ k[X]) ⊂ N ⊂
k[G] ⊗ k[X]. This is possible because M (k[G] ⊗ k[X]) 6= k[G] ⊗ k[X]: otherwise if we have
equality of both terms, applying the Reynolds operator pk[G]⊗k[X] we obtain a contradiction.
To prove (3), let y ∈ (G ∗H X) \ π(W ); then the H-stable subsets W and π −1 (y) are
disjoint. If we call J1 and J2 their ideals in G × X, then J1 + J2 = k[G] ⊗ k[X]; using Lemma
6.11, we deduce the existence of f ∈ H J1 , g ∈ H J2 such that f + g = 1. Then f separates
π(W ) and y and this means that π(W ) is closed.
To prove that the quotient is not only semigeometric but geometric, we prove first that
the H-orbits in G × X are closed subsets of G × X. If (g, x) ∈ G × X, then OH (g, x) is
closed in gH × X. Indeed, OH (g, x) is the graph of the morphism gH → X, gh 7→ h−1 x
and, therefore, the orbit is closed in G × X; see also Exercise 15. The explicit description
above also shows that the orbits have the same dimension (equal to dim H). This implies —
in the notations of Definition 3.10 — that the set Ω(G × X) = G × X and hence it follows
from Theorem 3.13 that the semigeometric quotient is indeed geometric.
Next we prove that ιX is a closed immersion. Recall that Im(ιX ) = ρ−1 (1H) = p(H×X).
In order to prove that ιX is a closed immersion, we need to prove that
Ker (ιX )] = I ρ−1 (1H) ⊂ IndG
H
H k[X] = k[G] ⊗ k[X] ,

where (ιX )] (f ⊗g) = f (1)g. By definition, an element fi ⊗gi ∈ H k[G]⊗k[X] belongs to
P
Ker (ιX )] if and only if fi (1)gi (x) = 0 for all x ∈ X, if and only if fi (h)gi (h−1 ·x) = 0
P P
for all h ∈ H, x ∈ X. In other words,
Ker (ιX )] = I H × X ∩ H k[G] ⊗ k[X] = I p(H × X) = I ρ−1 (1H) .

Observation 6.13. Theorem 6.12 as well as Lemma 6.11 are particular cases of results
concerning linearly reductive actions of an algebraic group on an affine algebraic variety;
see [50] — where this theory was first developed — and [52], where the mentioned results
are proved.
7 Revisiting affinized quotients

In Chapter 7, Section 5, we introduced the concept of affinized quotient with a view
towards its applications in relation to the concept of observability. With the new elements of
geometric invariant theory introduced in this chapter, we can explore further this concept: if
the affinized quotient exists, i.e., if the algebra of invariants is finitely generated, additional
information can be obtained concerning the orbits and the fibers of the canonical projection
on the quotient.
For convenience in the notations, through this section we will consider right actions of
an algebraic group G on an affine variety X, and the corresponding left actions of G on
k[X].
Theorem 7.1. Let G be an affine algebraic group acting on an affine variety X, such
that G k[X] is finitely generated, and let π : X → X/aff G be the affinized quotient. Consider
the following conditions:
G G
(1) k[X] = k[X] .
(2) There exists a nonempty open subset W ⊂ X/aff G such that π −1 (x) = O(y), with
y ∈ X max , for all x ∈ W . Moreover, W can be taken of the form (X/aff G)f = Xf /aff G,
with f ∈ G k[X].
G G
Then condition (1) implies (2), and conversely: if (2) holds then k[X] ⊂ k[X] is
G G
a purely inseparable finite extension. In particular, if char k = 0, then k[X] = k[X] .
Proof. In order to prove the implication (1) =⇒ (2), let X0 ⊂ X be as in Rosenlicht’s

theorem 5.3, and take ρ : X0 → X0 //G to be the geometric quotient. Then we have the
following commutative diagram
/X
X0 (14.2)
ρ π

X0 //G / X/aff G
ϕ
Since
k(X0 //G) ∼
G
= G k(X0 ) = G k(X) =

k[X] = k[X/aff G] = k(X/aff G),
it follows that the geometric quotient X0 //G and the affinized quotient X/aff G are bi-
rationally equivalent via ϕ. Hence, there exists an open subset V ⊂ X0 //G such that
ϕ : V → X/aff G is an open immersion. Replacing X0 by U = π −1 (V ) we can assume that
V
ϕ is an open immersion, with ρ = π .
X0
Let V ⊂ X/aff G be an open subset such that, for each w ∈ V , the fiber π −1 (w) has
all its irreducible components of maximal dimension (as follows from Chevalley’s theorem
2.6.4). Let m = dim X − dim X/aff G be the dimension of such a fiber. It follows from
diagram (14.2) that m is the maximal dimension of the orbits of G on X. Moreover, if w ∈
W = (X0 //G) ∩ V ⊂ X/aff G, then there exists a unique irreducible component Z ⊂ π −1 (w)
such that Z = O(x), with x ∈ X0 . Let Y ⊂ π −1 (w) another irreducible component; then
Y ⊂ X \ X0 . Let A ⊂ W be the set of points whose fiber is reducible:
A = w ∈ W : π −1 (w) 6= π −1 (w) ∩ X0 .

Since A ⊂ π(X \ X0 ), it follows that if A is dense in X0 , then X \ X0 dominates X/aff G.

Hence, the maximal dimension of a fiber of π is
X\X0
dim(X \ X0 ) − dim X/aff G < dim X − dim X/aff G = m.

−1
But for a ∈ A, dim π (a) = m, which is a contradiction. Hence, for all w ∈
X\X0
W \ A 6= ∅, we have that π −1 (w) = O(x), with x ∈ X0 .

G
Let now f ∈ k[X] such that (X/aff G)f ⊂ W \ A. Then Xf satisfies assertion (2).
G
Assume now that (2) holds, with W = (X/aff G)f , f ∈ k[X]. Then π −1 (W ) = Xf ,
G G G G G G
k[X] = k[X]f = k[Xf ] and k[X] = k[X]f = k[Xf ] . Thus, we can
assume without loss of generality that X is such that π −1 (y) = G · x, with x ∈ X max , for
all y ∈ X/aff G. Let X0 ⊂ X be as in Rosenlicht’s theorem and consider the commutative
diagram

X0 /X
ρ π

X0 //G / X/aff G
ϕ
Let y, z ∈ X0 //G be such that ϕ(y) = ϕ(z). Then ρ−1 (y) and ρ−1 (z) are contained in
−1
π ϕ(y) , and hence ρ−1 (y) = ρ−1 (z). It follows that ϕ is injective. Since ϕ is dominant,
G G
it follows that ϕ induces a finite, purely inseparable extension k[X] ⊂ k[X] .
Observation 7.2. (1) Observe that if G is reductive, then G k[X] is finitely generated,
and thus Theorem 7.1 applies.
(2) Let X × G → X an action satisfying condition (1) of Theorem 7.1, and O, O0 ⊂
max
X ∩ Xf two orbits, with f as in condition (2) of that theorem. Then there exists
h ∈ k[Xf /aff G] = G k[X]f such that f π(O) = 0 and f π(O0 ) = 1. Hence G k[X]f ,

or, equivalently, the affinized quotient π : Xf → Xf /aff G separates orbits in X max ∩ Xf .

Corollary 7.3. Let G be an algebraic group acting on an affine algebraic variety,
G G G G
in such a way that k[X] = k[X] . Then there exists f ∈ k[X] such that k[X]f
max
separates the orbits in X ∩ Xf .
G
Proof. From Theorem 7.5.6, there exists h ∈ k[X] such that k[Xh ] = k[X]h is finitely
generated. Since
G G G
k[Xh ] = k[X] = k[Xh ]
G
it follows from Theorem 7.1 that there exists g ∈ G k[Xh ] = G k[X]h such that k (Xh )g =
G
k[X]gh separates the orbits in X max ∩ Xhg .
8 Further results on invariants of finite groups

Classical invariant theorists — see, for example, [185] — always considered “a first
fundamental problem”: to find the “fundamental invariants,” i.e., generators of the k-algebra
of all the invariants, and a “second fundamental problem” that would solve the question of
finding the relations between the fundamental invariants.
In particular, in the case of a finite group the first fundamental problem was solved
by E. Noether, who proved that there are a finite number of fundamental invariants with
degrees bounded by the order of the group (see [135] and [136]).
It is natural to look for linear actions of finite groups for which the fundamental invari-
ants have no relations. Around 1955, first in [168] by a case by case discussion, and then
in [25] using a general method that will be discussed in this section, it was proved that a
finite linear group over a field of characteristic zero is generated by reflections if and only
if the algebra of its invariants is generated by algebraically independent elements. From
a geometric viewpoint, this result means that the quotient of an affine space by a group
generated by reflections is again an affine space — of the same dimension.
In Exercise 22, we present an example of a finite group acting with nontrivial relations
between the fundamental invariants; see also [170, Chap. 4, Sect. 5] or [97]).
In this section we prove that if G is a finite group that acts linearly in the affine space kn
G
and is generated by reflections, then the algebra of invariants k[X1 , . . . , Xn ] is a polynomial
algebra in n variables. The interested reader may consult [170] for the converse. The proof
we present is extracted from [51].
The mentioned problem concerning the freeness of the algebra of invariants can be
formulated for arbitrary reductive groups. In this general context there is a close relationship
G
between this property and the freeness of the algebra k[V ] as a k[V ]-module. A third
ingredient that also enters into the picture is the equidimensionality property of all the
fibers of the morphism π : V → V /G. This equidimensionality property is automatically
satisfied in the case of a finite group. See [144] for an interesting discussion and a large
bibliography about these topics.
The reader that is interested in the general invariant theory of finite groups may look
at [170] and the references appearing there, or [169] for a more recent exposition — where,
for example, a proof of the converse of Theorem 8.7 is presented.
8.1 Invariants of graded algebras

The theorem that follows can be considered as an abstraction of the basic techniques
used in [25].
Theorem 8.1. Let S be a graded commutative k-algebra without zero divisors, and
P, ∆1 , ∆2 , . . . , ∆n : S → S a family of k-linear homogeneous maps such that:
(1) The maps ∆i are of degree −1, P is of degree 0 and P (1) = 1.
(2) The maps ∆i and P are S ∆ -linear, i.e., for all x ∈T
S ∆ and y ∈ S, P (xy) = xP (y) and
∆ n
∆i (xy) = x∆i (y), where ∆ = {∆1 , . . . , ∆n } and S = i=1 Ker(∆i ).
(3) If I ⊂ S is a homogeneous ∆i -stable ideal, i = 1, . . . , s, then I is P -stable and the
sequence
/ S/I / S/I ⊕ · · · ⊕ S/I
P
b c1 ⊕···⊕∆
∆ d n
S/I
is exact. Here Pb, ∆
c1 , . . . , ∆
cn denote the maps induced by P, ∆1 , . . . , ∆n on S/I.
Then S is a free homogeneously generated S ∆ -module with rank equal to dimk S/S+
∆
S.
Proof. Observe that if x, y ∈ S ∆ , then ∆i (xy) = x∆i (y) = x · 0 = 0 and S ∆ is a

graded k-subalgebra of S. As each of the ∆i has degree −1, 1 ∈ S ∆ . Moreover, if x ∈ S ∆ ,
then P (x) = P (x.1) = xP (1) = x. If we take I = 0, condition (3) implies that for all
i = 1, . . . , n, ∆i P = 0, i.e., P (x) ∈ S ∆ for all x ∈ S. Thus, P is a projection of S into the
subalgebra S ∆ .
∆
= k>0 Sk∆ and consider the ideal I = S+ ∆
L
Write S+ S ⊂ S. If {eα + I : α ∈ A} is a
k-basis of S/I, with eα homogeneous for α ∈ A, then B = {eα : α ∈ A} is a family of free
S ∆ -generators of S.
Indeed, call M the graded S ∆ -subalgebra of S generated by B. We prove by induction

on dP∈ N that Md = Sd . As {eα + I} generate S/I over k, there exist aα ∈ k such that
∆
1 − P α aα eα ∈ S+ S. If we look at the degree zero terms of the above equation we get
1− aα eα = 0, so that for at least one α we have that deg(eα ) = 0. Then
α:deg(eα )=0
M0 = k = S0 and the first step of the induction is established.
Suppose that Me = Se for all e < d. If f ∈ Sd , then there exist aα ∈ k such that
∆
P
f − P α aα eα ∈ P
I. Hence, there exist homogeneous elements rβ ∈ S+ , fβ ∈ S such that
f − α aα eα = β fβ rβ . Taking the terms of degree d on the above equation we obtain
X X
f− aα eα = fβ rβ .
{α:deg(eα )=d} {β:deg(fβ )=d−deg(rβ )}
As deg(rβ ) > 0, d − deg(rβ ) = deg(fβ ) < d, we deduce from the inductive hypothesis
that β, fβ ∈ M and this implies that f ∈ M .
Now we prove that B is free over S ∆ . First we prove that for any relation x1 y1 + x2 y2 +
· · · + xm ym = 0 with xi ∈ S ∆ , yi ∈ S and yi homogeneous, we have x1 ∈ S ∆ x2 + · · · + S ∆ xm
∆
or y1 ∈ S+ S = I. This implies that B is free over S ∆ . Indeed, given a relation of the form
x1 e1 +x2 e2 +· · ·+xm em = 0, with xi ∈ S ∆ , as e1 6∈ I we deduce that x1 = z2 x2 +· · ·+zm xm
with z2 , . . . , zm ∈ S ∆ . Hence,
(z2 x2 + · · · + zm xm )e1 + · · · + xm em = x2 (z2 e1 + e2 ) + · · · + xm (zm e1 + em ) = 0.
Suppose that z2 e1 + e2 ∈ I. Then, if deg(z2 ) = 0 we obtain a k-linear dependence

relation between e1 and e2 in S/I and this is impossible. If deg(z2 ) > 0 then z2 e1 ∈ I and
hence e2 ∈ I; this is also a contradiction. Then, z2 e1 + e2 6∈ I.
It follows that x2 ∈ S ∆ x3 + · · · + S ∆ xm . Iterating this procedure we obtain a relation
(t1 e1 +· · ·+tm−1 em−1 +em )xm = 0, with ti ∈ S ∆ , and then t1 e1 +· · ·+tm−1 em−1 +em = 0.
If we write ti = si +pi , with deg(si ) = 0 and deg(pi ) > 0, then s1 e1 +· · ·+sm−1 em−1 +em ∈
∆ ∆
S+ S, and this contradicts the k-independence of the ei modulo S+ S.
We have to prove that for any relation x1 y1 +x2 y2 +· · ·+xm ym = 0 with xi ∈ S ∆ , yi ∈ S,
then x1 ∈ S ∆ x2 + · · · + S ∆ xm or y1 ∈ S+
∆
S = I.
The proof proceeds by induction on the degree of y1 .
If y1 = 0, there is nothing to prove, and if y1 ∈ k∗ , we write x1 = x2 z2 + · · · + xm zm .
Applying P to this equation we obtain that x1 = P (x1 ) = x2 P (z2 ) + · · · + xm P (zm ) and
x1 ∈ S ∆ x2 + · · · + S ∆ xm .
Suppose that deg(y1 ) = d > 0 and that the assertion is established for all elements
of smaller degree. If we apply the operator ∆i to the original relation we get x1 ∆i (y1 ) +
x2 ∆i (y2 ) + · · · + xm ∆i (ym ) = 0. If ∆i (y1 ) = 0 for all i, as y1 has positive degree it belongs
∆
to S+ and we are done.
If for some i, ∆i (y1 ) 6= 0, as this element has degree d − 1 we conclude by induction
that x1 ∈ S ∆ x2 + · · · + S ∆ xm or ∆i (y1 ) ∈ S+ ∆
S for all i = 1, . . . , n. In the first case the
∆
proof is finished; assume that for all i, ∆i (y1 ) ∈ S+ S and consider the exact sequence
/ S/S+ / S/S+
P c1 ⊕···⊕∆
∆ d n
∆ ∆ ∆ ∆
b
S/S+ S S S ⊕ · · · ⊕ S/S+ S.
∆

As y1 = y1 + S+ S ∈ Ker ∆c1 ⊕ · · · ⊕ ∆
cn , then y1 ∈ Im(Pb). In other words, P (y1 ) − y1 ∈
∆ ∆ ∆
S+ S, and as y1 ∈ S+ , we deduce that P (y1 ) ∈ S+ S, so that y1 ∈ S+ S.
8.2 Polynomial subalgebras of polynomial algebras

In the previous paragraph we proved that S is a free S ∆ -module, generated by homo-
geneous elements. Next we prove that under additional hypothesis on the degrees of the
∆
generators of S+ , if S is a polynomial algebra so is S ∆ .
The proof we present is similar to the standard one (see, for example, [16] or [170]).
Theorem 8.2. Let k be a field of an arbitrary characteristic and R ⊂ S a graded
extension of commutative algebras without zero divisors. Assume that:
(1) The extension R ⊂ S is integral and flat.
(2) If I ⊂ R is a homogeneous ideal, then IS ∩ R = I.
(3) If we order in terms of the cardinals the family of all sets of homogeneous generators of
the ideal R+ ⊂ R, then there exists a minimal set in this family with the additional property
that the degrees of all its elements are prime with p, the characteristic exponent of k.
(4) The algebra S is polynomial and generated by elements of degree one.
Then R is a polynomial algebra with the same number of generators as S.
Proof. First observe that in the situation above, if S is integral over R, then S is
finitely generated as an R-module. Since S is a finitely generated k-algebra, it follows from
Theorem 1.2.4 that R is a finitely generated k-algebra. Hence R is Noetherian and R+ is
finitely generated as an R-module.
Let F = {f1 , . . . , fm } be a set of homogeneous generators such that p is prime with
di = deg(fi ). Then F generates R as a k-algebra (see Exercise 1.2 and the introduction to
Chapter 10). We want to prove that F is algebraically independent over k.
Assuming that there exists a nonzero homogeneous polynomial h ∈ k[X1 , . . . , Xm ] such
that h(f1 , . . . , fm ) = 0 we may take it of minimal degree. Call hi = ∂h/∂Xi ∈ k[X1 , . . . , Xm ]
its partial derivatives for i = 1, . . . , m.
Let J = h1 (f1 , . . . , fm ), . . . , hm (f1 , . . . , fm ) R ⊂ R be the ideal generated by
hi (f1 , . . . , fm ), i = 1, . . . , m. From the hypothesis on the minimality of the degree of h
we deduce that at least one of the generators hi (f1 , . . . , fm ), i = 1, . . . , m of the ideal J is
nonzero. Hence we can assume that there exists a minimal 1 ≤ s ≤ m, such that
J = h1 (f1 , . . . , fm ), . . . , hs (f1 , . . . , fm ) R
,
Ps for all s + 1 ≤ j ≤ m, there exists ri,j ∈ R such that hj (f1 , . . . , fm ) =

and
i=1 ri,j hi (f1 , . . . , fm ).
Write the polynomial algebra S as S = k[T1 , . . . , Tn ], with deg Tl = 1, l = 1, . . . , n. If
we differentiate h(f1 , . . . , fm ) = 0 with respect to Tl , l = 1, . . . , n we have:
m
X
0 = ∂/∂Tl h(f1 , . . . , fm ) = hi (f1 , . . . , fm )∂fi /∂Tl =
i=1
s
X m X
X s
hi (f1 , . . . , fm )∂fi /∂Tl + ri,j hi (f1 , . . . , fm ) ∂fj /∂Tl = (14.3)
i=1 j=s+1 i=1
Xs Xm
hi (f1 , . . . , fm ) ∂fi /∂Tl + ri,j ∂fj /∂Tl .
i=1 j=s+1
Pm
In order to prove that ∂fi /∂Tl + j=s+1 ri,j ∂fj /∂Tl ∈ R+ S, consider the R-linear map
s
φ : Rs → R, φ(r1 , . . . , rs ) = s
P
i=1 hi (f1 , . . . , fm )ri , and its extension φS : S → S. By
definition, J = Im(φ), and since S is a flat R-module by condition (1), we deduce that the
short exact sequence
0 / Ker(φ) / Rs φ
/J /0
remains exact after tensoring with S. Hence
0 / Ker(φ)S / Ss φS
/ JS /0
is exact and Ker(φS ) = Ker(φ)S.

s
Ps In order to prove that Ker(φ) ⊂ R+ , consider (r1 , . . . , rs ) ∈ R such that
i 6= 0, then its degree zero component equals zero; other-
i=1 hi (f1 , . . . , fm )ri = 0. If rP
s
wise, we would have a relation i=1 hi (f1 , . . . , fm )λi = 0, with λi ∈ k and λi 6= 0 for some
i. This contradicts the minimality of s, the number of generators of the ideal J.
It follows from Equation 14.3 that
m
X
∂fi /∂Tl + ri,j ∂fj /∂Tl ∈ Ker(φS ) = Ker(φ)S ⊂ R+ S.
j=s+1
Hence, as f1 , . . . , fm generate R+ , there exist si,l,j ∈ S, 1 ≤ j ≤ m, 1 ≤ i ≤ s, 1 ≤ l ≤ n,

such that for all i, l we have that
m
X m
X
∂fi /∂Tl + ri,j ∂fj /∂Tl = si,l,j fj . (14.4)
j=s+1 j=1
Multiplying in Equation 14.4 both sides by Tl and adding all the terms we obtain that
for all 1 ≤ i ≤ s,
n
X m
X n
X m X
n
X
Tl ∂fi /∂Tl + ri,j Tl ∂fj /∂Tl = Tl si,l,j fj .
l=1 j=s+1 l=1 j=1 l=1
Pn Pn
Clearly, ti,j = l=1 Tl si,l,j ∈ S+ and, using Euler’s relation di fi = l=1 Tl ∂fi /∂Tl , we
obtain that for 1 ≤ i ≤ s,
m
X s
X m
X
di fi + ri,j dj fj = ti,j fj + ti,j fj ,
j=s+1 j=1 j=s+1
and hence we can rewrite the above equation as

s
X m
X
d i fi − ti,j fj = (ti,j − dj ri,j )fj .
j=1 j=s+1
In particular,
d1 f1 − t1,1 f1 = t1,2 f2 + · · · + t1,s fs + (t1,s+1 − ds+1 r1,s+1 )fs+1 + · · · + (t1,m − dm r1,m )fm ,
and as t1,1 f1 has no homogeneous term of degree d1 , it follows that d1 f1 ∈ Sf2 + · · · + Sfm .
Condition (3) of the hypothesis implies that f1 ∈ (Sf2 + · · · + Sfm ) ∩ R, and from
condition (2) we deduce that f1 ∈ Rf2 + · · · + Rfm and this contradicts the minimality of
{f1 , . . . , fm }.
Since R ⊂ S is an integral finitely generated extension of polynomial algebras, it follows
that R and S have the same number of generators — apply, for example, Theorem 1.2.6.
Observation 8.3. Observe that in the case that the base field is of characteristic zero,
condition (3) of Theorem 8.2 is automatically satisfied.
The following lemma guarantees that if S is freely finitely generated by homogeneous

elements as an R-module, then condition (1) and (2) of Theorem 8.2 are satisfied.
Lemma 8.4. Let k be of arbitrary characteristic and R ⊂ S be a graded extension of
k-algebras. Assume that the R-module S has a finite basis of homogeneous elements. Then
the extension R ⊂ S is integral and flat. Moreover, if I ⊂ R is a homogeneous ideal, then
IS ∩ R = I.
Proof. It is clear that the extension R ⊂ S is integral and flat.

Let {s1 , . . . , sr } be a homogeneous finite basis of S over R and 1 = b1 s1 + · · · + br sr ,
with bi ∈ R. It is clear that there exists i such that deg si = 0; therefore we can assume
that {1, s2 , . . . , sr } is an homogeneous basis.
P
If I ⊂ R is P a homogeneous ideal, let ξ ∈ IS ∩ R and write ξ = aj tj , with aj ∈ I,
tj ∈ S. As tj = bi,j si , with bi,j ∈ R,
X X
ξ= aj bi,j si .
i j
P P P
Moreover, as ξ ∈ R, then ξ = ξ1 = a b 1 + i>1 j aj bi,j si . Being
P j j 1,j
{1, s2 , . . . , sr } an R- basis, it follows that ξ = j aj b1,j ∈ I.
We can summarize the main results obtained until now as follows (see theorems 8.1 and
8.2).
Theorem 8.5. Let S = k[T1 , . . . , Tn ] be a polynomial algebra with its usual grading and
assume that there exist k-linear homogeneous operators P, ∆1 , ∆2 , . . . , ∆n : S → S such
that:
(1) The operators ∆i are of degree −1, P is of degree 0 and P (1) = 1.
(2) The operators ∆i and P are S ∆ -linear.
(3) Every homogeneous ∆i -stable ideal I ⊂ S, i = 1, . . . , n, is P -stable and the sequence
/ S/I / S/I ⊕ · · · ⊕ S/I

P
b c1 ⊕···⊕∆
∆ d n
S/I
is exact.
∆
(4) The k-space S/S+ S is finite dimensional.
(5) If we order in terms of the cardinals the family of all sets of homogeneous generators
∆
of the ideal S+ ⊂ S ∆ , then there exists a minimal set in this family with the additional
property that the degrees of all its elements are prime with p, the characteristic exponent of
k.
Then S ∆ is a polynomial algebra in the same number of variables as S.
8.3 The case of a group generated by reflections

In this paragraph we show that if char k = 0, then the theorem of Chevalley–Shephard–
Todd can be proved using the methods developed above.
We start by reviewing some well-known definitions and constructions. An invertible
linear map T : kn → kn is called a pseudo-reflection if it is diagonalizable and dimk Ker(T −
Id) = n − 1. Call fT : kn → k a linear functional dual to Ker(T − Id), aT the T -eigenvalue

different from 1 and vT the aT -eigenvector of T such that fT (vT ) = 1. Then T = Id +(aT −
1)fT |vT and T −1 = Id +(a−1
T − 1)fT |vT .
The action of T on kn induces an action of T on (kn )∗ by the formula (T · f )(v) =
f (T v), i.e., T · f = f + (a−1
−1
T − 1)f (vT )fT . This action can be extended to the symmetric
algebra S = S (kn )∗ , and the following result can be easily proved by induction (see
Exercise 23).
Lemma 8.6. In the situation above there exists a linear operator ∆T : S → S, homoge-
neous of degree −1, such that for all ξ ∈ S,
T · ξ − ξ = fT ∆T (ξ) and ∆T (ξη) = ξ∆T (η) + η∆T (ξ) + fT ∆T (ξ)∆T (η).
Let G ⊂ GLn (k) be a finite linear group. Let s ∈ G be a pseudo-reflection in kn and

∆s : S → S as above. Let {s1 , . . . , sm } ⊂ G be the set of pseudo-reflections in G and call
∆1 , . . . , ∆m the corresponding operators ∆i = ∆si .
First we prove that S ∆ = G S.
S ∆ = ξ ∈ S : ∆i (ξ) = 0 , i = 1, . . . , m = {ξ ∈ S : si · ξ = ξ , i = 1, . . . , m} =

{ξ ∈ S : g · ξ = ξ ∀ g ∈ G} = G S.
Next we prove that if G is generated by s1 , . . . , sm , then the operators P =

RS , ∆1 , . . . , ∆m satisfy the hypothesis of Theorem 8.5 — RS is the Reynolds operator.
(1) This condition follows from the definitions.
(2) It is clear that P is G S-linear, and if we apply the formula ∆i (ξη) = ξ∆i (η) + ∆i (ξ)η +
fi ∆i (ξ)∆i (η) for ξ ∈ S, η ∈ S G , we conclude that the operators ∆i are G S-linear.
(3) If I is a ∆i -stable ideal, then si · I ⊂ I for all i = 1, . . . , m and hence I is G-stable.
If ξ + I is such that ∆i (ξ) ∈ I for i = 1, . . . , m, then si · ξ − ξ ∈ I for all i. As the set
{si : i = 1, . . . , m} generate G, we conclude that g · ξ − ξ ∈ I, for all g ∈ G. Adding all the
above relations we conclude that P (ξ) − ξ ∈ I and ξ + I ∈ Im(Pb). Hence the sequence:
/ S/I / S/I ⊕ · · · ⊕ S/I

P
b c1 ⊕···⊕∆
∆ d n
S/I
is exact.
(4) This condition follows from the fact that the extension G S ⊂ S is integral (see Theorem
7.6.2).
Condition (5) is automatically satisfied, since we are in the characteristic zero case.
Then, we can apply Theorem 8.5 and obtain the following result.
Theorem 8.7 (Chevalley–Shephard–Todd). Let char k = 0 and assume that G ⊂
GLn is a finite subgroup generated by pseudo-reflections. Then the algebra of invari-
G
ants k[X1 , . . . , Xn ] ⊂ k[X1 , . . . , Xn ] is a polynomial algebra in n indeterminates. More-
over, (kn , π) is the geometric quotient, where π : kn → kn is the morphism π(x) =
f1 (x), . . . , fn (x) , with {f1 , . . . , fn } a set of algebraically independent generators of the
G
k-algebra k[X1 , . . . , Xn ].
Observation 8.8. In Exercise 24 we apply Theorem 8.5 to a situation where the
polynomial subalgebra is not constructed as the invariants of a group.
Example 8.9. Consider the symmetric group Sn acting on kn by permutation of the
Another random document with
no related content on Scribd:
E. Smith, “The Teaching of Arithmetic,” Teachers College Record,
Vol. X, No. 1.
[14] E. L. Thorndike, “Handwriting,” Teachers College Record,
Vol. XI, No. 2; Stone, Arithmetical Abilities and Some of the
Factors Determining them.
[15] Quoted by Johnson in a monograph on “The Problem of
Adapting History to Children in the Elementary School,” Teachers
College Record, Vol. IX, p. 319.
[16] Teachers College Record, Vol. IX, pp. 319-320.
[17] “Stenographic Reports of High School Lessons,” Teachers
College Record, September, 1910, pp. 18-26.
[18] Baldwin, Industrial School Education. A most helpful
discussion of industrial work.
[19] W. S. Jackman, “The Relation of School Organization to
Instruction,” The Social Education Quarterly, Vol. I, pp. 55-69;
Scott, Social Education.
[20] Allen, Civics and Health, p. 53.
[21] Dewey, Moral Principles in Education.
[22] See chapter on Social Phases of the Recitation.
[23] Moral Training in the Public Schools, p. 41. The essay by
Charles Edward Rugh.
[24] Bagley, Classroom Management, Chapter XIV.
[25] See discussion of the study lesson, ante.
[26] McMurry, How to Study, Chapter III.
[27] See ante, Chapter XI.
[28] Adapted from a plan prepared by Lida B. Earhart, Ph.D., for
the author’s syllabus on Theory and Practice of Teaching.
[29] Some discussion of the course of study as an instrument in
supervision is given in the chapter on “The Teacher in Relation to
the Course of Study.”
[30] For a discussion of the doctrine of formal discipline, and for
bibliography, see Thorndike, Educational Psychology, 1903
edition, Chapter VIII; Heck, Mental Discipline.
[31] James E. Russell, “The School and Industrial Life,”
Educational Review, Vol. XXXVIII, pp. 433-450.
[32] E. L. Thorndike, “Handwriting,” Teachers College Record,
Vol. XI, No. 2.
[33] Cubberley, School Funds and their Apportionment; Elliott,
Fiscal Aspects of Education; Strayer, City School Expenditures.
[34] In proceeding to the part of the study that is necessarily
largely composed of tables, it may be well to state the position of
the author regarding the partial interpretations offered in
connection with the tables. It is that the entire tables give by far
the best basis for conclusions; that for a thorough comprehension
of the study they should be read quite as fully as any other part;
and that they should be regarded as the most important source of
information rather than the brief suggestive readings which are
liable to give erroneous impressions, both because of the
limitations of a single interpretation and the lack of space for
anything like full exposition.
[35] M = Median, which is the representation of central tendency
used throughout this study. It has the advantages over the
average of being more readily found, of being unambiguous, and
of giving less weight to extreme or erroneous cases.
[36] For reliability of measures of reasoning ability, see Appendix,
p. 100.
[37] As stated in Part I, p. 17, a score is arbitrarily set at one. The
fact that the zero point is unknown in both reasoning and
fundamentals makes these scores less amenable to ordinary
handling than they might at first thought seem. Hence, entire
distributions are either printed or placed on file at Teachers
College.
[38] For the data from which these calculations were made, see
first column of table XXI, p. 52, and the first columns of tables III
and IV, p. 21. The absence of known zero points makes such
computations inadvisable except in connection with the more
reliable evidence of the preceding table.
[39] And it is the opinion of the author that the chances are much
better that one would get a school with a superior product in
education.
The following pages contain advertisements of Macmillan
books on education, pedagogy, etc.
A Cyclopedia of Education
Edited by PAUL MONROE, Ph.D.
Professor of the History of Education. Teachers College, Columbia University,
Author of “A Text-Book in the History of Education,” “Brief
Course in the History of Education,” etc.
The need of such work is evidenced: By the great mass of varied educational literature
showing an equal range in educational practice and theory; by the growing importance
of the school as a social institution, and the fuller recognition of education as a social
process; and by the great increase in the number of teachers and the instability of
tenure which at the same time marks the profession.
The men who need it are: All teachers, professional men, editors, ministers, legislators, all
public men who deal with large questions of public welfare intimately connected with
education—every one who appreciates the value of a reference work which will give
him the outlines of any educational problem, the suggested solutions, the statistical
information, and in general the essential facts necessary to its comprehension.
Among the departmental Editors associated with Dr. Monroe are Dr. Elmer E. Brown,
U. S. Commissioner of Education, Prof. E. F. Buchner, of Johns Hopkins, Dr. WM. H.
Burnham, Clark University, M. Gabriel Compayré, Inspector-General of Public
Instruction, Paris, France, Prof. Wilhelm Münch, of Berlin University, Germany, Prof.
John Dewey, of Columbia University, Dr. Ellwood P. Cubberly, Stanford University,
Cal., Prof. Foster Watson, of the University College of Wales, Dr. David Snedden,
Commissioner of Education for the State of Massachusetts, and others.
Send for a descriptive circular and list of contributors to Volume I
To be completed in five large octavo volumes, each $5.00
T H E M A C M I L L A N C O M PA N Y
64-66 Fifth Avenue, New York
A LIST OF BOOKS FOR TEACHERS
Published by The Macmillan Company
ADAMS, John. Exposition and Illustration in Teaching.

Cloth. viii + 428 pages. $1.25
ARMSTRONG, Henry E. The Teaching of Scientific Method and Other Papers
on Education.
Cloth. xxvii + 504 pages. $1.75
ARNOLD, Felix. A Text-book of School and Class Management. I. Theory and
Practice.
Cloth. 12mo. xxii + 409 pages. Index. $1.25
II. Administration and Hygiene. Cloth. xii + 292 pages. $1.00
—Attention and Interest. Cloth. viii + 272 pages. $1.00
BAGLEY, William Chandler. Classroom Management: Its Principles and
Technique. By William Chandler Bagley, Director, School of Education,
University Of Illinois.
Cloth. 12mo. xvii + 352 pages. $1.25
—The Educative Process. Cloth. 12mo. xix + 358 pages. $1.25
BROWN, John Franklin. The American High School. By John Franklin Brown,
Ph.D., formerly Professor in Education and Inspector of High Schools for the
State University of Iowa.
Cloth. xii + 498 pages. 12mo. $1.25
BUTLER, Nicholas Murray. The Meaning of Education, and Other Essays
and Addresses. By Nicholas Murray Butler, President of Columbia University.
Cloth. 12mo. xii + 230 pages. $1.00
CHUBB, Percival. The Teaching of English. By Percival Chubb, Principal of
High School Department, Ethical Culture School, New York.
COLLAR, George, and CROOK, Charles W. School Management and
Methods of Instruction. By George Collar and Charles W. Crook, London.
Cloth. 12mo. viii + 336 pages. $1.00
CRONSON, Bernard. Methods in Elementary School Studies. By Bernard
Cronson, A.B., Ph.D., Principal of Public School No. 3, Borough of Manhattan,
City of New York.
Cloth. 12mo. 167 pages. $1.25
—Pupil Self-Government. Cloth. 12mo. ix + 107 pages. $.90
CUBBERLEY. Syllabus of Lectures on the History of Education. With Selected
Bibliographies and Suggested Readings. By Ellwood P. Cubberley. Second
Edition, revised and enlarged. In two parts.
Part I, v + 129 pages, $1.50 net; Part II, xv + 361 pages, $1.50
Complete in one volume, $2.60
DE GARMO, Charles. Interest and Education. By Charles De Garmo, Professor
of the Science and Art of Education in Cornell University.
—The Principles of Secondary Education.
Vol. I, Studies. Cloth. 12mo. xii + 299 pages. $1.25
Vol. II, Processes of Instruction. xii + 200 Pages. $1.00
Vol. III, Ethical Training. x + 220 pages. $1.00
DEXTER, Edwin Grant. A History of Education in the United States. By Edwin
Grant Dexter, Professor of Education in the University of Illinois.
Cloth. xxi + 665 pages. 8vo. $2.00
HUEY, Edmund B. The Psychology and Pedagogy of Reading. By Professor
Edmund B. Huey, of the Western University of Pennsylvania.
Cloth. 12mo. xvi + 469 pages. $1.40
JONES, Olive M., LEARY, Eleanor G., and QUISH, Agnes E. Teaching
Children to Study. The Group System applied.
Illustrated. Cloth. viii + 193 pages. 12mo. $.80
KILPATRICK, Van Evrie. Departmental Teaching in Elementary Schools.
Cloth. 12mo. xiii + 130 pages, 16mo. $.60
KIRKPATRICK, Edwin A. Fundamentals of Child Study. By Professor Edwin A.
Kirkpatrick, Principal of State Normal School, Fitchburg, Mass.
Cloth. 12mo. xxi + 384 pages. $1.25
—Genetic Psychology. Cloth, xv + 373 pages. $1.25
LAURIE, S. S. Institutes of Education. 3d ed. Cloth. xii + 391 pages. $1.90
MAJOR, David R. First Steps in Mental Growth. A Series of Studies in the
Psychology of Infancy. By David R. Major, Professor of Education in the Ohio
State University.
Cloth, xiv + 360 pages. 12mo. $1.25
THE McMURRY SERIES Each, cloth, 12mo.

General Method
—The Elements of General Method. By Charles A. McMurry.
323 pages. $.90
—The Method of the Recitation. By Charles A. McMurry and Frank M. McMurry,
Professor of the Theory and Practice of Teaching, Teachers College,
Columbia University.
xi + 329 pages. $.90
Special Method. By Charles A. McMurry.
—Special Method in Primary Reading and Oral Work with Stories.
vii + 103 pages. $.60
—Special Method in the Reading of English Classics.
vi + 254 pages. $.75
—Special Method in Language in the Eight Grades.
viii + 192 pages. $.70
—Special Method in History.
vii + 291 pages. $.75
—Special Method in Arithmetic.
vii + 225 pages. $.70
—Special Method in Geography.
xi + 217 pages. $.70
—Special Method in Elementary Science.
ix + 275 pages. $.75
—Nature Study Lessons for Primary Grades. By Mrs. Lida B. McMurry, with an
Introduction by Charles A. McMurry.
xi + 191 pages. $.60
Course of Study in the Eight Grades.
Vol. I. Grades I to IV. vii + 236 pages. $.75
Vol II. Grades V to VIII. v + 226 pages. $.75
MONROE, Paul. A Brief Course in the History of Education. By Paul Monroe,
Ph.D., Professor in the History of Education, Teachers College, Columbia
University.
Cloth. 8vo. xviii + 409 pages. $1.25
The Philosophy of Education

By HERMAN HARRELL HORNE, Ph.D.
Professor of the History of Philosophy and of the History of Education, New York
University
Cloth, 8vo, xvii + 295 pages, $1.50
A connected series of discussions on the foundations of education in the related
sciences of biology, physiology, sociology, and philosophy, and a thoroughgoing
interpretation of the nature, place, and meaning of education in our world. The newest
points of view in the realms of natural and mental science are applied to the
understanding of educational problems. The field of education is carefully divided, and
the total discussion is devoted to the philosophy of education, in distinction from its
history, science, and art.
The Psychological Principles of Education

Cloth, 12mo, xiii + 435 pages, $1.75
The relationship of this book to the author’s “Philosophy of Education” is that, whereas
the first was mostly theory with some practice, this is mostly practice with some theory.
This volume lays the scientific foundations for the art of teaching so far as those
foundations are concerned with psychology. The author is the “middleman” between the
psychologist and the teacher, taking the theoretical descriptions of pure psychology and
transforming them into educational principles for the teacher. In the Introduction the
reader gets his bearings in the field of the science of education. The remainder of the
book sketches this science from the standpoint of psychology, the four parts of the
work, Intellectual Education, Emotional Education, Moral Education, and Religious
Education, being suggested by the nature of man, the subject of education. A special
feature is the attention paid to the education of the emotions and of the will.
Idealism in Education
Or First Principles in the Making of Men and Women
Author of “The Philosophy of Education” and “The Psychological Principles of
Education”
Cloth, 12mo, xxi + 183 pages, index, $1.25 by mail, $1.34
Professor Horne here discusses three things which he regards as fundamental in the
building of human character,—Heredity, Environment, and Will. His method of handling
these otherwise heavy subjects, makes the book of interest, even to the general reader.
By WILLIAM CHANDLER BAGLEY
Director of the School of Education, University of Illinois
Craftsmanship in Teaching
Cloth, 12mo, 247 pages, $1.25
Readers of “The Educative Process” and “Classroom Management” by Director W. C.
Bagley of the University of Illinois will welcome the author’s new book on
“Craftsmanship in Teaching.” The book is made up of a series of addresses given
before educational gatherings, the subject of the first one giving the book its name. In
these addresses the personality of the author is more in evidence than is possible in his
more systematic work, but the same sane, scientific point of view is apparent
throughout.
Classroom Management
Cloth, xvii + 332 pages, $1.25
This book considers the problems that are consequent upon the massing of children
together for purposes of instruction and training. It aims to discover how the unit-group
of the school system—the “class”—can be most effectively handled. The topics
commonly included in treatises upon school management receive adequate attention;
the first day of school; the mechanizing of routine; the daily programme; discipline and
punishment; absence and tardiness, etc.
The Educative Process

Cloth, xix + 358 pages, $1.25
The book aims to prevent a waste of energy on the part of the young teacher by setting
forth a systematic and comprehensive view of the task that is to be accomplished by
the school, with the working principles for the attainment of the end. The best idea for
the author’s plan of treatment can be had from his division of the book. Part I discusses
the function of education and of the school in biological and sociological terms. Part II
continues the same topic from the psychological standpoint. Part III deals with the
functioning of experience in its relation to the educative process. Part IV treats of the
relation of education to the three periods of child-development: the transitional, the
formative, the adolescent. Part V considers educational values and the necessity of
ideals in the educative process, and Part VI concludes with the technique of teaching.
TRANSCRIBER’S NOTE
Obvious typographical errors and punctuation errors have been corrected after
careful comparison with other occurrences within the text and consultation of
external sources.
Some hyphens in words have been silently removed, some added, when a
predominant preference was found in the original book.
Except for those changes noted below, all misspellings in the text, and
inconsistent or archaic usage, have been retained.
Pg 69: ‘a singe problem’ replaced by ‘a single problem’.
Pg 113: ‘Professon Johnson’ replaced by ‘Professor Johnson’.
Pg 136: ‘find situtions’ replaced by ‘find situations’.
Pg 150: ‘actally demanded’ replaced by ‘actually demanded’.
Pg 189: ‘was comformable’ replaced by ‘was conformable’.
Pg 236: ‘genuine motive’ replaced by ‘genuine motives’.
Pg 244: ‘I. Abstract.’ replaced by ‘II. Abstract.’.
Pg 258: The note ‘Footnotes on opposite page.’ has been removed from the
bottom of TABLE III.
Pg 260: ‘XXII’ (first row in the table) replaced by ‘XXIII’.
Pg 272: ‘Syntax of etymology’ replaced by ‘Syntax or etymology’.
Pg 273: ‘c.’ inserted in front of ‘The Influence of the’.
*** END OF THE PROJECT GUTENBERG EBOOK A BRIEF
COURSE IN THE TEACHING PROCESS ***
Updated editions will replace the previous one—the old editions

will be renamed.
Creating the works from print editions not protected by U.S.

copyright law means that no one owns a United States copyright
in these works, so the Foundation (and you!) can copy and
distribute it in the United States without permission and without
paying copyright royalties. Special rules, set forth in the General
Terms of Use part of this license, apply to copying and
distributing Project Gutenberg™ electronic works to protect the
PROJECT GUTENBERG™ concept and trademark. Project
Gutenberg is a registered trademark, and may not be used if
you charge for an eBook, except by following the terms of the
trademark license, including paying royalties for use of the
Project Gutenberg trademark. If you do not charge anything for
copies of this eBook, complying with the trademark license is
very easy. You may use this eBook for nearly any purpose such
as creation of derivative works, reports, performances and
research. Project Gutenberg eBooks may be modified and
printed and given away—you may do practically ANYTHING in
the United States with eBooks not protected by U.S. copyright
law. Redistribution is subject to the trademark license, especially
commercial redistribution.
START: FULL LICENSE

THE FULL PROJECT GUTENBERG
LICENSE
PLEASE READ THIS BEFORE YOU DISTRIBUTE OR USE THIS WORK
To protect the Project Gutenberg™ mission of promoting the

free distribution of electronic works, by using or distributing this
work (or any other work associated in any way with the phrase
“Project Gutenberg”), you agree to comply with all the terms of
the Full Project Gutenberg™ License available with this file or
online at www.gutenberg.org/license.
Section 1. General Terms of Use and

Redistributing Project Gutenberg™
electronic works
1.A. By reading or using any part of this Project Gutenberg™
electronic work, you indicate that you have read, understand,
agree to and accept all the terms of this license and intellectual
property (trademark/copyright) agreement. If you do not agree to
abide by all the terms of this agreement, you must cease using
and return or destroy all copies of Project Gutenberg™
electronic works in your possession. If you paid a fee for
obtaining a copy of or access to a Project Gutenberg™
electronic work and you do not agree to be bound by the terms
of this agreement, you may obtain a refund from the person or
entity to whom you paid the fee as set forth in paragraph 1.E.8.
1.B. “Project Gutenberg” is a registered trademark. It may only

be used on or associated in any way with an electronic work by
people who agree to be bound by the terms of this agreement.
There are a few things that you can do with most Project
Gutenberg™ electronic works even without complying with the
full terms of this agreement. See paragraph 1.C below. There
are a lot of things you can do with Project Gutenberg™
electronic works if you follow the terms of this agreement and
help preserve free future access to Project Gutenberg™
electronic works. See paragraph 1.E below.
1.C. The Project Gutenberg Literary Archive Foundation (“the
Foundation” or PGLAF), owns a compilation copyright in the
collection of Project Gutenberg™ electronic works. Nearly all the
individual works in the collection are in the public domain in the
United States. If an individual work is unprotected by copyright
law in the United States and you are located in the United
States, we do not claim a right to prevent you from copying,
distributing, performing, displaying or creating derivative works
based on the work as long as all references to Project
Gutenberg are removed. Of course, we hope that you will
support the Project Gutenberg™ mission of promoting free
access to electronic works by freely sharing Project
Gutenberg™ works in compliance with the terms of this
agreement for keeping the Project Gutenberg™ name
associated with the work. You can easily comply with the terms
of this agreement by keeping this work in the same format with
its attached full Project Gutenberg™ License when you share it
without charge with others.
1.D. The copyright laws of the place where you are located also
govern what you can do with this work. Copyright laws in most
countries are in a constant state of change. If you are outside
the United States, check the laws of your country in addition to
the terms of this agreement before downloading, copying,
displaying, performing, distributing or creating derivative works
based on this work or any other Project Gutenberg™ work. The
Foundation makes no representations concerning the copyright
status of any work in any country other than the United States.
1.E. Unless you have removed all references to Project

Gutenberg:
1.E.1. The following sentence, with active links to, or other

immediate access to, the full Project Gutenberg™ License must
appear prominently whenever any copy of a Project
Gutenberg™ work (any work on which the phrase “Project
Gutenberg” appears, or with which the phrase “Project
Gutenberg” is associated) is accessed, displayed, performed,
viewed, copied or distributed:
This eBook is for the use of anyone anywhere in the United

States and most other parts of the world at no cost and with
almost no restrictions whatsoever. You may copy it, give it
away or re-use it under the terms of the Project Gutenberg
License included with this eBook or online at
www.gutenberg.org. If you are not located in the United
States, you will have to check the laws of the country where
you are located before using this eBook.
1.E.2. If an individual Project Gutenberg™ electronic work is

derived from texts not protected by U.S. copyright law (does not
contain a notice indicating that it is posted with permission of the
copyright holder), the work can be copied and distributed to
anyone in the United States without paying any fees or charges.
If you are redistributing or providing access to a work with the
phrase “Project Gutenberg” associated with or appearing on the
work, you must comply either with the requirements of
paragraphs 1.E.1 through 1.E.7 or obtain permission for the use
of the work and the Project Gutenberg™ trademark as set forth
in paragraphs 1.E.8 or 1.E.9.
1.E.3. If an individual Project Gutenberg™ electronic work is

posted with the permission of the copyright holder, your use and
distribution must comply with both paragraphs 1.E.1 through
1.E.7 and any additional terms imposed by the copyright holder.
Additional terms will be linked to the Project Gutenberg™
License for all works posted with the permission of the copyright
holder found at the beginning of this work.
1.E.4. Do not unlink or detach or remove the full Project

Gutenberg™ License terms from this work, or any files
containing a part of this work or any other work associated with
Project Gutenberg™.
1.E.5. Do not copy, display, perform, distribute or redistribute
this electronic work, or any part of this electronic work, without
prominently displaying the sentence set forth in paragraph 1.E.1
with active links or immediate access to the full terms of the
Project Gutenberg™ License.
1.E.6. You may convert to and distribute this work in any binary,
compressed, marked up, nonproprietary or proprietary form,
including any word processing or hypertext form. However, if
you provide access to or distribute copies of a Project
Gutenberg™ work in a format other than “Plain Vanilla ASCII” or
other format used in the official version posted on the official
Project Gutenberg™ website (www.gutenberg.org), you must, at
no additional cost, fee or expense to the user, provide a copy, a
means of exporting a copy, or a means of obtaining a copy upon
request, of the work in its original “Plain Vanilla ASCII” or other
form. Any alternate format must include the full Project
Gutenberg™ License as specified in paragraph 1.E.1.
1.E.7. Do not charge a fee for access to, viewing, displaying,

performing, copying or distributing any Project Gutenberg™
works unless you comply with paragraph 1.E.8 or 1.E.9.
1.E.8. You may charge a reasonable fee for copies of or

providing access to or distributing Project Gutenberg™
electronic works provided that:
• You pay a royalty fee of 20% of the gross profits you derive from
the use of Project Gutenberg™ works calculated using the
method you already use to calculate your applicable taxes. The
fee is owed to the owner of the Project Gutenberg™ trademark,
but he has agreed to donate royalties under this paragraph to
the Project Gutenberg Literary Archive Foundation. Royalty
payments must be paid within 60 days following each date on
which you prepare (or are legally required to prepare) your
periodic tax returns. Royalty payments should be clearly marked
as such and sent to the Project Gutenberg Literary Archive
Foundation at the address specified in Section 4, “Information
about donations to the Project Gutenberg Literary Archive
Foundation.”
• You provide a full refund of any money paid by a user who

notifies you in writing (or by e-mail) within 30 days of receipt that
s/he does not agree to the terms of the full Project Gutenberg™
License. You must require such a user to return or destroy all
copies of the works possessed in a physical medium and
discontinue all use of and all access to other copies of Project
Gutenberg™ works.
• You provide, in accordance with paragraph 1.F.3, a full refund of

any money paid for a work or a replacement copy, if a defect in
the electronic work is discovered and reported to you within 90
days of receipt of the work.
• You comply with all other terms of this agreement for free
distribution of Project Gutenberg™ works.
1.E.9. If you wish to charge a fee or distribute a Project

Gutenberg™ electronic work or group of works on different
terms than are set forth in this agreement, you must obtain
permission in writing from the Project Gutenberg Literary
Archive Foundation, the manager of the Project Gutenberg™
trademark. Contact the Foundation as set forth in Section 3
below.
1.F.
1.F.1. Project Gutenberg volunteers and employees expend

considerable effort to identify, do copyright research on,
transcribe and proofread works not protected by U.S. copyright
law in creating the Project Gutenberg™ collection. Despite
these efforts, Project Gutenberg™ electronic works, and the
medium on which they may be stored, may contain “Defects,”
such as, but not limited to, incomplete, inaccurate or corrupt
data, transcription errors, a copyright or other intellectual
property infringement, a defective or damaged disk or other
medium, a computer virus, or computer codes that damage or
cannot be read by your equipment.
1.F.2. LIMITED WARRANTY, DISCLAIMER OF DAMAGES -

Except for the “Right of Replacement or Refund” described in
paragraph 1.F.3, the Project Gutenberg Literary Archive
Foundation, the owner of the Project Gutenberg™ trademark,
and any other party distributing a Project Gutenberg™ electronic
work under this agreement, disclaim all liability to you for
damages, costs and expenses, including legal fees. YOU
AGREE THAT YOU HAVE NO REMEDIES FOR NEGLIGENCE,
STRICT LIABILITY, BREACH OF WARRANTY OR BREACH
OF CONTRACT EXCEPT THOSE PROVIDED IN PARAGRAPH
1.F.3. YOU AGREE THAT THE FOUNDATION, THE
TRADEMARK OWNER, AND ANY DISTRIBUTOR UNDER
THIS AGREEMENT WILL NOT BE LIABLE TO YOU FOR
ACTUAL, DIRECT, INDIRECT, CONSEQUENTIAL, PUNITIVE
OR INCIDENTAL DAMAGES EVEN IF YOU GIVE NOTICE OF
THE POSSIBILITY OF SUCH DAMAGE.
1.F.3. LIMITED RIGHT OF REPLACEMENT OR REFUND - If

you discover a defect in this electronic work within 90 days of
receiving it, you can receive a refund of the money (if any) you
paid for it by sending a written explanation to the person you
received the work from. If you received the work on a physical
medium, you must return the medium with your written
explanation. The person or entity that provided you with the
defective work may elect to provide a replacement copy in lieu
of a refund. If you received the work electronically, the person or
entity providing it to you may choose to give you a second
opportunity to receive the work electronically in lieu of a refund.
If the second copy is also defective, you may demand a refund
in writing without further opportunities to fix the problem.
1.F.4. Except for the limited right of replacement or refund set

forth in paragraph 1.F.3, this work is provided to you ‘AS-IS’,
WITH NO OTHER WARRANTIES OF ANY KIND, EXPRESS
OR IMPLIED, INCLUDING BUT NOT LIMITED TO
WARRANTIES OF MERCHANTABILITY OR FITNESS FOR
ANY PURPOSE.
1.F.5. Some states do not allow disclaimers of certain implied

warranties or the exclusion or limitation of certain types of
damages. If any disclaimer or limitation set forth in this
agreement violates the law of the state applicable to this
agreement, the agreement shall be interpreted to make the
maximum disclaimer or limitation permitted by the applicable
state law. The invalidity or unenforceability of any provision of
this agreement shall not void the remaining provisions.
1.F.6. INDEMNITY - You agree to indemnify and hold the

Foundation, the trademark owner, any agent or employee of the
Foundation, anyone providing copies of Project Gutenberg™
electronic works in accordance with this agreement, and any
volunteers associated with the production, promotion and
distribution of Project Gutenberg™ electronic works, harmless
from all liability, costs and expenses, including legal fees, that
arise directly or indirectly from any of the following which you do
or cause to occur: (a) distribution of this or any Project
Gutenberg™ work, (b) alteration, modification, or additions or
deletions to any Project Gutenberg™ work, and (c) any Defect
you cause.
Section 2. Information about the Mission of

Project Gutenberg™
Project Gutenberg™ is synonymous with the free distribution of
electronic works in formats readable by the widest variety of
computers including obsolete, old, middle-aged and new
computers. It exists because of the efforts of hundreds of
volunteers and donations from people in all walks of life.
Volunteers and financial support to provide volunteers with the

assistance they need are critical to reaching Project

Textbook Actions and Invariants of Algebraic Groups Second Edition Ferrer Santos Ebook All Chapter PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Textbook Actions and Invariants of Algebraic Groups Second Edition Ferrer Santos Ebook All Chapter PDF

Uploaded by

Copyright:

Available Formats

Actions and invariants of algebraic

groups Second Edition Ferrer Santos

Actions and Invariants of Algebraic Groups, Second

Modernization and Political Actions in the Brazilian

Algebraic Groups The Theory of Group Schemes of Finite

Algebraic Groups The Theory of Group Schemes of Finite

Theory of Groups and Symmetries Finite groups Lie

Longitudinal Multivariate Psychology Emilio Ferrer

Theory of Groups and Symmetries: Representations of

An Introduction to Quantum and Vassiliev Knot

variety is related to the local isotriviality of the canonical projection π : G → G/H — in

2 Actions by reductive groups: the semigeometric quotient

Proof. If U ⊂ X is an open saturated subset, then X \ U is closed and G-stable,

π π −1 (y) ∩ C ∩ π π −1 (y) ∩ D . Since (G\X, π) is a semigeometric quotient, we conclude

that π −1 (y) ∩ C ∩ π −1 (y) ∩ D 6= ∅, and hence

π −1 (y) ∩ C ∩ D = π −1 (y) ∩ C ∩ π −1 (y) ∩ D 6= ∅.

Since C, D are closed in π −1 (V ) and π −1 (y) ⊂ π −1 (V ), it follows that C ∩D = π −1 (V )∩

Proof. Let Z be an algebraic variety and f : X → Z a morphism constant along the

If x1 , x2 ∈ X are such that π(x1 ) = π(x2 ), then

Since f −1 (U ) is an open G-stable subset of X, we conclude that X \ f −1 (U ) is a closed

morphism of k-algebras (π ] )−1 ◦ f ] : OZ (U ) → OY fb−1 (U ) . Since U is affine, (π ] )−1 ◦ f ]

induces a morphism gU : fb−1 (U ) → U . As

we deduce that gU ◦ π =f . As π is surjective, fb = gU and hence fb is a

J1 ∩ G R + J2 ∩ G R ⊂ (J1 + J2 ) ∩ G R ⊂ rad (J1 ∩ G R) + (J2 ∩ G R) .

Moreover, if G is linearly reductive, then

r1 ∈ J1 and r2 ∈ J2 , consider r1 + J1 ∩ J2 ∈ J1 /(J1 ∩ J2 ) and r2 + J1 ∩ J2 ∈ J2 /(J1 ∩ J2 ). As

We prove now that if G is a reductive group acting on an affine algebraic variety X,

Proof. We need to prove condition (1) to (4) of Definition 2.1.

Hence, π(W ) is closed in Y .

Proof. This is a straightforward consequence of Lemma 2.4.

and ar,s = 0 unless r = s. Hence,

Observe that the principal open subset A2xy ⊂ A2 is G-stable and

π are the orbits (see Exercise 7).

In particular, this is an example of a semigeometric quotient that is not geometric.

3 Actions by reductive groups: the geometric quotient

Hence, π(x1 ) 6= π(x2 ). The converse is clear.

is a geometric quotient for the action of G on π −1 (V ) — therefore, if G acts on X with

Proof. In view of Lemma 2.4, if V ⊂ Y is an open subset, then (V, ππ−1 (V ) ) is a

If H ⊂ G is a reductive subgroup, then the orbits of the action by right translations

Ω(X) = x ∈ X max : O(x) is closed in X .

Next we establish some properties of these subsets.

Proof. If (Z, λ : U→ Z) is an orbit space, then Theorem2.6.4 guarantees that d : U →

is open in X. As X is irreducible, Vu ∩ X max 6= ∅, i.e., u ∈ Vu ⊂ X max and the result

Proof. If Y 0 = Y \ π X \ X max , since π X \ X max is closed — X max is open and

G-stable — then Y 0 is open.

Next we show that π−1 (Y 0 ) = Ω(X). First we prove that Ω(X) ⊂ π −1 (Y 0 ) — or

Definition 3.16. Let G be an affine algebraic group acting on an algebraic variety X.

Observation 3.17. Observe that (Xsoc )soc = Xsoc .

Ω(X) = π −1 (G\X) \ π(X \ X max ) .

In particular, Ω(X) is a (possibly empty) G-stable open subset.

Proof. It is clear that Ω(X) ⊂ π −1 (G\X) \ π(X \ X max ) . Since X \ X max is a

(2) π : X \ X max → G\X is surjective.

(3) X max does not contain any closed orbit.

is bijective and purely inseparable, i.e., ϕ is a bijective

Let Z ⊂ Xsoc be an irreducible component such that π is dominant. Then Xsoc ⊂ Z,

4 Canonical forms of matrices: a geometric perspective

the characteristicpolynomial and call π : Mn (k) → An , the morphism given as π(A) =

Proof. As an arbitrary matrix is similar to a matrix in Jordan form, it is enough to

If g = diag(1, a, . . . , an−1 ) ∈ GLn , then g −1 As g = As and g −1 An g = aAn . Thus, for all

GLnA = {g ∈ GLn : gA = Ag} ⊂ GLn ⊂ Mn (k),

where the dimension of the (i, j)-summand

Mn (k)max = A ∈ Mn (k) : A has a cyclic vector ,

In particular — recall the definition of D in Theorem 4.5 — Ω Mn (k) = π −1 (An \ D)