Motivation
Orbit–stabilizer tells us how many points an orbit has. There is a stronger statement behind that count: the stabilizer determines the entire action on the orbit, up to an appropriate notion of isomorphism. This makes a subgroup a way to describe a system of symmetries, even when the subgroup is not normal and its cosets do not form a quotient group.
We develop that statement and then ask a more precise question. If two sets carry actions of the same group, which maps between them respect every group operation? The answer turns a condition on all points into a single inclusion between subgroups. Its bijective case explains both the classification of transitive actions and a normalizer formula for their automorphisms.
Use the definitions of group actions, stabilizers and cosets from the preceding notes. Composition of functions is read from right to left. Groups and sets may be infinite unless a theorem explicitly assumes finiteness.
Definitions
Maps that preserve an action
Definition 2.3.1
(Equivariant maps and transitive actions)
Let carry left actions of the same group . A map is -equivariant if
A -set isomorphism is a bijective equivariant map. A nonempty -set is transitive if any point can be moved to any other point by some element of .
The inverse of a bijective equivariant map is equivariant. Indeed, if , then , and applying the inverse proves the required identity. Thus this definition of isomorphism really is symmetric.
Equivariance is more restrictive than being a bijection of the underlying sets. It requires the same named element to act compatibly on both sides; one cannot silently replace by an automorphism of . Also, a transitive action is required to be nonempty, so choosing a base point never invokes an empty-set exception.
For any subgroup , left multiplication defines an action on the set of left cosets:
This is well defined: replacing by for does not change . It is transitive, and the stabilizer of is exactly . No multiplication between cosets is being defined here, so normality is unnecessary.
Every orbit is a coset action
Theorem 2.3.2
(The stabilizer reconstructs a transitive action)
Let be a transitive -set and choose . With , the map
is a -set isomorphism. Consequently every transitive action is isomorphic to a coset action, including actions of infinite groups.
Proof
(Constructing the isomorphism from cosets)
If , then fixes , so . Conversely, equality of these points implies and therefore equality of the cosets. This proves well-definedness and injectivity at once. Transitivity supplies surjectivity. Finally,
which proves equivariance. This proof does more than compare cardinalities: it supplies the actual map that preserves the action.
Changing the chosen point changes the stabilizer by conjugation. If , then fixes exactly when fixes , so . This suggests that conjugate subgroups should describe the same transitive action. The next result proves the precise statement and also handles maps that are not isomorphisms.
All maps are determined at one coset
Theorem 2.3.3
(Equivariant maps between coset actions)
For subgroups , the equivariant maps correspond exactly to the cosets satisfying
The corresponding map is
Every such map is surjective. It is injective, and hence an isomorphism, exactly when .
Proof
(The image of the base coset determines the map)
Suppose first that is equivariant and write . Each fixes , so it must fix . Thus , equivalently . Equivariance then forces . In particular, there is no further freedom once is chosen.
Conversely, suppose the subgroup inclusion holds. If , write with . Then
The formula therefore defines a map on cosets, and proves equivariance. Every target coset has preimage , so the map is onto. Replacing by with changes neither the image coset nor the condition, since conjugation by preserves .
For injectivity, observe that
Equality of the original cosets is the stronger condition . We already know . If the two subgroups are equal, the conditions coincide and is injective. If the inclusion is strict, choose . Then and are different cosets with the same image. This proves the criterion without a finite-set argument.
Concept lensStructural
Equivariant maps as fixed points
The proof shows more than the theorem states. For any -set , evaluation at the base coset, , is a bijection from the equivariant maps onto the fixed-point set . Necessity is the first paragraph of the proof; conversely each gives the well-defined equivariant map . With , the coset is fixed by exactly when , which recovers the condition of the theorem.
Theorem 2.3.4
(Classification up to equivariant isomorphism)
The transitive -sets and are isomorphic exactly when and are conjugate in . If the isomorphism is additionally required to send the distinguished coset to , it exists exactly when .
Proof
(Equality case, with and without a base point)
The first assertion is precisely the equality case just proved. For the pointed assertion, the image coset is , so one may choose , and the equality criterion becomes . Thus remembering a base point retains an actual subgroup; forgetting it retains only a conjugacy class of subgroups.
Worked example 2.3.5
(Isomorphic stabilizers can describe different actions)
Let be the Klein four-group, with and . Put and . Both subgroups are abstractly cyclic of order two. Both coset sets have two points.
Nevertheless their actions are not isomorphic. In , the element fixes every coset, since is abelian and . In , it interchanges the two cosets, since . An equivariant bijection would have to preserve this behavior of the same element , which is impossible. Equivalently, conjugation in an abelian group fixes each subgroup, so the distinct subgroups are not conjugate.
Automorphisms and the normalizer
Write for the group of equivariant bijections , using function composition. It is usually much smaller than the symmetric group of the underlying set.
Theorem 2.3.6
(Automorphisms of a transitive action)
For every subgroup ,
An explicit isomorphism sends to the map
Proof
(Automorphisms as normalizer cosets)
First, is normal in by the definition of the normalizer, so the displayed quotient is a group. An equivariant self-map is bijective precisely when its image of is a coset with . These are exactly the cosets represented by normalizer elements. Taking gives all automorphisms in the stated form.
The inverse in the formula matters. Direct calculation gives
Thus is a homomorphism, not an order-reversing map. Its kernel consists exactly of the elements , since is equivalent to . The first isomorphism theorem now proves the claim. This also verifies that the choice of representative for causes no ambiguity.
Proof X-ray
Why a right translation needs an inverse
The action of is by multiplication on the left, while an equivariant automorphism moves a representative by multiplication on the right. If one writes , then : the multipliers appear in reverse order. Writing corrects that reversal and gives an ordinary homomorphism from . Checking the value at an arbitrary coset makes the direction visible.
When is finite, the theorem gives exactly equivariant automorphisms. It is useful to distinguish three subgroups: fixes the chosen point; controls action automorphisms; the core is the kernel of the action of itself. These answer different questions.
Worked example 2.3.7
(The natural three-point action has no nontrivial equivariant automorphism)
Let act on , and choose the base point . Its stabilizer is . A permutation normalizing must conjugate its unique nonidentity element to itself. Hence it preserves the set and fixes , so . The equivariant automorphism group is therefore trivial.
One can see the result directly: an equivariant map must send to a point fixed by , and only is fixed. Transitivity then determines the whole map as the identity. Although the underlying set admits six bijections, only one commutes with every permutation in this action.
Worked example 2.3.8
(A normal stabilizer gives translations of a quotient)
Take cyclic of order six and . There are three cosets. Since is abelian, , and the equivariant automorphism group is cyclic of order three. For instance sends respectively to .
The action is transitive but not faithful: its kernel is . It is not free either, since the nonidentity element fixes every point. After passing to the quotient , however, it becomes the regular left action of a three-element group on itself. Normality makes this quotient interpretation possible; transitivity alone would not.
A small index can force normality
Theorem 2.3.9
(Normality at the least prime divisor)
Let be a finite group, and let be the smallest prime dividing . If has index , then is normal and is cyclic of order .
Proof
(Counting the image of the coset action)
Let act on its cosets of , giving a homomorphism . Its image is transitive, so orbit–stabilizer shows that divides . Also divides both and : use the first isomorphism theorem for the first divisibility and Lagrange in for the second.
Every prime divisor of is therefore at least , because it divides , and at most , because it divides . Only can occur. There is exactly one factor of in , so . The stabilizer in of any one of the cosets consequently has order one. An element of fixes the coset exactly when it belongs to ; because the corresponding stabilizer in is trivial, this is also exactly when it belongs to . Thus is normal. Its quotient has prime order and is cyclic.
Proof X-ray
Squeezing the image between two divisibility bounds
The argument studies the image rather than , because is constrained from two sides. Being isomorphic to , it has order dividing , so its prime divisors are at least ; this is where minimality of is used. As a subgroup of it has order dividing , so its prime divisors are at most and ; this is where the index is used. Transitivity supplies . Hence , and a transitive group of order on points has trivial stabilizers. The close converts “fixes the coset ” into “acts trivially”, so .
For example, any subgroup of index three in a group of order is normal, because three is the smallest prime dividing . This statement is conditional on the subgroup existing; the proof does not construct one. The same argument includes the familiar index-two theorem.
Counterexample mode
Prime index alone does not force normality
One might expect every subgroup of prime index to be normal, since index two always works. Take , of prime index three. Conjugation by sends to , so is not normal. In the proof above, the coset action has kernel , so its image has order six: the prime two survives in the image, and the order cannot be forced down to three. The repaired statement is the theorem itself: the index must be the smallest prime dividing , which for is two, not three.
The finite-set hypothesis behind bijective self-maps
Every equivariant map between transitive actions is surjective, as proved above. A self-map of a finite transitive action is therefore automatically bijective. For infinite actions, that last conclusion fails; the subgroup criterion explains precisely how.
Worked example 2.3.10
(An infinite transitive action with a noninjective equivariant self-map)
Let be the group of affine bijections of of the form
where . Composition preserves this form, and the inverse is , again of this form. Let be the subgroup of integer translations and let . Then
The map is consequently well defined and surjective on the transitive coset set . It is not injective: , but , so
This is a concrete strict-inclusion case of the map theorem. It also explains why the automorphism theorem uses equality , rather than merely requiring a self-map to be well defined.
Common mistakes
Common mistake
(Abstractly isomorphic stabilizers are not enough)
An abstract isomorphism does not classify transitive actions of a fixed . Their positions inside matter, and the correct condition is conjugacy. A bijection of the coset sets also need not be equivariant.
Common mistake
(The inclusion runs )
A map exists exactly when there is some with , with this direction of inclusion. The condition follows by asking which elements of must fix the proposed image . Checking a base point is a reliable way to recover the formula without memorizing it.
Common mistake
(Transitive, faithful and free are different conditions)
Finally, transitive, faithful and free are distinct properties. Transitive means one orbit; faithful means a trivial action kernel; free means every point stabilizer is trivial. The three-point action is faithful and transitive but not free. The cyclic-six example is transitive with a nontrivial kernel. The regular action of any group on itself is both faithful and free.
Exercises
Quick check
(Let . Prove that is equivariant and describe its fiber over . When is finite, compute the size of every fiber and determine exactly when this map is injective.)
Solution · Answer
This is in the map theorem, since . The fiber over is the set , naturally the coset set even when is not normal in . Every other fiber is its translate: over it is . For finite each has size , and the map is injective exactly when . Calling the fiber a coset set does not introduce a group operation on it.
Quick check
(Let act on itself by left multiplication. Prove directly that every equivariant self-map has the form for a unique . Explain how the inverse convention gives even when is nonabelian.)
Solution · Answer
Put . Equivariance forces , and any such right translation is equivariant and has inverse right translation by . Thus every self-map here is an automorphism. However . The homomorphism giving the stated isomorphism is , as in the theorem; omitting the inverse would reverse multiplication.
Quick check
(Suppose and are finite and have the same size. Show that any equivariant map between them is an isomorphism. Does equal size alone guarantee the existence of such a map?)
Solution · Answer
Every equivariant map is onto, and a surjection between finite sets of equal size is injective. Equal size does not ensure existence: the distinct order-two subgroups of the Klein four-group in the worked counterexample give two coset sets of size two with no equivariant isomorphism, and therefore no equivariant map between them. The failure is a missing subgroup inclusion, not a cardinality obstruction.
Quick check
(Let have order and let be a subgroup of order . Prove that is normal and determine . State which part of the argument would fail for a subgroup of order two in .)
Solution · Answer
The index is , and three is the least prime divisor of . The normality theorem gives and . For the subgroup of order two in , the index is also three but the group order has the smaller prime divisor two. An image in can then retain that factor and have order six, so its point stabilizers need not be trivial.
Quick check
(If is abelian and , determine when an equivariant map exists and count all such maps when the target is finite. Compute .)
Solution · Answer
Conjugation does nothing, so the map condition is simply , independently of the chosen coset . If the inclusion fails there are no maps. If it holds, each coset of supplies exactly one map, giving maps for finite target. All subgroups of an abelian group are normal, so the automorphism theorem gives . This includes , whose one-point action has only its identity automorphism.