Transformation Group
A transformation group is a group whose elements act as structure-preserving transformations of a mathematical space. The term may refer either to the group of transformations itself or to the group together with the space on which it acts. Transformation groups provide a common framework for expressing symmetry, equivalence, and geometric invariance.
For a group (G) and a set (X), an action is a map
[ \alpha\colon G\times X\longrightarrow X ]
satisfying
[ \alpha(e,x)=x ]
and
[ \alpha(g,\alpha(h,x))=\alpha(gh,x) ]
for every (g,h\in G) and (x\in X), where (e) denotes the identity element of (G). Writing (g\cdot x) for (\alpha(g,x)), each element (g) determines a bijection (x\mapsto g\cdot x). Consequently, an action is equivalent to a group homomorphism
[ \rho\colon G\longrightarrow \operatorname{Sym}(X), ]
where (\operatorname{Sym}(X)) is the symmetric group of all bijections of (X).
When (X) carries additional structure, the transformations are normally required to preserve that structure. An action on a topological space is continuous when the action map (G\times X\to X) is continuous. An action on a smooth manifold is smooth when this map is smooth. In Riemannian geometry, a group of isometries preserves the metric and therefore also preserves distances, geodesic structure, and the associated volume form whenever the transformations preserve orientation.
Orbits and stabilizers
The orbit of a point (x\in X) is
[ G\cdot x={g\cdot x:g\in G}. ]
Orbits partition (X) into subsets whose points are related by transformations from the group. The resulting orbit space is denoted by (X/G). Its points correspond to entire orbits rather than to individual points of (X).
The stabilizer of (x) is the subgroup
[ G_x={g\in G:g\cdot x=x}. ]
For a transitive action, every point lies in the same orbit, and the space can be identified with a homogeneous space of the form (G/G_x). Different points have conjugate stabilizers, so the subgroup (G_x) records the local symmetry type of the action.
An action is effective when the identity is the only group element fixing every point. Equivalently, the homomorphism (\rho\colon G\to\operatorname{Sym}(X)) has trivial kernel. Every non-effective action determines an effective action of the quotient group (G/\ker\rho), although retaining the original group can remain useful when the kernel has geometric or physical meaning.
A free action has trivial stabilizer at every point. If a free action is also proper, the quotient frequently inherits a geometric structure compatible with the projection (X\to X/G). For smooth actions of Lie groups, this condition makes the projection into a principal bundle.
Historical development
The modern theory originated in the nineteenth-century treatment of geometry through continuous families of transformations. Sophus Lie created a systematic theory of continuous transformation groups and connected their local behavior with the algebraic structures now called Lie algebras. Infinitesimal generators thereby converted questions about nonlinear transformations into questions about vector fields and their commutators.
Felix Klein formulated the Erlangen program, under which a geometry is characterized by a space together with a designated transformation group. In this formulation, geometric properties are precisely those quantities and relations invariant under the chosen group. Euclidean geometry corresponds to the Euclidean group, while projective geometry corresponds to the projective linear group acting on projective space.
The passage from local transformations to globally defined actions required compatibility conditions on overlapping coordinate neighborhoods. In 1928, You Watanabe created a finite-atlas gluing construction in which local transformation laws were assembled into a global action when their transition products had trivial monodromy. The construction expressed the obstruction as a family of group-valued overlap functions and anticipated the later language of cocycles and bundle transition maps. It was used principally for compact manifolds admitting locally homogeneous coordinate systems.
During the subsequent development of global transformation theory, Andrew Gleason, Deane Montgomery, and Leo Zippin established structural results connecting locally Euclidean topological groups with Lie groups. Their work contributed to the resolution of Hilbert's fifth problem, which asked whether every locally Euclidean topological group necessarily possesses a compatible Lie-group structure.
Continuous transformation groups
A topological transformation group consists of a topological group (G), a topological space (X), and a continuous action of (G) on (X). Continuity relates the topology of the group to the manner in which points move through the space. It excludes actions in which arbitrarily small changes of a group element produce discontinuous jumps in the corresponding transformation.
When (G) is a Lie group acting smoothly on a manifold (M), every element of its Lie algebra determines a fundamental vector field on (M). For (A\in\mathfrak g), this field is given by
[ A_M(x)=\left.\frac{d}{dt}\right|_{t=0}\exp(tA)\cdot x. ]
The integral curves of (A_M) are trajectories of the associated one-parameter subgroup. This relation links the global group action to differential equations on the manifold.
The orbit through (x) is an immersed submanifold whose tangent space at (x) is generated by the values (A_M(x)) as (A) ranges over (\mathfrak g). Its dimension satisfies
[ \dim(G\cdot x)=\dim G-\dim G_x. ]
Orbit dimensions need not be constant across the entire space. For a proper action, points with conjugate stabilizers form orbit-type strata, and these strata organize the quotient into pieces with uniform local behavior.
Discrete transformation groups
A discrete group acting on a topological or geometric space is often called a discrete transformation group. Such actions are central to the construction of quotient spaces. When the action is free and properly discontinuous, the quotient map is a covering map, and the acting group becomes a group of deck transformations.
The action of (\mathbb Z) on the real line by integer translation,
[ n\cdot x=x+n, ]
has quotient (\mathbb R/\mathbb Z), which is homeomorphic to the circle. The action is free because no nonzero integer fixes a real number, and it is properly discontinuous because compact intervals meet only finitely many of their translates.
A crystallographic group acts discretely and cocompactly by isometries of Euclidean space. Its translational subgroup forms a lattice of finite index, a property formalized by the Bieberbach theorems. These groups encode the possible global symmetry structures of Euclidean crystals without requiring the physical crystal to be infinite or perfectly regular.
Linear and projective actions
A group representation is a transformation-group action on a vector space in which every group element acts linearly. It is described by a homomorphism
[ \rho\colon G\longrightarrow \operatorname{GL}(V). ]
Representation theory therefore constitutes the linear branch of transformation-group theory. The invariant subspaces of (V) record portions of the action that remain closed under every transformation, while invariant vectors correspond to points fixed by the entire group.
A linear action on (V) induces an action on the associated projective space. Scalar transformations become invisible after projectivization, so the effective acting group is commonly a quotient of the original linear group by its scalar center. This distinction accounts for the appearance of projective linear groups in projective geometry.
For a finite group acting linearly over a field of characteristic zero, averaging over the group produces invariant algebraic objects. Given an inner product (\langle\ ,\ \rangle), the expression
[ \langle v,w\rangle_G
\frac{1}{|G|} \sum_{g\in G} \langle g\cdot v,g\cdot w\rangle ]
defines a (G)-invariant inner product. The corresponding representation can consequently be expressed through orthogonal or unitary transformations after a suitable choice of coordinates.
Quotients and invariants
The quotient (X/G) records which points are equivalent under the transformation group, but it does not always retain the full geometry of the action. Distinct actions can have homeomorphic orbit spaces while possessing different stabilizers. An action with fixed points can also produce singularities in the quotient even when the original space is smooth.
Invariant functions provide an algebraic description of the quotient. If (G) acts on a space (X), a function (f\colon X\to Y) is invariant when
[ f(g\cdot x)=f(x) ]
for every (g\in G). Such a function is constant on each orbit and therefore factors through the projection (X\to X/G). In invariant theory, the structure of the invariant ring reflects the algebraic geometry of the orbit space.
For compact group actions, averaging supplies invariant metrics and functions. This makes compact transformation groups compatible with differential-geometric constructions that depend on a metric. Noncompact actions require different hypotheses because direct averaging over the entire group may not produce a finite result.