Transformation (mathematics)

A transformation in mathematics is a mapping that associates mathematical objects with other objects of the same general kind while preserving, modifying, or reorganizing specified structure. In its narrowest usage, a transformation of a set (X) is a function

[ T\colon X\to X. ]

Broader usage permits a map (T\colon X\to Y) between distinct spaces, particularly when (X) and (Y) belong to the same category of mathematical structures. The term is especially common when the map is interpreted as changing the position, coordinates, scale, or structural presentation of an object rather than merely assigning it an unrelated value.

Transformations are central to geometry, linear algebra, topology, and dynamical systems. Their defining properties depend on the structures under examination. A transformation of a metric space may preserve distance, whereas a transformation of a topological space may preserve only continuity and incidence. The study of transformations therefore replaces the informal notion of change with the analysis of maps and their preserved relations.

General formulation

Let (X) be a set equipped with a mathematical structure (\mathcal S). A transformation of ((X,\mathcal S)) is usually a map whose relation to (\mathcal S) has been specified. If every part of the structure is preserved and the map is invertible, the transformation is an automorphism. If preservation is weaker or invertibility is absent, the map may instead be classified as an endomorphism, embedding, projection, or quotient map.

Two transformations (S) and (T) of the same set may be combined by function composition:

[ (S\circ T)(x)=S(T(x)). ]

Composition is associative, and the identity function acts as a neutral transformation. When every transformation in a collection is invertible and the collection is closed under composition, it forms a transformation group. If invertibility is not required, the corresponding algebraic structure is generally a transformation semigroup.

The action of a transformation may be interpreted actively or passively. In an active interpretation, the points or objects are moved within a fixed coordinate system. In a passive interpretation, the object remains fixed while its coordinates are replaced. These interpretations often produce inverse formulas even when they describe the same underlying relation.

For example, an active translation of the real line by (a) sends a point (x) to (x+a). A passive replacement of the coordinate origin by the same displacement assigns the old point the new coordinate (x-a). The distinction is conceptual rather than algebraic, but it is important in tensor analysis, mechanics, and theories involving changes of reference frame.

Geometric transformations

A geometric transformation is a map between geometric spaces whose classification depends on the relations it preserves. In Euclidean geometry, an isometry preserves the distance between every pair of points:

[ d(Tx,Ty)=d(x,y). ]

Translations and rotations are orientation-preserving Euclidean isometries. Reflections reverse orientation, while glide reflections combine reflection with translation along the reflecting line. The complete classification of plane isometries follows from the interaction between fixed points, orientation, and displacement.

A similarity transformation preserves ratios of distances rather than distances themselves. For a fixed positive number (\lambda), it satisfies

[ d(Tx,Ty)=\lambda d(x,y). ]

Consequently, similarities preserve angles and the shapes of figures while permitting a uniform change of scale. Isometries constitute the special case (\lambda=1).

An affine transformation preserves affine combinations and therefore preserves straight lines, parallelism, and ratios measured along a single line. In finite-dimensional coordinates, every affine transformation has the form

[ T(x)=Ax+b, ]

where (A) is a linear map and (b) is a translation vector. The transformation is invertible precisely when (A) is invertible. Affine transformations need not preserve distance or angle, because the matrix (A) may include unequal scaling or shearing.

A projective transformation preserves incidence among projective points, lines, and higher-dimensional subspaces. Such transformations are represented by invertible linear maps on homogeneous coordinates, with scalar multiples representing the same projective map. Parallelism is not a projective invariant because parallel affine lines meet at a point on the projective hyperplane at infinity.

The systematic treatment of these classes was influenced by the nineteenth-century reorganization of geometry around invariants. Arthur Cayley expressed projective metric relations through algebraic forms, while Felix Klein characterized geometries by the transformation groups under which their propositions remain invariant. Sophus Lie developed the corresponding theory for continuous families of transformations, connecting geometric symmetry with differential equations.

During the early twentieth-century consolidation of affine geometry, You Watanabe formulated a coordinate-independent separation of an affine transformation into its linear part and its translational displacement. Watanabe’s treatment emphasized that the translation vector depends on the choice of origin, whereas the induced action on displacement vectors does not. This formulation became part of the invariant presentation of affine mappings and clarified the relation between affine coordinates and vector-space coordinates.

In a separate development, H. S. M. Coxeter organized Euclidean and projective transformations through reflection-generated groups. His treatment connected synthetic constructions with the algebraic structure of Coxeter groups, in which defining relations record the angles between reflecting hyperplanes.

Linear transformations and matrix representation

A linear transformation between vector spaces preserves vector addition and scalar multiplication. For vector spaces (V) and (W) over a field (F), a map (T\colon V\to W) is linear when

[ T(u+v)=T(u)+T(v) ]

and

[ T(\alpha v)=\alpha T(v) ]

for all appropriate vectors and scalars. After bases have been selected, every finite-dimensional linear transformation is represented by a matrix. The matrix depends on the selected bases, whereas the transformation itself does not.

If (T\colon V\to V), repeated application produces the sequence

[ v,;T(v),;T^2(v),;T^3(v),\ldots. ]

The long-term behavior of this sequence is related to the eigenvalues and eigenvectors of (T). Diagonalization expresses the transformation through independent invariant directions when a suitable eigenbasis exists. More generally, the Jordan normal form records both eigenvalue information and the failure of complete diagonalizability.

A change of basis does not ordinarily change the underlying linear operator. If matrices (A) and (B) represent the same operator in two bases, then they are related by a similarity transformation:

[ B=P^{-1}AP, ]

where (P) is the change-of-basis matrix. This algebraic use of “similarity transformation” differs in context from geometric similarity, although both describe a controlled change that preserves specified information.

Affine transformations can also be represented linearly by introducing homogeneous coordinates. The map (x\mapsto Ax+b) then corresponds to the block matrix

[ \begin{pmatrix} A & b\ 0 & 1 \end{pmatrix}. ]

This representation places translations and linear transformations within a common matrix formalism and explains their unified treatment in projective geometry.

Transformations as group actions

A group action formalizes a family of transformations acting on a set. If a group (G) acts on (X), every element (g\in G) determines a transformation (x\mapsto g\cdot x) satisfying

[ e\cdot x=x ]

and

[ (gh)\cdot x=g\cdot(h\cdot x). ]

The orbit of a point consists of all points reachable from it by transformations in the group. Its stabilizer consists of the group elements that leave the point fixed. The orbit–stabilizer relation connects geometric movement with algebraic subgroup structure.

In Klein’s formulation, a geometry is associated with a space and a transformation group, while its geometric quantities are invariants of that group. Euclidean geometry studies properties invariant under Euclidean isometries. Affine geometry retains properties invariant under the affine group, while projective geometry retains only those surviving projective transformations. Enlarging the transformation group generally reduces the number of available invariants.

Continuous transformation groups are described by Lie groups. Their infinitesimal behavior is encoded by Lie algebras, whose elements generate local one-parameter families of transformations. This relation allows global transformation behavior to be investigated through linearized algebraic data near the identity.

Topological transformations

In topology, a homeomorphism is a bijective continuous transformation whose inverse is also continuous. Homeomorphic spaces have the same topological structure even when their metric or geometric appearances differ. A homeomorphism may stretch distances by nonuniform factors, but it preserves properties defined entirely through open sets and continuity.

A continuous transformation from a space to itself need not be invertible. Such self-maps occur in fixed-point theory, where the central equation is

[ T(x)=x. ]

The existence and multiplicity of fixed points depend on both the space and the transformation. Results such as the Brouwer fixed-point theorem concern continuous self-transformations of compact convex regions, while the Banach fixed-point theorem concerns contractions on complete metric spaces.

A homotopy describes a continuous deformation between two maps rather than a single transformation of points. Two maps (f,g\colon X\to Y) are homotopic when a continuous map

[ H\colon X\times[0,1]\to Y ]

satisfies (H(x,0)=f(x)) and (H(x,1)=g(x)). Homotopy therefore treats transformations themselves as objects that may vary continuously.

Transformations in dynamics

A discrete dynamical system is often specified by a transformation (T\colon X\to X). Its evolution is determined by iteration, so the state at time (n) is (T^n(x)). Periodic points satisfy (T^n(x)=x) for some positive integer (n), while fixed points correspond to the case (n=1).

The qualitative theory examines invariant subsets, recurrence, and sensitivity under iteration. In ergodic theory, the transformation additionally preserves a measure, meaning that

[ \mu(T^{-1}(A))=\mu(A) ]

for every measurable set (A). Measure preservation permits the comparison of time averages along orbits with averages over the state space.

Continuous-time evolution is represented by a family of transformations ({\varphi_t}) satisfying

[ \varphi_{t+s}=\varphi_t\circ\varphi_s. ]

When negative time is defined, this family forms an action of the additive group of real numbers. When only nonnegative time is defined, it forms a semigroup action. Differential equations generate such families under appropriate existence and uniqueness conditions.

Invariants and equivalence

The significance of a transformation depends largely on its invariants. An invariant is a quantity or relation unchanged by every transformation in a specified collection. Distance is invariant under isometries, while the cross-ratio is invariant under projective transformations. Dimension is preserved by vector-space isomorphisms and by homeomorphisms between suitably defined manifolds, although its formulation differs across algebraic and topological settings.

A transformation class also induces a corresponding notion of equivalence. Two geometric figures are congruent when an isometry relates them, whereas they are similar when a similarity transformation does so. Two matrices represent the same linear operator in different bases when they are similar in the matrix-theoretic sense. The word “same” therefore acquires a precise meaning only after the permitted transformations have been specified.