1 Concept and Definition

An affine transformation is a function between geometric spaces that preserves straightness and parallelism. It may shift, rotate, scale, or shear a figure, but it does not usually preserve distances or angles. In Euclidean settings, such maps provide a general framework for describing motions and deformations that keep lines mapped to lines.

A key feature of affine transformations is that they act consistently on entire geometric configurations. A line segment remains a line segment, a pair of parallel lines remains parallel, and the relative arrangement of points on a line is preserved in a broad sense. This makes affine maps central in geometry and in applications where shape structure matters more than exact metric measurements.

1.1 Affine maps vs. linear maps

A linear map sends the origin to the origin and preserves addition and scalar multiplication. An affine map extends this idea by allowing a translation after the linear part. Thus, a linear transformation is a special case of an affine transformation, namely one with no translation term.

Because affine maps need not fix the origin, they are often better suited to describing real geometric changes. A pure translation, for example, is affine but not linear. This distinction is important when a coordinate system is chosen arbitrarily rather than centered at a natural geometric origin.

1.2 Coordinate form and geometric meaning

In coordinates, an affine transformation typically has the form \[ x \mapsto Ax + b, \] where \(A\) is a matrix representing the linear part and \(b\) is a translation vector. The matrix determines how directions are stretched, rotated, or sheared, while the vector shifts the result in space.

Geometrically, this form separates two kinds of effects. The matrix changes the shape of infinitesimal patterns and relative directions, whereas the translation repositions the entire image. This decomposition is one reason affine transformations are easy to analyze and compose.

1.3 Properties preserved by affine transformations

Affine transformations preserve collinearity, parallelism, and the ratio of lengths along a single line. They also preserve intersections of lines and the incidence relations among points, lines, and planes. More generally, they map affine subspaces to affine subspaces of the same or lower dimension.

What they do not preserve are metric properties such as exact length, angle, and circle shape. A circle, for instance, may become an ellipse under a non-uniform scaling or shear. Even so, the transformed figure retains a recognizable structural relationship to the original.

1.4 Relationship to convexity and collinearity

Affine maps preserve convexity: the image of a convex set is convex. This follows from the fact that affine combinations of points are carried to affine combinations of their images. As a result, polygons, polytopes, and other convex bodies keep their overall convex character.

Collinearity is also preserved. If several points lie on the same line, their images will lie on a line as well. This makes affine transformations especially useful in computational geometry, where preserving the combinatorial structure of point sets is often essential.

2 Algebraic Formulations

Affine transformations admit compact algebraic descriptions. These descriptions are useful for proving general properties, computing images of points, and combining multiple transformations efficiently. The most common formulations use matrices, vectors, and homogeneous coordinates.

2.1 Matrix representation in standard coordinates

In an \(n\)-dimensional coordinate system, an affine map can be written as \[ x' = Ax + b, \] with \(A\) an \(n \times n\) matrix and \(b\) an \(n\)-component vector. This representation applies directly to points expressed as coordinate columns.

The matrix \(A\) governs the internal geometry of the map. If \(A\) is invertible, the affine transformation is one-to-one and onto its image; if not, it may collapse dimensions and lose information. The vector \(b\) then adjusts the position of the resulting figure.

2.2 Linear part and translation vector

The linear part and the translation vector play distinct roles. The linear part determines how basis directions are transformed, while the translation moves the entire configuration without changing its shape by itself. Together they define the full action of the affine map.

This split is not unique in an absolute sense, because it depends on the chosen origin. If the origin changes, the translation term changes as well, even when the geometric map remains the same. The linear component, however, is tied to the directional behavior of the transformation.

2.3 Homogeneous coordinates

Homogeneous coordinates provide a convenient way to represent affine transformations as matrix multiplications. In this setting, a point \(x\) is embedded into a higher-dimensional vector by appending an extra coordinate, and the affine map becomes a single matrix acting on that augmented vector.

This approach simplifies composition and implementation. Translation, which is awkward to encode in ordinary linear algebra, becomes part of a unified matrix framework. It is widely used in computer graphics and robotics because multiple transformations can be combined through ordinary matrix multiplication.

2.4 Composition of affine transformations

The composition of two affine transformations is again affine. If one map is \(x \mapsto A_1x + b_1\) and another is \(x \mapsto A_2x + b_2\), their composition has the form \[ x \mapsto A_2(A_1x+b_1)+b_2 = (A_2A_1)x + (A_2b_1+b_2). \] This closure property makes affine transformations form a natural algebraic group when the linear parts are invertible.

Composition is generally order-sensitive. Applying a rotation before a translation can yield a different result from applying the translation first. This noncommutativity is one of the main reasons affine transformations are best handled with explicit formulas or matrix products.

2.5 Inverses and conditions for invertibility

An affine transformation is invertible exactly when its linear part is invertible. In that case, the inverse map is also affine and can be written explicitly. If \(x' = Ax + b\), then \[ x = A^{-1}(x' - b). \]

When \(A\) is singular, the transformation collapses some directions and cannot be reversed uniquely. Such a map may flatten a plane into a line or send a three-dimensional region into a lower-dimensional set. Invertibility therefore depends entirely on the matrix component, not on the translation alone.

3 Geometric Effects

Affine transformations can be understood by their visible effects on shapes. Depending on the linear part, they may move objects rigidly, stretch them, tilt them, or combine several such changes at once. The translation component then relocates the entire result.

3.1 Translation

Translation shifts every point by the same vector. It changes position but leaves size, orientation, and shape unchanged. Since all points move equally, all distances and angles are preserved under translation alone.

Translations are among the simplest affine transformations and serve as the basic motion model in many coordinate systems. They are also the most direct example of an affine map that is not linear, because they do not send the origin to itself.

3.2 Rotation and its role within affine maps

Rotation is a linear transformation that preserves distances and angles. Within the broader affine family, it often appears as part of the linear component, sometimes combined with scaling or shear. A rotation followed by a translation yields a rigid motion expressed in affine form.

Rotations are important because they preserve orientation and metric structure, unlike many other affine effects. When used together with translation, they model ordinary turning and repositioning of objects in the plane or in space.

3.3 Scaling (uniform and non-uniform)

Scaling changes the size of a figure by multiplying coordinates along one or more axes. Uniform scaling uses the same factor in every direction and preserves angles, whereas non-uniform scaling uses different factors and can distort shapes into stretched forms.

In affine terms, scaling is encoded in the linear matrix. A positive factor preserves orientation, while a negative factor reflects the figure across an axis or plane. Non-uniform scaling is especially common in modeling, where objects may need to be elongated or compressed selectively.

3.4 Shear transformations

A shear shifts points in one direction by an amount proportional to another coordinate. This tilts shapes without necessarily changing area in all cases, and it can convert rectangles into parallelograms. Shears are typical examples of transformations that preserve parallel lines while altering angles.

Shear transformations are useful in design, typography, and visualization because they provide controlled distortion. They are also valuable in linear algebra as examples of matrices that are simple to write but geometrically significant.

3.5 General deformation as a combination of linear effects and translation

Most affine transformations can be viewed as combinations of rotation, scaling, shear, and translation. The exact decomposition may not be unique, but the overall effect is always the same: a structured deformation of space that preserves affine relationships.

This combination viewpoint helps explain why affine maps are so flexible. A single formula can describe many different geometric actions, from rigid motion to strong distortion, while still maintaining line structure and parallelism.

4 Classification by Special Cases

Affine transformations form a broad family that contains several important subclasses. These special cases are distinguished by additional preserved quantities, such as distance, angle, area, or orientation. Their classification helps connect affine geometry with more familiar geometric motions.

4.1 Isometries within the affine group

Isometries preserve distances exactly. In Euclidean space, they include translations, rotations, reflections, and their combinations. Every isometry is affine, but not every affine transformation is an isometry.

Within the affine group, isometries are the most rigid examples. They leave the shape and size of objects unchanged, making them fundamental in geometry, physical modeling, and rigid-body motion.

4.2 Similarity transformations

Similarity transformations preserve angles and the shape of figures up to uniform scaling. They may include a common scale factor, a rotation or reflection, and a translation. Circles remain circles, and ratios of corresponding lengths are preserved.

Similarities are intermediate between isometries and general affine transformations. They allow size changes but still maintain the essential form of geometric objects. This makes them useful in mapmaking, pattern recognition, and technical drawing.

4.3 Volume scaling and determinant interpretation

The determinant of the linear part measures signed volume scaling. In two dimensions, it corresponds to signed area scaling; in three dimensions, to signed volume scaling. A determinant of zero indicates collapse to a lower-dimensional image.

The sign of the determinant also reflects orientation. A positive value preserves orientation, while a negative value reverses it. This interpretation provides a compact way to classify how an affine transformation changes space.

4.4 Rank-deficient non-invertible affine maps

If the linear part has rank smaller than the ambient dimension, the affine transformation is not invertible. Such maps can project space onto a line, a plane, or another lower-dimensional set. They often arise when one wants to discard certain coordinates or compress data.

Rank deficiency means that some distinct points share the same image. Although this prevents exact reversal, the map can still be useful in visualization and dimensional reduction. Its geometric effect is a controlled loss of dimension.

4.5 Projections as special affine-like mappings

Projections send points to a subset by dropping or collapsing one or more directions. Orthogonal and oblique projections are closely related to affine maps, especially when expressed in coordinates. They are often represented by affine or linear formulas depending on the setup.

Projections are important because they provide a simple model for viewing three-dimensional objects in two dimensions. Even when not invertible, they preserve enough structure to show relative layout and alignment.

5 Affine Subspaces and Invariant Structures

Affine transformations interact naturally with lines, planes, and higher-dimensional affine subspaces. These objects form the basic “flat” structures of affine geometry, and they behave predictably under affine maps. This makes them central to both theory and applications.

5.1 Images of lines, planes, and parallel sets

A line is mapped to a line, unless it is collapsed to a point by a non-invertible map. Likewise, a plane is sent to a plane or to a lower-dimensional affine set. Parallel families of lines or planes remain parallel after an affine transformation.

This preservation of flat structure is one of the defining features of affine geometry. It allows complex figures to be analyzed by tracking their supporting lines and planes rather than their exact metric details.

5.2 Affine combinations and barycentric coordinates

An affine combination is a weighted sum of points whose coefficients add to one. Such combinations are preserved by affine transformations. This property explains why affine maps preserve centroids and other weighted geometric constructions.

Barycentric coordinates express a point relative to the vertices of a simplex using affine combinations. They are especially useful in triangles and tetrahedra, where they provide a coordinate system adapted to shape interpolation and geometric computation.

5.3 Fixed points and affine subspaces

A fixed point is a point that remains unchanged under the transformation. Some affine maps have none, some have one, and some have infinitely many. The set of fixed points, when nonempty, can itself form an affine subspace.

The existence and structure of fixed points depend on the equation \(Ax+b=x\). This can be rewritten as \((A-I)x=-b\), showing that the linear part determines whether a solution exists and how many there are.

5.4 Invariants under affine transformations

Affine transformations preserve a range of structural features, including collinearity, parallelism, affine ratios on a line, and convexity. They also preserve the property of being an affine subspace. These invariants are often more important than metric quantities in purely geometric arguments.

Because of these preserved features, affine maps are well suited for studying configurations up to shape-preserving deformation. They reveal what is intrinsic to a layout and what depends on the choice of coordinates or scale.

6 Affine Transformations in Computation

Affine transformations are widely used in computational settings because they are efficient, composable, and easy to represent numerically. They appear in rendering pipelines, image processing, motion tracking, and geometric modeling. Their matrix form makes them especially suitable for implementation.

6.1 Applications in computer graphics

In computer graphics, affine transformations position, rotate, scale, and skew objects in a scene. They are used to move models from local coordinates into world coordinates and then into camera or screen coordinates. This chain of transformations is a standard part of rendering.

Affine maps are also used for animation and articulated motion. By combining several transformations, programmers can describe the placement of parts of a figure relative to each other in a compact way.

6.2 Image warping and resampling

Image warping applies an affine transformation to the coordinates of pixels or samples. This can correct orientation, align images, or create visual effects such as tilts and stretches. Because pixel locations are discrete, the transformed image usually requires resampling.

Resampling introduces interpolation choices and possible artifacts. Even when the underlying transformation is simple, the rasterized result must approximate continuous geometry on a grid. This is why affine warping combines geometry with numerical approximation.

6.3 Transform pipelines and transform order

Modern systems often apply several transformations in sequence. The order matters because matrix multiplication is not commutative. A typical pipeline may first scale an object, then rotate it, and finally translate it to its target position.

Transform pipelines are used to separate modeling, viewing, and projection steps. Each step has a clear geometric role, and the combined effect is represented by a single affine matrix when possible. This organization reduces complexity in both software and analysis.

6.4 Numerical stability and implementation considerations

Computing with affine transformations requires attention to floating-point precision. Repeated composition can accumulate roundoff error, and near-singular matrices may be difficult to invert reliably. Stable implementations often prefer direct construction or careful factorization.

Practical systems also distinguish between points and vectors in code, since translations should affect points but not direction vectors. This distinction prevents logical errors and helps preserve the intended geometric meaning of each operation.

7 Connections to Other Mathematical Concepts

Affine transformations sit at the intersection of geometry, linear algebra, and coordinate methods. They are closely related to several other concepts that organize geometric reasoning. These connections help explain why affine ideas appear across so many areas of mathematics.

7.1 Affine geometry basics

Affine geometry studies properties that remain unchanged under affine transformations. It focuses on incidence, parallelism, and affine combinations rather than distances or angles. In this setting, the notion of “straightness” is primary.

The subject provides a coordinate-free perspective on many familiar constructions. By removing metric assumptions, affine geometry isolates the structural features shared by a wide class of shapes and mappings.

7.2 Relation to vector spaces and coordinate changes

Vector spaces provide the algebraic language underlying affine transformations, but affine spaces do not require a distinguished origin. This makes affine geometry a natural extension of linear algebra when position matters separately from direction.

Changes of coordinates often involve affine transformations. A new coordinate frame may shift and reorient the description of a figure without changing the figure itself. This distinction between the object and its coordinates is central to geometry and mechanics.

7.3 Barycentric coordinates and triangulations

Barycentric coordinates are especially useful in triangulated domains. They describe points relative to the vertices of a simplex and behave naturally under affine maps. This makes them valuable for interpolation, mesh processing, and geometric algorithms.

Triangulations break complex regions into simple pieces. Since affine maps preserve the combinatorial layout of each simplex, they support many local computations that assemble into a global result.

Projective transformations generalize affine transformations by allowing points at infinity and more flexible line behavior. Every affine transformation can be viewed as a special case of a projective transformation that preserves the hyperplane at infinity.

The contrast is useful because affine maps preserve parallelism, while projective maps generally do not. This difference marks a shift from affine geometry to a broader framework used in perspective and vanishing-point geometry.

8 Worked Examples and Intuition

Concrete examples help clarify how affine transformations behave. By examining simple cases in two and three dimensions, one can see how the algebraic formulas correspond to visible changes in shape and position. These examples also show how affine maps are built from basic components.

8.1 2D examples: shear and non-uniform scaling

A two-dimensional shear might send \((x,y)\) to \((x+y,y)\). Horizontal lines remain horizontal, while vertical lines slant. Squares become parallelograms, but the image still has straight sides and parallel opposite edges.

Non-uniform scaling might send \((x,y)\) to \((2x,\tfrac12 y)\). Here the figure stretches in one direction and shrinks in the other. A circle under this map generally becomes an ellipse, illustrating the loss of metric symmetry while preserving affine structure.

8.2 3D examples: composing rotations with translations

In three dimensions, a rotation about an axis can be combined with a translation to move an object rigidly. For example, one may rotate a model part and then place it at a specific location in space. The resulting transformation is affine and invertible.

This type of composition is common in robotics and animation. It allows an object to be oriented and positioned independently, which is essential for describing articulated motion and scene assembly.

8.3 Constructing an affine map from point correspondences

An affine transformation in \(n\) dimensions is determined by its action on \(n+1\) points, provided those points are affinely independent. Given corresponding source and target points, one can solve for the matrix and translation that map one set to the other.

This construction is often used in registration and interpolation problems. Once enough point pairs are known, the affine map can be recovered algebraically and then applied to other points in the same coordinate system.

8.4 Interpreting transformed coordinate systems

An affine transformation can be viewed either as moving a shape or as changing the coordinate system used to describe it. These two perspectives are mathematically equivalent but intuitively different. In one view, the object changes place; in the other, the observer’s frame changes.

This dual interpretation helps explain why affine transformations are so versatile. They describe both actual geometric deformation and re-expression of geometry in new coordinates, making them a bridge between shape and representation.