Affine space

Lore & Background

In an affine space, there is no distinguished origin, unlike in a vector space. Points cannot be added together arbitrarily, but affine combinations—weighted sums with coefficients summing to 1—are meaningful and define barycentric coordinates. Any vector space can be viewed as an affine space by forgetting the special role of the zero vector; when considered as a point, the zero vector is called the origin. Adding a fixed vector to a linear subspace produces an affine subspace, often described as translating the linear subspace away from the origin. In finite dimensions, such an affine subspace is the solution set of an inhomogeneous linear system, while the displacement vectors are solutions of the corresponding homogeneous linear system.

Reader's Guide

The concept of affine space is significant because it isolates the geometric properties of parallelism and ratio of lengths from those of distance and angle, providing a framework for affine geometry. This abstraction allows mathematicians to study structures where the notion of an origin is not fixed, making affine spaces fundamental in fields such as computer graphics, physics, and algebraic geometry. The dimension of an affine space is defined as the dimension of its associated vector space of translations. Affine subspaces, including affine lines, planes, and hyperplanes, are central to understanding linear systems and geometric transformations. The informal characterization by Marcel Berger—'An affine space is nothing more than a vector space whose origin we try to forget about'—highlights how affine spaces arise naturally from vector spaces by adding translations to linear maps. This perspective clarifies why affine combinations (linear combinations with coefficients summing to 1) are invariant under choice of origin, a property that distinguishes affine from linear structures.

Did You Know?

• An affine space has no distinguished point that serves as an origin.
• Through any pair of points in an affine space, an infinite straight line can be drawn.

Stripping Euclidean Geometry to Its Parallel Core

Affine space occupies a distinctive niche between pure geometry and algebraic structure. Rather than inheriting the full metric apparatus of Euclidean space—where distances, angles, and measurements define the landscape—affine space deliberately discards those notions. What survives is a framework organized entirely around parallelism and the proportional relationships between parallel line segments. This makes it the natural home of affine geometry, a discipline focused on properties that persist under transformations preserving parallelism. The basic building blocks are points, conceived as zero-dimensional locations with no size or shape. From these, higher-dimensional objects emerge: a line through two points, a plane through three non-collinear points, and, in general, a k-dimensional flat through k plus one points in general position. The parallelism axiom anchors the whole structure: within any plane, two lines either meet at exactly one point or remain parallel indefinitely, and through any given point one can draw a unique line parallel to a specified one. Lines sharing this parallel relationship are said to possess a common direction, forming equivalence classes that give the space its geometric character.

Points Without an Anchor: The Geometry of Locations

One of the most striking features of affine space is the deliberate absence of a privileged origin. In a vector space, the zero vector plays a special role, and vectors can be freely added, scaled, and combined. In an affine space, no such distinguished point exists. Points cannot be added together, nor can they be multiplied by scalars in any meaningful sense. What the structure does permit is the notion of displacement: the difference between two points yields a free vector, also called a translation or displacement vector. Adding such a vector to a point produces a new point, effectively translating the original by that amount. Beyond this, the only legitimate algebraic operation on points is the affine combination—a weighted sum in which the numerical coefficients add up to exactly one. The result is always another point in the same space, and the coefficients themselves define a barycentric coordinate system relative to the flat determined by the original points. This restriction is not a limitation but a feature: it ensures the geometry is independent of any arbitrary choice of reference point, making the structure genuinely affine rather than linear.

The Alice-and-Bob Paradox of Origin

A particularly illuminating way to grasp affine structure comes from imagining two observers who disagree about where the origin lies. Alice knows the true origin, while Bob insists that some other point p is the origin. When Bob attempts to add two vectors a and b using the parallelogram rule from his chosen origin, he actually computes p plus (a minus p) plus (b minus p)—a result that differs from Alice's straightforward sum. This disagreement extends to any linear combination: Alice and Bob will generally produce different answers for the same formal expression. Yet a remarkable exception exists. If the coefficients in a linear combination sum to one, both observers arrive at the identical point. For instance, the combination λa plus (1−λ)b yields the same location regardless of which origin is assumed, because Bob's computation simplifies algebraically to the same expression. This invariance under a shift of origin is precisely what defines the affine structure. While only Alice has access to the full linear structure, both share the affine structure—the collection of all combinations whose coefficients total one. A set equipped with this shared structure is, by definition, an affine space.

Affine Subspaces and the Shadow of Linear Algebra

The relationship between affine and vector spaces is one of elegant duality. Any vector space can be reinterpreted as an affine space simply by forgetting which element serves as the zero vector; in that affine reading, the former zero vector is merely the origin, one point among many. Conversely, given a vector space, one can generate affine subspaces by taking a linear subspace and translating every element of it by a fixed displacement vector. The resulting set is no longer anchored at the origin—it has been shifted away—and is therefore affine rather than linear. In finite-dimensional settings, this construction has a concrete algebraic interpretation: an affine subspace is precisely the solution set of an inhomogeneous system of linear equations, while the collection of all displacement vectors connecting points within that subspace forms the solution set of the corresponding homogeneous system, which is a genuine linear subspace. The dimension of an affine space is defined as the dimension of its associated translation vector space. A one-dimensional affine space is an affine line, a two-dimensional one is an affine plane, and a subspace of dimension n minus one within an n-dimensional space is called an affine hyperplane.