Ma · Mathematics

What Is an Affine Space?

Published By the 9to5Science editorial team

At a glance

  1. 1. Separate points and vectors

    An affine space has points and a vector space of displacements, but no preferred zero point.

  2. 2. Compare two points

    Subtracting one point from another gives a vector; adding a vector to a point gives another point.

  3. 3. Choose an origin if useful

    Picking a point lets you assign coordinates, but the choice is a convenience, not part of the space.

  4. 4. Use affine combinations

    Weighted combinations of points are intrinsic when their coefficients add up to one.

Advertisement

An affine space looks much like a vector space geometrically: it has lines, directions, and displacements. The key difference is that its elements are points, not vectors, so no point is singled out as the zero. This is the right framework for discussing geometry without making an arbitrary choice of origin.

Definition: points and displacements

An affine space AA over a vector space VV is a nonempty set of points equipped with an operation that takes a point P∈AP\in A and a vector v∈Vv\in V to another point, written P+v∈AP+v\in A. The operation must satisfy two properties:

  1. Moving by zero does nothing, and successive moves combine as expected: P+0=PP+0=P and (P+u)+v=P+(u+v)(P+u)+v=P+(u+v).
  2. For every pair of points P,Q∈AP,Q\in A, there is exactly one vector v∈Vv\in V such that P+v=QP+v=Q.

That unique vector is called the displacement from PP to QQ, and is written Q−PQ-P. Thus, subtracting points gives a vector, while adding a vector to a point gives a point. Subtracting two vectors gives another vector, but adding two points is not an operation supplied by the definition.

The vector space VV is sometimes called the translation space or direction space of AA. If VV has dimension nn, then AA is an nn-dimensional affine space.

A familiar example

The ordinary plane can be viewed as an affine space whose translation space is R2\mathbb{R}^2. A point PP can be moved by a vector v=(a,b)v=(a,b): move aa units horizontally and bb units vertically. Given two points, their difference is the vector from the first to the second.

For a coordinate example, take points with coordinates P=(2,1)P=(2,1) and Q=(6,5)Q=(6,5). Their displacement is

Q−P=(6−2,5−1)=(4,4).Q-P=(6-2,5-1)=(4,4).

This says how to move from PP to QQ. The midpoint is the point reached by moving halfway along that displacement:

P+12(Q−P)=(2,1)+(2,2)=(4,3).P+\frac{1}{2}(Q-P)=(2,1)+(2,2)=(4,3).

The coordinates here use a chosen origin. The geometric displacement and midpoint do not depend on which origin you choose, even though the coordinate descriptions may change.

Why there is no preferred origin

In a vector space, the zero vector is part of the structure. It gives a natural way to interpret every vector as a point. In an affine space, no point has that special role. You can choose any point OO as an origin and represent a point PP by the vector P−OP-O. This gives a one-to-one correspondence between points of AA and vectors of VV.

But a different choice of origin gives different position vectors. If you replace OO with O′O', every coordinate vector is shifted by the same amount. Differences between points remain unchanged. This is why affine geometry treats locations and displacements differently: a location depends on the origin, but a displacement does not.

Affine combinations of points

Although you cannot add points arbitrarily, you can form an affine combination of points when the coefficients sum to one. For points P1,…,PkP_1,\ldots,P_k and scalars λ1,…,λk\lambda_1,\ldots,\lambda_k with ∑iλi=1\sum_i\lambda_i=1, the combination is independent of which point is temporarily chosen as a reference. For example, the midpoint of PP and QQ is

12P+12Q,\frac{1}{2}P+\frac{1}{2}Q,

where the notation means the midpoint construction, not ordinary addition of points. More generally, the point with affine parameter tt on the line from PP to QQ is P+t(Q−P)P+t(Q-P); it lies on the segment when 0≤t≤10\le t\le 1, and the midpoint occurs at t=12t=\frac{1}{2}.

The condition that the coefficients add to one is essential. If you change the origin, the extra shift cancels in an affine combination precisely when the coefficients sum to one. Such combinations describe constructions like points on lines, centroids, and convex combinations.

Affine spaces and affine subspaces

An affine space is the whole setting of points and displacements. An affine subspace is a subset of an affine space formed by translating a vector subspace. If WW is a vector subspace of VV and PP is a point, then

P+W={P+w:w∈W}P+W=\{P+w:w\in W\}

is an affine subspace. In the plane, a line is a point translated along a one-dimensional direction space; a plane in three-dimensional space is a point translated along a two-dimensional direction space.

A linear subspace must contain the zero vector. An affine subspace need not contain the chosen origin. For example, the line y=1y=1 in R2\mathbb{R}^2 is an affine line, but it is not a vector subspace because it does not contain (0,0)(0,0).

Common mistakes

  • Calling points vectors: Points and vectors can have the same coordinate tuples after choosing an origin, but they play different roles. A point minus a point is a vector; a point plus a vector is a point.
  • Assuming an origin is intrinsic: Coordinates require an origin, but the affine space itself does not select one.
  • Confusing affine and linear subspaces: A translated vector subspace is affine; it is linear only when it contains the origin.
  • Combining points with coefficients that do not sum to one: Such an expression generally changes when the origin changes, so it is not an origin-independent affine combination.

References

  1. Marcel Berger, Geometry I

Spotted an error? Send a correction. Read how we write and check articles in our editorial policy.