1.812 min read

Solution Space Shape, Affine Sets, and Dimension

Once we know a system is consistent, the next question is: what does the solution set look like? The answer is always an affine set — a flat, shifted version of a subspace.

For a system of $m$ equations in $n$ unknowns with rank $r$ (the effective number of constraints), the solution set has dimension $n - r$ . When no exceptional behavior occurs, $r = m$ , giving dimension $n - m$ .

An affine set of dimension $k$ is: a single point (dimension 0), a line (dimension 1), a plane (dimension 2), or higher-dimensional flat. These are always "translated" versions of vector subspaces — they don't have to pass through the origin.

Concretely: the general solution is a particular solution (one specific solution) plus any solution to the homogeneous system (which forms a subspace). This is the affine structure.

Formal View

Definition 1.8 — Affine Set

A subset

S \subseteq \mathbb{R}^n

is an affine set (or flat) if there exists a subspace

V \subseteq \mathbb{R}^n

and a point

\mathbf{p}

such that

S = \mathbf{p} + V := \{\mathbf{p} + \mathbf{v} : \mathbf{v} \in V\}

The dimension of

S

equals the dimension of

V

Theorem 1.8 — Solution Sets are Affine

Let

\mathbf{x}_p

be a particular solution to

A\mathbf{x} = \mathbf{b}

, and let

V = \{\mathbf{x} : A\mathbf{x} = \mathbf{0}\}

be the null space. Then the solution set is the affine set

\{\mathbf{x}_p + \mathbf{v} : \mathbf{v} \in V\}

Its dimension equals the number of free variables =

n - \text{rank}(A)

Why This Matters

The structure of solution sets is not arbitrary — it is always a flat, shift of a subspace.

Image deblurring: the solution set being affine means we can parameterize all possible deblurred images
In control theory, the set of feasible control inputs often forms an affine set
Affine structure enables efficient sampling from solution sets in Monte Carlo methods

Learning Resources

Null space and column space

3Blue1Brown

Understanding the structure of solution sets geometrically.

10 min

Quiz

Question 1

A consistent system with $n = 4$ unknowns and rank $r = 3$ has solution set of dimension:

Question 2

The solution set of a consistent linear system always passes through the origin.

Common Mistakes

Thinking the expected dimension $n - m$ always holds — exceptional cases (redundant or contradictory equations) change this.
Confusing the solution set (an affine set) with the null space (a subspace). They are related but distinct.