The Column Space, the Null Space and the Rank

I. Review: A Transformation “Squashes” Space

An invertible transformation preserves the full dimension of space. But a non-invertible transformation squashes space flat — and then you need to ask two questions:

  1. Where do all the output vectors “land”? → the column space
  2. Which vectors get squashed to the origin? → the null space

II. The Column Space

Definition

The column space of a matrix MM is the set of all possible outputs:

Col(M)={Mx∣x∈Rn}\text{Col}(M) = \{ M\mathbf{x} \mid \mathbf{x} \in \mathbb{R}^n \}

Geometric meaning: what the transformed space looks like.

Why it is called the “column space”

Recall that Mx=x1⋅(column 1)+x2⋅(column 2)+⋯M\mathbf{x} = x_1\cdot(\text{column 1}) + x_2\cdot(\text{column 2}) + \cdots

The output is always a linear combination of the columns! Hence:

column space=the span of all the column vectors\text{column space} = \text{the span of all the column vectors}
Example: M = [[1, 2], [2, 4]] (column 2 = column 1 × 2)
column space = span{(1,2)} = a straight line
⟹ every output of this transformation lies on the line y = 2x

Deciding whether b\mathbf{b} lies in the column space

The equation Mx=bM\mathbf{x} = \mathbf{b} has a solution ⟺ b∈Col(M)\mathbf{b} \in \text{Col}(M).

Geometrically: b\mathbf{b} has to lie inside the transformed space for some x\mathbf{x} to map onto it.

III. The Null Space

Definition

The null space of a matrix MM is the set of all vectors mapped to the origin:

Nul(M)={x∣Mx=0}\text{Nul}(M) = \{ \mathbf{x} \mid M\mathbf{x} = \mathbf{0} \}

Geometric meaning: the vectors that get “squashed flat to the origin” in the course of the transformation.

IV. The Rank and the Rank-Nullity Theorem

The definition of the rank

The rank = the dimension of the column space = the dimension of the transformed space.

  • Full rank (rank = 2 for a 2×2 matrix): the transformation squashes nothing, and it is invertible
  • Rank deficient (rank = 1): squashed into a line
  • rank = 0: squashed into a single point

The Rank-Nullity Theorem

 dim⁡(Col(M))+dim⁡(Nul(M))=n \boxed{\,\dim(\text{Col}(M)) + \dim(\text{Nul}(M)) = n\,}

where nn is the dimension of the input space (the number of columns of the matrix).

Geometric intuition: the input space is split into two parts — “the part that becomes the output space” (the column space) and “the part that gets squashed to the origin” (the null space). The two together make up exactly the whole dimension of the input space.

Example: M = [[1, 2], [2, 4]] (2×2, squashed into a line)
column space: one line → dimension 1
null space: the solutions of Mx = 0 are x₂ = -x₁/2 (a line) → dimension 1
rank-nullity: 1 + 1 = 2 ✓

The unification with the determinant and invertibility

ConditionMeaning
det ≠ 0full rank, the null space contains only the zero vector, invertible
det = 0rank deficient, the null space contains nonzero vectors, not invertible

V. A Complete Example

M=(1224),det⁡=1×4−2×2=0M = \begin{pmatrix} 1 & 2 \\ 2 & 4 \end{pmatrix}, \quad \det = 1\times4 - 2\times2 = 0
  1. Column space: the two columns are collinear, so Col(M)=span{(1,2)}\text{Col}(M) = \text{span}\{(1,2)\} (a line, dimension 1)
  2. Null space: solve (1224)(x1x2)=(00)\begin{pmatrix} 1 & 2 \\ 2 & 4 \end{pmatrix}\begin{pmatrix} x_1 \\ x_2 \end{pmatrix} = \begin{pmatrix} 0 \\ 0 \end{pmatrix}
    • This gives x1+2x2=0x_1 + 2x_2 = 0, that is, x=t(−21)\mathbf{x} = t\begin{pmatrix} -2 \\ 1 \end{pmatrix} (a line, dimension 1)
  3. Rank: rank = 1
  4. Verify the rank-nullity theorem: 1 + 1 = 2 ✓
  5. Conclusion: MM is not invertible, and the equation Mx=bM\mathbf{x} = \mathbf{b} either has no solution (b\mathbf{b} is not in the column space) or has infinitely many solutions

Summary

Symbols Used in This Article

SymbolTypePronunciation/ExplanationMeaning in This Article
Col(M)\text{Col}(M)Mathematical symbolcolumn space of MThe column space: the set of all possible outputs of the matrix MM
Nul(M)\text{Nul}(M)Mathematical symbolnull space of MThe null space: the set of vectors mapped to the origin 0\mathbf{0} by MM
Mx=0M\mathbf{x} = \mathbf{0}Mathematical symbolM x equals zeroThe homogeneous system whose solutions form the null space
dim⁡(Col(M))+dim⁡(Nul(M))=n\dim(\text{Col}(M)) + \dim(\text{Nul}(M)) = nMathematical symbolrank-nullityThe rank-nullity theorem: the dimension of the column space plus that of the null space equals the input dimension
nnMathematical symbolnThe dimension of the input space (the number of columns of the matrix)
Rn\mathbb{R}^nMathematical symbolR nThe nn-dimensional real vector space, appearing in x∈Rn\mathbf{x} \in \mathbb{R}^n
z\mathbf{z}Mathematical symbolvector zA nonzero vector in the null space, so that Mz=0M\mathbf{z} = \mathbf{0}

Found an issue with this content?

Tell us what could be improved.