Column Space & Null Space — Column Space and Null Space...

SECTION 1

Historical Context & Motivation

Matrices are everywhere — from computer graphics to weather prediction. But for centuries, mathematicians wondered: what can a matrix actually do? When you multiply a matrix by different vectors, which outputs are possible, and which inputs give you zero? These questions led to two powerful ideas called the column space and the null space of a matrix.

1750s

Early Systems of Equations

Mathematicians like Gabriel Cramer and Leonhard Euler developed methods to solve systems of linear equations, laying the groundwork for matrix theory.

1850s

Arthur Cayley Formalizes Matrices

British mathematician Arthur Cayley introduced the modern idea of a matrix as a rectangular array of numbers with its own rules of algebra.

1870s

Rank and Nullity Emerge

Georg Frobenius and others studied the rank (dimension of the column space) and nullity (dimension of the null space), connecting them in a beautiful theorem.

1900s–Today

Modern Applications

Column space and null space ideas power machine learning, search engines, data compression, and the GPS system on your phone.

The central question is this: given a matrix A, what outputs can Ax produce, and which inputs x get "crushed" to the zero vector? Answering these two questions reveals the structure hiding inside every matrix.

SECTION 2

Core Principles & Definitions

Before we dive into column space and null space, let's make sure we're comfortable with a few building blocks. A matrix is a grid of numbers arranged in rows and columns. A vector is a list of numbers stacked in a column (think of it like a single column of the matrix). When you multiply a matrix A by a vector x, you get a new vector b. We write this as Ax = b.

Column Space (Col A)

The set of all possible outputs b you can get from Ax = b. It's built from all combinations of the columns of A.

Null Space (Nul A)

The set of all inputs x that produce zero — that is, every vector x where Ax = 0.

Linear Combination

A recipe for mixing vectors: multiply each vector by a number (called a scalar) and add the results together.

Span

The collection of every possible linear combination of a set of vectors. The column space is the span of the matrix's columns.

✦ KEY TAKEAWAY

Think of a matrix like a vending machine. The column space is the menu of all snacks the machine can possibly give you. The null space is the set of button combinations that give you nothing — you press buttons but nothing comes out! Every matrix has both a menu (column space) and a set of "do-nothing" inputs (null space).

SECTION 3

Visual Explanation

Let's visualize what a matrix does. Imagine the matrix A as a machine that takes in a 2D vector (an input arrow) and produces a 2D output arrow. The diagram below shows how the columns of A determine the column space, and how certain input vectors land on the zero vector.

The shaded region on the right represents the column space — all outputs b that are reachable. The pink dashed vector on the left is a null space vector: it gets mapped to the zero point on the right.

Notice the key relationship: the column space lives in the output space (it shows what the matrix can produce), while the null space lives in the input space (it shows which inputs get "squished" to zero). These two spaces together tell you almost everything about how a matrix behaves.

SECTION 4

Mathematical Framework

Now let's see the formal definitions. Suppose A is an m × n matrix with columns c₁, c₂, …, cₙ.

COLUMN SPACE DEFINITION

Col(A) = { b : b = x₁c₁ + x₂c₂ + … + xₙcₙ for some scalars x₁, x₂, …, xₙ }

This means the column space is every possible linear combination of the columns of A. If a vector b is in Col(A), then the equation Ax = b has at least one solution.

NULL SPACE DEFINITION

Nul(A) = { x : Ax = 0 }

The null space collects every input vector x that produces the zero vector when multiplied by A. The zero vector itself (x = 0) is always in the null space.

RANK–NULLITY THEOREM

dim(Col A) + dim(Nul A) = n

Here, n is the number of columns. The dimension of the column space is called the rank, and the dimension of the null space is called the nullity. Together they always add up to the number of columns.

💡 Why Does This Matter?

The rank–nullity theorem is like a conservation law. If the column space is "big" (high rank), then the null space must be "small" (low nullity), and vice versa. A matrix can't have a big menu of outputs and also crush lots of inputs to zero — there's a trade-off!

SECTION 5

How to Find Column Space & Null Space

To find these two spaces, we use a technique called row reduction (also known as Gaussian elimination). You convert the matrix to row echelon form (REF) or reduced row echelon form (RREF). The diagram below walks through the process for both spaces.

Flowchart showing how to find both spaces. For the column space, identify pivot columns in RREF, then go back to the original matrix and take those columns. For the null space, solve Ax = 0 by expressing pivot variables in terms of free variables.

⚠️ Important Warning

When finding the column space, always use the columns from the original matrix A, not the row-reduced version! Row reduction changes the column space, so you only use RREF to identify which columns are pivot columns.

SECTION 6

Worked Example

Let's work through a complete example with a small matrix. We'll find both the column space and the null space step by step.

Find Col(A) and Nul(A) for a 3×3 Matrix

Step 1 — Write the matrixConsider the matrix A = [[1, 2, 3], [2, 4, 6], [0, 1, 1]]. Written as rows: Row 1 is [1, 2, 3], Row 2 is [2, 4, 6], Row 3 is [0, 1, 1].

A has 3 rows and 3 columns, so n = 3.

Step 2 — Row reduce to RREFReplace Row 2 with Row 2 − 2 × Row 1: [2 − 2, 4 − 4, 6 − 6] = [0, 0, 0]. The matrix becomes [[1, 2, 3], [0, 0, 0], [0, 1, 1]]. Swap Row 2 and Row 3: [[1, 2, 3], [0, 1, 1], [0, 0, 0]]. Replace Row 1 with Row 1 − 2 × Row 2: [1, 0, 1], [0, 1, 1], [0, 0, 0].

RREF = [[1, 0, 1], [0, 1, 1], [0, 0, 0]]. Pivots are in columns 1 and 2.

Step 3 — Find the column spaceThe pivot columns are columns 1 and 2. Go back to the original matrix A. Column 1 of A is [1, 2, 0] and column 2 of A is [2, 4, 1].

Col(A) = Span{ [1, 2, 0], [2, 4, 1] }. The rank is 2.

Step 4 — Find the null spaceColumn 3 is a free variable column (no pivot). Set x₃ = t (a free parameter). From the RREF, Row 2 gives x₂ + x₃ = 0, so x₂ = −t. Row 1 gives x₁ + x₃ = 0, so x₁ = −t. The solution is x = t × [−1, −1, 1].

Nul(A) = Span{ [−1, −1, 1] }. The nullity is 1.

Step 5 — Verify rank–nullityWe found rank = 2 and nullity = 1. Check: rank + nullity = 2 + 1 = 3 = n (the number of columns). ✓

The rank–nullity theorem is confirmed!

SECTION 7

Column Space vs. Null Space — Key Differences

Column space and null space are two sides of the same coin. They describe different aspects of the same matrix, and understanding their differences is crucial.

Side-by-side comparison of column space and null space

Feature	Column Space (Col A)	Null Space (Nul A)
Question it answers	What outputs can Ax produce?	What inputs make Ax = 0?
Lives in	Output space (ℝᵐ)	Input space (ℝⁿ)
Built from	Pivot columns of original A	Solutions to Ax = 0
Dimension name	Rank	Nullity
Always contains	The zero vector (always reachable via x = 0)	The zero vector (A × 0 = 0 always)

✦ KEY TAKEAWAY

Think of a projector shining light on a wall. The column space is like the wall area that gets lit up — the range of images the projector can create. The null space is like the set of adjustments you could make to the input slide that produce no change on the wall. A powerful projector (high rank matrix) lights up more of the wall, leaving fewer "invisible" adjustments.

SECTION 8

Connections to Advanced Topics

Column space and null space are just the beginning. As you continue in linear algebra, you'll meet two more related spaces: the row space and the left null space. Together, these four spaces form what mathematician Gilbert Strang calls the "Fundamental Theorem of Linear Algebra."

How introductory concepts connect to advanced linear algebra

Concept	What You Know Now	Where It Leads
Column Space	Span of columns of A; the "reachable outputs"	Image/range of a linear transformation; orthogonal complement of left null space
Null Space	Solutions to Ax = 0; the "invisible inputs"	Kernel of a linear transformation; orthogonal complement of row space
Rank–Nullity	rank + nullity = number of columns	Generalizes to dimension theorem for linear maps between any vector spaces

In data science, the null space helps identify redundant features in a dataset. In engineering, it reveals the "degrees of freedom" in a mechanical system. The column space tells engineers which forces or loads a structure can withstand. These are not just abstract ideas — they have real, powerful applications.

SECTION 9

Practice Problems

PROBLEM 1 — CONCEPTUAL

In your own words, explain the difference between the column space and the null space of a matrix. Where does each one "live" — in the input space or the output space?

PROBLEM 2 — BASIC CALCULATION

Let A = [[1, 3], [2, 6]]. Row reduce A to RREF. What is the rank of A? What is the nullity of A?

PROBLEM 3 — INTERMEDIATE

Using the matrix A = [[1, 3], [2, 6]] from Problem 2, find a vector that spans the null space of A. Then verify your answer by computing Ax.

PROBLEM 4 — APPLIED

A video game uses a 3×3 matrix T to transform character positions. After row reduction, T has pivots in columns 1 and 3. The original columns of T are c₁ = [2, 0, 1], c₂ = [4, 0, 2], c₃ = [1, 3, 0]. What is the column space of T, and what does it mean for the game's characters?

PROBLEM 5 — CRITICAL THINKING

If a 4×6 matrix A has a null space with dimension 4, what is the rank of A? Can the equation Ax = b have a solution for every possible vector b in ℝ⁴? Explain your reasoning using the rank–nullity theorem.

SUMMARY

Summary

Every matrix A has two fundamental spaces. The column space (Col A) is the set of all possible outputs b in the equation Ax = b — it's the span of the pivot columns of the original matrix. The null space (Nul A) is the set of all input vectors x that get mapped to zero — it's found by solving Ax = 0 after row reduction.

The dimensions of these two spaces are linked by the rank–nullity theorem: rank + nullity = n (the number of columns). A high rank means the matrix can reach many outputs; a high nullity means many inputs are "invisible" to the matrix. To find the column space, identify pivot columns via row reduction and select those columns from the original matrix. To find the null space, solve for the free variables in the RREF system.