Row, Column, and Space — Relationships Among Row Space, Column Space, and Null Space - Linear Algebra

SECTION 1

Historical Context & Motivation

For centuries, mathematicians have been solving systems of equations — sets of rules with multiple unknowns. Think of a mystery where you have several clues, and you need to figure out the values that make every clue true at the same time. By the 1800s, mathematicians began organizing these clues into neat rectangular grids called matrices. This organization made it much easier to spot patterns and solve problems.

As mathematicians studied matrices more carefully, they realized that the rows and columns of a matrix each tell a different part of the story. The row space, column space, and null space became the three main "lenses" for understanding what a matrix can and cannot do. Together, they form a complete picture.

1693 • Leibniz Uses Determinants

Gottfried Wilhelm Leibniz introduces a method for solving systems of equations using determinants, an early step toward matrix theory.

1850 • Sylvester Coins "Matrix"

James Joseph Sylvester invents the word "matrix" (Latin for "womb") to describe a rectangular array of numbers that generates determinants.

1878 • Frobenius & Rank

Ferdinand Georg Frobenius defines the rank of a matrix, connecting the number of independent rows to the number of independent columns for the first time.

1955 • The Four Fundamental Subspaces

Mathematicians formalize the four fundamental subspaces of a matrix, showing how row space, column space, null space, and left null space fit together perfectly.

The big question that drove this entire field forward was: When you multiply a matrix by a vector, what outputs are possible, and what inputs get "lost" (sent to zero)? Understanding row space, column space, and null space answers this question completely.

SECTION 2

Core Principles & Definitions

Before we dive into how these spaces relate, let's define each one clearly. Imagine a matrix as a machine: you feed in a vector (a list of numbers), and the machine spits out another vector. The three spaces describe what the machine can produce and what it ignores.

Column Space

The column space is the set of all possible outputs when you multiply the matrix by any input vector. It is built from all combinations of the matrix's columns.

Row Space

The row space is the set of all combinations of the matrix's rows. It lives in the "input side" and tells you which input directions the matrix actually uses.

Null Space

The null space is the set of all input vectors that the matrix sends to the zero vector. These are the inputs the matrix completely ignores.

Rank

The rank is the number of truly independent rows (or columns). It equals the dimension of both the row space and the column space.

✦

KEY TAKEAWAY

Think of a matrix like a spotlight on a stage. The column space is the lit area on the wall (what the audience can see). The row space is the part of the stage actually illuminated by the light. The null space is the dark part of the stage — no matter what happens there, the audience sees nothing. The rank tells you how wide the spotlight beam is.

SECTION 3

Visual Explanation — The Big Picture

The diagram below shows how a matrix A acts as a transformation from the input space (on the left) to the output space (on the right). Notice how the input space splits into two perpendicular parts: the row space and the null space. The matrix maps the row space onto the column space, and it crushes the entire null space down to the zero vector.

The input space splits into the row space (dimension r) and the null space (dimension n − r). They are perpendicular. The matrix maps the row space onto the column space and crushes the null space to zero.

This diagram captures the most important idea in the entire lesson. The row space and null space are perpendicular complements inside the input space — their dimensions always add up to the total number of columns n. Meanwhile, the column space lives in the output space and has the same dimension as the row space. That shared dimension is the rank.

SECTION 4

Mathematical Framework

Let's put numbers to the ideas. Suppose A is an m × n matrix (m rows and n columns) with rank r. We can describe the three spaces and their dimensions using two elegant equations.

RANK EQUALS COLUMN RANK

dim(Row Space) = dim(Column Space) = r

The number of independent rows always equals the number of independent columns. This value is the rank r of the matrix.

RANK–NULLITY THEOREM

r + dim(Null Space) = n

Here n is the number of columns. The rank r plus the dimension of the null space (called the nullity) always equals n. If the matrix has 5 columns and rank 3, the null space must have dimension 5 − 3 = 2.

NULL SPACE DEFINITION

Null(A) = { x ∈ ℝⁿ : Ax = 0 }

The null space is the collection of all vectors x such that Ax equals the zero vector. In everyday language: it's every input that the matrix "ignores."

💡

Why Does Row Rank = Column Rank?

When you row-reduce a matrix to echelon form, each pivot marks one independent row AND one independent column. The number of pivots is the same whether you count by rows or by columns. That's why the two dimensions match!

There is also a fourth space called the left null space, which is the null space of Aᵀ (the transpose of A). It lives in the output space and is perpendicular to the column space. Its dimension is m − r. Together, these four spaces are called the four fundamental subspaces of a matrix.

SECTION 5

Detailed Breakdown — The Four Fundamental Subspaces

The diagram below organizes all four subspaces into a single picture. Notice the perpendicularity relationships: in the input space, the row space is perpendicular to the null space; in the output space, the column space is perpendicular to the left null space. Every vector in ℝⁿ can be split into a row-space part and a null-space part, and every vector in ℝᵐ can be split into a column-space part and a left-null-space part.

All four fundamental subspaces and their dimensions. The row space and null space are perpendicular in ℝⁿ; the column space and left null space are perpendicular in ℝᵐ.

Subspace	Lives In	Dimension	How to Find It
Row Space	ℝⁿ (input)	r	Non-zero rows of row echelon form
Column Space	ℝᵐ (output)	r	Pivot columns from original matrix A
Null Space	ℝⁿ (input)	n − r	Solve Ax = 0 using free variables
Left Null Space	ℝᵐ (output)	m − r	Solve Aᵀy = 0

Summary of the four fundamental subspaces for an m × n matrix of rank r

SECTION 6

Worked Example

Let's find all three main spaces for a concrete matrix and verify the rank–nullity theorem.

Finding Row Space, Column Space, and Null Space

1
Step 1 — Write the Matrix

Consider the 2 × 3 matrix A = [[1, 2, 3], [2, 4, 6]]. This matrix has m = 2 rows and n = 3 columns.

A is a 2 × 3 matrix
2
Step 2 — Row Reduce to Echelon Form

Replace row 2 with (row 2 − 2 × row 1): [[1, 2, 3], [0, 0, 0]]. The second row becomes all zeros because row 2 was just 2 times row 1.

Echelon form: [[1, 2, 3], [0, 0, 0]]
3
Step 3 — Identify the Rank

There is only 1 non-zero row in the echelon form, and only 1 pivot (in column 1). So the rank r = 1.

Rank r = 1
4
Step 4 — Find the Row Space

The row space is spanned by the non-zero rows of the echelon form. So the row space = span{(1, 2, 3)}. It's a line through the origin in ℝ³. Its dimension is 1, matching the rank.

Row Space = span{(1, 2, 3)}, dim = 1
5
Step 5 — Find the Column Space

The pivot is in column 1, so we take column 1 from the original matrix A. Column 1 of A is (1, 2). The column space = span{(1, 2)}. It's a line through the origin in ℝ². Its dimension is also 1.

Column Space = span{(1, 2)}, dim = 1
6
Step 6 — Find the Null Space

Solve Ax = 0 using the echelon form. With x₁ as the pivot variable and x₂, x₃ as free variables: x₁ + 2x₂ + 3x₃ = 0, so x₁ = −2x₂ − 3x₃. Setting x₂ = 1, x₃ = 0 gives (−2, 1, 0). Setting x₂ = 0, x₃ = 1 gives (−3, 0, 1). The null space is spanned by these two vectors.

Null Space = span{(−2, 1, 0), (−3, 0, 1)}, dim = 2
7
Step 7 — Verify Rank–Nullity

rank + nullity = 1 + 2 = 3 = n (the number of columns). ✓ The theorem checks out!

1 + 2 = 3 ✓

SECTION 7

Comparing the Three Spaces

Students often mix up the row space and column space, or confuse the null space with "the leftover columns." The table below highlights the key differences and similarities so you can keep them straight.

Feature	Row Space	Column Space	Null Space
Where it lives	Input space ℝⁿ	Output space ℝᵐ	Input space ℝⁿ
Dimension	r (rank)	r (rank)	n − r (nullity)
Built from	Rows of A	Columns of A	Solutions to Ax = 0
Perpendicular to	Null space	Left null space	Row space
Key question it answers	Which input directions matter?	What outputs are reachable?	Which inputs are invisible?

Side-by-side comparison of the three main subspaces

✦

KEY TAKEAWAY

Think of a vending machine. The row space is like the set of button combinations the machine actually responds to. The column space is the menu of snacks you can actually get. The null space is like pressing buttons that are broken — no matter what combination you try with them, nothing comes out. The rank–nullity theorem guarantees that the working buttons plus the broken buttons account for every button on the machine.

SECTION 8

Connection to Advanced Theory

The ideas of row space, column space, and null space are not just classroom tools — they form the backbone of many advanced topics in mathematics, science, and engineering. The table below shows how these foundational concepts connect to more advanced areas you might encounter in college.

This Lesson's Concept	Advanced Extension
Column space (reachable outputs)	Image of a linear transformation — central in abstract algebra and functional analysis
Null space (invisible inputs)	Kernel of a linear map — key to understanding eigenvalues, SVD, and differential equations
Rank–nullity theorem	Dimension theorem for linear maps — generalizes to infinite-dimensional spaces in functional analysis
Perpendicularity of row space & null space	Orthogonal complements — fundamental to least-squares regression and data science
Four fundamental subspaces	Singular Value Decomposition (SVD) — used in image compression, recommendation systems, and machine learning

How today's concepts lead to advanced mathematics

If you continue studying linear algebra, you will encounter the Singular Value Decomposition (SVD), which beautifully decomposes any matrix into rotations and scalings that act on exactly these four subspaces. The row space gets rotated into the column space, and the null space gets rotated into the left null space. Everything you learned today is the foundation for that powerful tool.

SECTION 9

Practice Problems

Question 1: CONCEPTUAL

In your own words, explain why the dimension of the row space must always equal the dimension of the column space. Use the idea of pivots in your explanation.

Show answer

When we row-reduce a matrix to echelon form, each pivot sits at the intersection of one independent row and one independent column. The total number of pivots counts both the independent rows and the independent columns. Since the same set of pivots determines both counts, the dimension of the row space (number of independent rows) must equal the dimension of the column space (number of independent columns). This shared value is the rank of the matrix.

Question 2: BASIC CALCULATION

Let A = [[1, 0, 2], [0, 1, 3]]. Find the rank, the dimension of the null space, and a vector that spans the null space.

Show answer

The matrix is already in reduced row echelon form with pivots in columns 1 and 2, so the rank r = 2. By the rank–nullity theorem, dim(Null Space) = n − r = 3 − 2 = 1. To find the null space, solve Ax = 0: x₁ + 2x₃ = 0 and x₂ + 3x₃ = 0. Setting x₃ = 1 gives x₁ = −2 and x₂ = −3. So the null space is spanned by (−2, −3, 1).

Question 3: INTERMEDIATE

Consider the matrix B = [[1, 2, 1], [3, 6, 3], [2, 4, 2]]. Find the row space, column space, and null space. Verify the rank–nullity theorem.

Show answer

Row-reducing: R₂ ← R₂ − 3R₁ and R₃ ← R₃ − 2R₁ gives [[1, 2, 1], [0, 0, 0], [0, 0, 0]]. The rank is 1. The row space is span{(1, 2, 1)}. The pivot is in column 1, so the column space is span{(1, 3, 2)} (column 1 of the original B). For the null space, x₁ = −2x₂ − x₃ with x₂ and x₃ free, giving span{(−2, 1, 0), (−1, 0, 1)} with dimension 2. Check: 1 + 2 = 3 = n. ✓

Question 4: APPLIED

A chemical reaction system has 3 reactions and 4 chemical species. The stoichiometry matrix is S = [[1, −1, 0, 0], [0, 1, −1, 0], [1, 0, −1, 0]]. Find the rank of S and the dimension of its null space. What does the null space tell you about the chemical species?

Show answer

Row-reduce: R₃ ← R₃ − R₁ gives [[1, −1, 0, 0], [0, 1, −1, 0], [0, 1, −1, 0]]. Then R₃ ← R₃ − R₂ gives [[1, −1, 0, 0], [0, 1, −1, 0], [0, 0, 0, 0]]. The rank is 2, so dim(Null Space) = 4 − 2 = 2. This means there are 2 independent conservation laws among the 4 species. The null space vectors represent combinations of species whose total amounts are conserved no matter how the reactions proceed.

Question 5: CRITICAL THINKING

Suppose A is a 5 × 5 matrix and you know that Ax = 0 has only the trivial solution x = 0. What can you conclude about the row space, column space, and rank of A? Can every vector in ℝ⁵ be an output of A? Explain your reasoning.

Show answer

If Ax = 0 has only the trivial solution, the null space is {0}, so the nullity is 0. By the rank–nullity theorem, rank + 0 = 5, so the rank is 5. This means the row space has dimension 5 in ℝ⁵, so it IS all of ℝ⁵. The column space also has dimension 5 in ℝ⁵, so it IS all of ℝ⁵. Therefore, yes — every vector in ℝ⁵ can be an output. The matrix is invertible: it loses no information and can reach every possible output. This is equivalent to saying the matrix has full rank.

SUMMARY

Lesson Summary

Summary

Every matrix has three key spaces that together reveal its structure. The row space (spanned by the matrix's rows) and the null space (all inputs sent to zero) live in the input space ℝⁿ and are perpendicular to each other. The column space (all reachable outputs) lives in the output space ℝᵐ. The rank r is the shared dimension of the row space and column space, and the rank–nullity theorem guarantees that r + dim(Null Space) = n, the number of columns.

To find these spaces, row reduce the matrix: non-zero rows of the echelon form span the row space, pivot columns of the original matrix span the column space, and free variables generate the null space. These three spaces, along with the left null space, form the four fundamental subspaces — the most important structural framework in all of linear algebra.