LINEAR ALGEBRA • LINEAR TRANSFORMATIONS

Linear Transformation Definition — Definition of a Linear Transformation

Discover how special functions between vectors preserve addition and scaling, unlocking powerful tools across math and science.

Historical Context & Motivation

Have you ever resized a photo, rotated a shape in a video game, or watched a 3-D movie? Behind all of these actions is a powerful mathematical idea called a linear transformation. For centuries, mathematicians worked with equations and geometric shapes separately. The concept of a linear transformation brought them together, creating a single language that connects algebra and geometry.

The story begins with systems of equations that people needed to solve for trade, navigation, and engineering. Over time, brilliant thinkers realized that certain kinds of functions — ones that play nicely with addition and multiplication — deserve a special name and special study.

1750s
Euler & Systems of Equations
Leonhard Euler studied systems of linear equations, laying groundwork for the idea of treating equations as transformations of variables rather than isolated problems.
1844
Grassmann's Vector Spaces
Hermann Grassmann published ideas about multi-dimensional spaces and operations on vectors, hinting at the structure that linear transformations would later formalize.
1855
Cayley & Matrices
Arthur Cayley introduced matrix notation, giving mathematicians a compact way to represent and compute linear transformations using rows and columns of numbers.
1888
Peano's Axioms for Vector Spaces
Giuseppe Peano formally defined vector spaces, providing the precise stage on which linear transformations act.
1900s+
Modern Applications Explode
Linear transformations became essential in computer graphics, data science, quantum physics, and machine learning — powering technology we use every day.

The central question this lesson addresses is: What exactly makes a function between vectors "linear," and why does that matter? By the end, you will be able to state the definition, test whether a function qualifies, and see why this idea is so useful.

Core Principles & Definitions

A linear transformation is a special kind of function that takes in a vector and gives back a vector. What makes it special? It must obey two rules — and only two. If a function follows both rules, it earns the title "linear." If it breaks even one of them, it does not.

1

Vectors In, Vectors Out

A linear transformation T takes a vector from one space (the domain) and produces a vector in another space (the codomain). Think of it as a machine: vector goes in, vector comes out.
2

Additivity (Rule 1)

If you add two vectors first and then transform the sum, you get the same result as transforming each vector separately and then adding the outputs. In symbols: T(u + v) = T(u) + T(v).
3

Homogeneity (Rule 2)

If you scale a vector by a number (a scalar) and then transform it, you get the same result as transforming first and then scaling. In symbols: T(c · v) = c · T(v).
4

Zero Goes to Zero

A quick consequence of the two rules: a linear transformation always sends the zero vector to the zero vector. If T(0) ≠ 0, the function cannot be linear.
KEY TAKEAWAY
Imagine a vending machine that follows two promises. Promise 1: if you insert two coins at once, you get the same snacks as inserting each coin separately. Promise 2: if you double the money, you double the snacks. A linear transformation is a "math machine" that keeps those same two promises — combining inputs and scaling inputs always gives predictable, consistent results.

Visual Explanation

The best way to understand a linear transformation is to see it in action. The diagram below shows two vectors u and v in the original (input) space on the left, and their images T(u) and T(v) in the output space on the right. Notice how the parallelogram formed by u and v maps to a parallelogram formed by T(u) and T(v). Straight lines stay straight, and the origin stays fixed.

The transformation T sends vectors u (cyan) and v (pink) from the input space to the output space. The dashed amber arrow shows that T(u + v) = T(u) + T(v) — the parallelogram shape is preserved.

In the diagram, the key visual clue is that the parallelogram rule is preserved. In a linear transformation, if two vectors form a parallelogram, their images also form a parallelogram. Grids of evenly spaced lines in the input become evenly spaced lines in the output (though they may be stretched, rotated, or reflected). Lines that are parallel before the transformation stay parallel afterward, and the origin never moves.

Mathematical Framework

Now let's write the definition precisely using mathematical notation. A function T from a vector space V to a vector space W is called a linear transformation if and only if it satisfies the following two properties for all vectors u and v in V and every scalar c.

ADDITIVITY PROPERTY
T(u + v) = T(u) + T(v)
T is the transformation, u and v are any two vectors in the domain. This says: transforming a sum equals the sum of the transforms.
HOMOGENEITY PROPERTY
T(c · v) = c · T(v)
c is any real number (scalar) and v is any vector. This says: transforming a scaled vector equals scaling the transformed vector.

These two properties are sometimes combined into a single condition called superposition:

SUPERPOSITION (COMBINED FORM)
T(c₁u + c₂v) = c₁ · T(u) + c₂ · T(v)
For any scalars c₁ and c₂ and any vectors u and v. This single equation captures both additivity and homogeneity at once.

In practice, many linear transformations from ℝⁿ to ℝᵐ can be represented by matrix multiplication. If A is an m × n matrix, then T(v) = A · v defines a linear transformation. You can verify: A(u + v) = Au + Av (additivity), and A(cv) = c(Av) (homogeneity). Matrices are the most common way you will encounter linear transformations in calculations.

MATRIX FORM
T(v) = A · v
A is a matrix with m rows and n columns, v is a column vector with n entries, and T(v) is the resulting column vector with m entries.

Linear vs. Non-Linear — Telling Them Apart

Not every function is a linear transformation. In fact, most functions you have already studied — like squaring a number or adding a constant — are not linear. The diagram below compares what a linear transformation does to a grid versus what a non-linear transformation does. Notice how the linear version keeps grid lines straight and evenly spaced, while the non-linear version warps and curves them.

Left (cyan): A linear transformation stretches the square horizontally but keeps all grid lines straight and evenly spaced. Right (red): The squaring function warps vertical grid lines into curves, violating the definition of linearity.
Common functions tested for linearity
FunctionLinear?Why / Why Not
T(x, y) = (2x, 3y)Yes ✓Scaling each component by a constant satisfies both rules.
T(x, y) = (x + 1, y)No ✗Adding 1 shifts the origin. T(0,0) = (1,0) ≠ (0,0).
T(x, y) = (−y, x)Yes ✓This is a 90° rotation. Rotations are linear transformations.
f(x) = x²No ✗f(2 + 3) = 25, but f(2) + f(3) = 4 + 9 = 13. Additivity fails.
T(x, y) = (x + y, x − y)Yes ✓Each output component is a linear combination of inputs — no squares, no constants added.
💡 Quick Test
If you are unsure whether a function is linear, try two fast checks. First, does T(0) = 0? If not, stop — it is not linear. Second, pick simple vectors (like (1, 0) and (0, 1)) and test additivity and homogeneity with small numbers. One failure is enough to rule it out.

Worked Example

Let's work through a full example. We will determine whether the function T(x, y) = (3x − y, 2x + 4y) is a linear transformation by checking both required properties.

Is T(x, y) = (3x − y, 2x + 4y) a linear transformation?
1
Step 1 — Set Up Two Generic VectorsLet u = (u₁, u₂) and v = (v₁, v₂) be any two vectors in ℝ². We also let c be any real number (scalar).
2
Step 2 — Check Additivity: Compute T(u + v)First, u + v = (u₁ + v₁, u₂ + v₂). Now apply T: T(u + v) = (3(u₁ + v₁) − (u₂ + v₂), 2(u₁ + v₁) + 4(u₂ + v₂)) = (3u₁ + 3v₁ − u₂ − v₂, 2u₁ + 2v₁ + 4u₂ + 4v₂)
T(u + v) = (3u₁ − u₂ + 3v₁ − v₂, 2u₁ + 4u₂ + 2v₁ + 4v₂)
3
Step 3 — Check Additivity: Compute T(u) + T(v)T(u) = (3u₁ − u₂, 2u₁ + 4u₂) and T(v) = (3v₁ − v₂, 2v₁ + 4v₂). T(u) + T(v) = (3u₁ − u₂ + 3v₁ − v₂, 2u₁ + 4u₂ + 2v₁ + 4v₂)
T(u + v) = T(u) + T(v) ✓ Additivity holds!
4
Step 4 — Check Homogeneity: Compute T(c · u)c · u = (cu₁, cu₂). Apply T: T(c · u) = (3(cu₁) − (cu₂), 2(cu₁) + 4(cu₂)) = (c(3u₁) − c(u₂), c(2u₁) + c(4u₂)) = c(3u₁ − u₂, 2u₁ + 4u₂) = c · T(u)
T(c · u) = c · T(u) ✓ Homogeneity holds!
5
Step 5 — ConclusionSince both additivity and homogeneity are satisfied for all vectors and all scalars, T(x, y) = (3x − y, 2x + 4y) is a linear transformation. Notice that each output component is built only from the inputs multiplied by constants and added together — no squares, no extra constants, no absolute values.
T is a linear transformation ✓

Strengths, Limitations & Common Pitfalls

Linear transformations are incredibly useful, but they also have boundaries. Understanding both their strengths and their limitations helps you know when and how to apply them.

Strengths vs. limitations of linear transformations
StrengthsLimitations
Can be represented by matrices, making calculations efficient even for huge data sets.Cannot model curved relationships like parabolas (y = x²) or exponential growth.
Predictable behavior: scaling the input scales the output by the same factor.Cannot include shifts or translations — T(0) must always equal 0.
Can be composed (chained together) and inverted, making them flexible building blocks.Real-world phenomena are often non-linear; linear transformations only approximate small portions.
Foundation for computer graphics: rotations, reflections, and scaling are all linear.Perspective projections (like 3-D to 2-D in video games) require an extension called affine transformations.
COMMON PITFALL
Students often confuse "linear function" (like y = 3x + 2) with "linear transformation." The function y = 3x + 2 is not a linear transformation because of the "+2" — it shifts the origin. A true linear transformation through ℝ¹ would be y = 3x (no added constant). Think of it this way: a linear transformation is like a perfectly fair recipe — double the ingredients, double the result, with no extra seasoning thrown in.

Connection to Advanced Topics

The definition of a linear transformation is the starting point for a huge web of ideas in mathematics and science. Once you know what a linear transformation is, you can explore how to measure it, classify it, and use it in sophisticated ways.

From this lesson to advanced linear algebra
This LessonWhere It Leads
Definition: T(u + v) = T(u) + T(v) and T(cv) = cT(v)Matrix representation: Every linear transformation from ℝⁿ to ℝᵐ can be written as multiplication by a unique matrix.
The zero vector maps to the zero vector.Kernel (null space): The set of all vectors that T sends to zero reveals important structural information.
Additivity and homogeneity as separate checks.Eigenvalues and eigenvectors: Special vectors whose direction is unchanged by T, used in vibration analysis, Google's PageRank, and quantum mechanics.
Geometric intuition: grid lines stay straight and evenly spaced.Singular Value Decomposition (SVD): Breaking any linear transformation into a rotation, a stretch, and another rotation — used in image compression and data science.

As you continue studying linear algebra, remember that the two simple rules — additivity and homogeneity — are the foundation for everything that comes next. Master them, and the rest of linear algebra will feel like natural extensions of the same core idea.

Practice Problems

PROBLEM 1CONCEPTUAL
State the two properties that a function T must satisfy to be called a linear transformation. Then explain in your own words why T(0) = 0 must be true for any linear transformation.
PROBLEM 2BASIC CALCULATION
Let T(x, y) = (4x, −2y). Verify that T is a linear transformation by checking additivity and homogeneity with u = (1, 3), v = (2, −1), and c = 5.
PROBLEM 3INTERMEDIATE
Determine whether T(x, y) = (x + y, x − y + 3) is a linear transformation. Justify your answer by testing the definition.
PROBLEM 4APPLIED
A computer graphics program rotates every point (x, y) on the screen by 90° counterclockwise using T(x, y) = (−y, x). A game designer wants to rotate the triangle with vertices A = (2, 0), B = (2, 3), and C = (5, 0). Find the new vertices and explain why T being a linear transformation guarantees the rotated triangle keeps its shape.
PROBLEM 5CRITICAL THINKING
Suppose T is a linear transformation from ℝ² to ℝ² and you know that T(1, 0) = (3, 1) and T(0, 1) = (−2, 4). Using only the definition of linearity, find T(5, −3). Then explain why knowing what T does to (1, 0) and (0, 1) is enough to determine T for every vector in ℝ².

Lesson Summary

A linear transformation is a function T between vector spaces that satisfies two rules: additivity — T(u + v) = T(u) + T(v) — and homogeneity — T(c · v) = c · T(v). These two properties can be combined into the single superposition condition: T(c₁u + c₂v) = c₁T(u) + c₂T(v). A key consequence is that T(0) = 0 always, and straight, evenly spaced grid lines remain straight and evenly spaced after the transformation.

Linear transformations from ℝⁿ to ℝᵐ can be represented by matrix multiplication, and they are completely determined by their action on a set of basis vectors. Common examples include rotations, reflections, scaling, and projections. Functions that add constants, square their inputs, or use absolute values are not linear. Mastering this definition opens the door to matrices, eigenvalues, and the many powerful tools of linear algebra.

Varsity Tutors • Linear Algebra • Linear Transformation Definition