Linear Transformation Theory- Mathematical Foundations

What Is a Linear Transformation?

A linear transformation is a function that maps vectors from one vector space to another while preserving two fundamental operations: vector addition and scalar multiplication. In plain terms, it bends, stretches, compresses, or rotates vectors without tearing or gluing them together.

The formal definition: A function T: V → W is linear if for all vectors u, v in V and all scalars c:

• T(u + v) = T(u) + T(v)
• T(cu) = cT(u)

If you see something that doesn't satisfy these conditions, it's not linear. Period.

Core Properties Every Mathematician Should Know

Linear transformations have predictable behaviors that make them useful in computation and theory.

Preservation of the Zero Vector

T(0) = 0. This follows directly from linearity. If your transformation maps the zero vector somewhere else, it's not linear.

Preservation of Linear Combinations

T(au + bv) = aT(u) + bT(v). This property extends to any finite combination of vectors. It's the backbone of matrix representation.

The Image of the Origin Stays the Origin

Geometrically, linear transformations always pass through the origin. You cannot translate vectors with a linear transformation. If you need translation, you're looking at affine transformations, which are a different beast.

Matrix Representation—The Practical View

Every linear transformation from ℝⁿ to ℝᵐ can be represented by an m×n matrix. The transformation T(x) = Ax gives you the output vector.

How do you find A? Apply T to each standard basis vector of ℝⁿ. The resulting columns form A:

• T(e₁) produces column 1
• T(e₂) produces column 2
• And so on

This is the standard way to construct the matrix of a linear transformation. No shortcuts, no tricks.

Types of Linear Transformations in ℝ² and ℝ³

Here is where geometry meets algebra. These are the transformations you'll encounter most often:

The determinant tells you how the transformation affects area (in 2D) or volume (in 3D). Zero determinant means the transformation collapses dimensions—information is lost.

Kernel and Range—What Gets Killed and What Survives

Kernel (Null Space)

The kernel of T is the set of all vectors that T maps to the zero vector. Mathematically: ker(T) = {v ∈ V : T(v) = 0}.

Think of it as the input that gets annihilated. If ker(T) contains only the zero vector, the transformation is injective (one-to-one).

Range (Image)

The range is the set of all output vectors. range(T) = {T(v) : v ∈ V}.

It's a subspace of the codomain. If range(T) equals the entire codomain, T is surjective (onto).

The Rank-Nullity Theorem

For a linear transformation T: V → W where V is finite-dimensional:

dim(ker(T)) + dim(range(T)) = dim(V)

This relationship is non-negotiable. It tells you exactly how much information survives the transformation.

Composition of Linear Transformations

When you chain transformations together, the result is still linear. If T₁: V → W and T₂: W → U, then T₂ ∘ T₁: V → U is linear.

Matrix-wise, the composition corresponds to matrix multiplication. If T₁(x) = Ax and T₂(y) = By, then T₂(T₁(x)) = B(Ax) = (BA)x.

Remember: matrix multiplication is not commutative. BA ≠ AB in general. The order matters.

Change of Basis—Same Transformation, Different Matrix

The matrix of a linear transformation depends on your choice of basis. Pick a different basis, get a different matrix. The transformation itself doesn't change—only how you represent it.

If P is the change-of-basis matrix from basis B to basis C, then the new matrix representation is:

A' = P⁻¹AP

Matrices A and A' are similar. They share the same determinant, trace, and eigenvalues. The algebraic properties stay constant; only the numerical representation changes.

Getting Started: How to Work With Linear Transformations

Here is the practical workflow for analyzing a linear transformation:

Step 1: Verify Linearity

Check the two conditions: T(u + v) = T(u) + T(v) and T(cu) = cT(u). Test with simple vectors if needed.

Step 2: Find the Matrix Representation

Apply T to each standard basis vector. Record the results as columns. That's your matrix A.

Step 3: Compute the Kernel

Solve Ax = 0. Use row reduction on the augmented matrix if necessary. The solution set is ker(T).

Step 4: Compute the Range

Find the column space of A. The pivot columns in row-reduced echelon form tell you which columns are linearly independent. That's your basis for range(T).

Step 5: Check Invertibility

If det(A) ≠ 0, the transformation is invertible. If det(A) = 0, it's not injective, and information is lost.

Where Linear Transformations Appear in the Real World

Computer graphics uses rotation, scaling, and projection matrices to render 3D scenes onto 2D screens. Every time you rotate an image or scale a shape in software, you're applying linear transformations.

Quantum mechanics describes quantum states as vectors and observables as linear operators. The math isn't optional here—it's the entire framework.

Data science uses linear transformations for dimensionality reduction. Principal Component Analysis applies linear transformations to find the directions of maximum variance in data.

Engineering systems are modeled with linear differential equations, which are solved using linear transformations to frequency domains.

The theory isn't abstract for the sake of being abstract. It underpins practical computation across disciplines.

Eigenvalues and Eigenvectors—The Invariants

An eigenvector of T satisfies T(v) = λv. The vector doesn't change direction—it only gets scaled by factor λ, the eigenvalue.

To find eigenvalues, solve det(A - λI) = 0. This characteristic polynomial gives you all eigenvalues directly.

Why does this matter? Eigenvectors define the axes along which a transformation acts as simple scaling. In many applications, switching to the eigenbasis diagonalizes the matrix, making computations trivial.