Geometric Applications - Linear Algebra

SECTION 1

Historical Context & Motivation

Long before computers could render 3-D video games or animate movie characters, mathematicians were searching for a clean, powerful way to describe shapes and their movements. The ancient Greeks studied geometry with rulers and compasses, but they lacked a system that could handle many points at once. Over centuries, new tools emerged — coordinates, vectors, and matrices — that turned geometry into something you could calculate step by step.

~300 BC • Euclid's Elements

Euclid organized Greek geometry into logical proofs. His work described shapes and transformations using words and diagrams, but had no algebraic formulas.

1637 • Descartes Invents Coordinate Geometry

René Descartes merged algebra and geometry by placing shapes on an x-y grid. Every point became a pair of numbers, opening the door to calculations.

1843 • Hamilton's Quaternions

William Rowan Hamilton extended the idea of complex numbers to describe rotations in 3-D space — an early step toward modern computer graphics.

1850s • Cayley & Sylvester Formalize Matrices

Arthur Cayley and James Sylvester developed matrix algebra, giving mathematicians a compact way to apply geometric transformations to many points simultaneously.

1970s–Today • Computer Graphics Revolution

Video games, animated films, and CAD software all rely on matrix transformations to move, rotate, and scale objects in real time on your screen.

The central question this lesson addresses is: How can we use vectors and matrices to move, resize, and reshape geometric figures in a precise, repeatable way? Once you see the pattern, you'll realize that every animation frame and every GPS map on your phone relies on these same ideas.

SECTION 2

Core Principles & Definitions

Before we start transforming shapes, let's nail down the key ideas you'll use throughout the lesson. Each concept builds on the one before it, so take a moment with each card below.

Vectors as Points

A vector is an ordered list of numbers like (3, 5). In geometry, each vector represents a point on the coordinate plane or a direction and distance from the origin.

Transformation

A transformation is any rule that takes every point of a shape and moves it to a new position. Slides, flips, turns, and resizing are all transformations.

Matrix Multiplication

A matrix is a rectangular grid of numbers. Multiplying a 2 × 2 matrix by a point's vector gives the transformed point. One matrix can encode an entire transformation.

Linear vs. Affine

Rotations, reflections, and scaling are linear transformations — they keep the origin fixed. Translation (sliding) is an affine transformation because it shifts the whole figure, origin included.

Composition

You can chain transformations by multiplying their matrices together. This is called composition. One combined matrix does the same work as applying several transformations in sequence.

✦

KEY TAKEAWAY

Think of a matrix like a recipe card for a shape. Just as a recipe tells you exactly how to change ingredients into a finished dish, a matrix tells you exactly how to change every point of a shape into its new position. One recipe card (matrix) can be reused on any shape, and you can stack recipe cards (multiply matrices) to combine effects.

SECTION 3

Visual Explanation — Transformations on a Grid

The diagram below shows the four main geometric transformations applied to the same triangle. Study how each vertex moves; the grid lines help you count units.

Each panel shows the same purple triangle before (original) and after one transformation. Translation shifts every point the same distance. Reflection mirrors across a line. Rotation turns around a center point. Scaling enlarges or shrinks from the origin.

Notice that in every case, the shape itself doesn't change its basic form — angles stay the same in translation, reflection, and rotation (these are called rigid transformations). Scaling changes the size but keeps the same proportions. The power of linear algebra is that each of these transformations can be expressed as a simple matrix multiplication.

SECTION 4

Mathematical Framework

Let's turn the visual ideas into algebra. Every point in 2-D can be written as a column vector. A 2 × 2 matrix multiplied by that vector produces the transformed point. Here are the key matrices.

SCALING MATRIX

S = [ sₓ 0 ] [ 0 sᵧ ]

sₓ scales in the x-direction; sᵧ scales in the y-direction. When sₓ = sᵧ, the shape grows or shrinks uniformly.

ROTATION MATRIX (θ COUNTERCLOCKWISE)

R(θ) = [ cos θ −sin θ ] [ sin θ cos θ ]

θ is the angle of rotation measured counterclockwise from the positive x-axis. For a 90° turn, cos 90° = 0 and sin 90° = 1.

REFLECTION OVER THE Y-AXIS

M_y = [ −1 0 ] [ 0 1 ]

This flips the x-coordinate's sign while keeping y the same. To reflect over the x-axis instead, swap the −1 and 1 on the diagonal.

TRANSLATION (VECTOR ADDITION)

P' = P + t, where t = [ tₓ ] [ tᵧ ]

Translation is not a matrix multiplication in 2 × 2 form; it is vector addition. tₓ shifts horizontally, tᵧ shifts vertically. You can fold translation into matrix form using 3 × 3 homogeneous coordinates, which advanced courses cover.

💡

How Matrix Multiplication Works (Quick Refresher)

To multiply a 2 × 2 matrix by a vector, take each row of the matrix and compute the dot product with the vector. For example: [ a b ] · [ x ] = a × x + b × y. The result goes into the corresponding row of the new vector.

SECTION 5

Detailed Breakdown — Composition of Transformations

In practice, you rarely apply just one transformation. A video game character might need to be rotated, scaled, and then translated — all in a single frame update. The beauty of matrices is that you can multiply several transformation matrices together into one combined matrix and apply it in a single step.

A unit square is first scaled by 2, then rotated 45°, then translated to the right and up. Each step's matrix (S, R) and vector (+t) combine to produce the final position in one operation.

Order matters! Rotating first and then scaling can give a different result than scaling first and then rotating. In matrix terms, A × B is generally not the same as B × A. Always apply transformations from right to left when reading the product: the matrix closest to the vector acts first.

Transformation	Matrix / Operation	Preserves Shape?
Translation	P' = P + t (vector addition)	Yes — rigid motion
Reflection	2 × 2 matrix with −1 on one diagonal	Yes — rigid motion (flipped orientation)
Rotation	2 × 2 matrix with cos θ, sin θ	Yes — rigid motion
Scaling	Diagonal matrix [sₓ, sᵧ]	Proportions yes, size no
Shear	Off-diagonal entry ≠ 0	No — angles change

Summary of 2-D transformations and their properties

SECTION 6

Worked Example — Rotating a Triangle

Let's apply what we've learned. Suppose triangle ABC has vertices A = (1, 0), B = (3, 0), and C = (2, 3). Rotate the triangle 90° counterclockwise about the origin.

Rotating Triangle ABC by 90° CCW

1
Step 1 — Write the Rotation Matrix

For θ = 90°, cos 90° = 0 and sin 90° = 1. The rotation matrix becomes R = [ 0 −1 ] / [ 1 0 ].
2
Step 2 — Transform Vertex A = (1, 0)

Multiply R by the vector (1, 0): new x = 0 × 1 + (−1) × 0 = 0; new y = 1 × 1 + 0 × 0 = 1.

A' = (0, 1)
3
Step 3 — Transform Vertex B = (3, 0)

New x = 0 × 3 + (−1) × 0 = 0; new y = 1 × 3 + 0 × 0 = 3.

B' = (0, 3)
4
Step 4 — Transform Vertex C = (2, 3)

New x = 0 × 2 + (−1) × 3 = −3; new y = 1 × 2 + 0 × 3 = 2.

C' = (−3, 2)
5
Step 5 — Verify with a Sketch

Plot the new vertices A'(0,1), B'(0,3), C'(−3,2). The triangle has the same side lengths and angles as the original — it has simply turned 90° counterclockwise around the origin.

Triangle A'B'C' = (0,1), (0,3), (−3,2)

SECTION 7

Strengths & Limitations of Matrix Transformations

Strengths	Limitations
One matrix encodes an entire transformation — efficient for computers to process millions of points.	2 × 2 matrices cannot represent translation; you need 3 × 3 homogeneous coordinates.
Matrices can be composed (multiplied) so many steps become one step.	Matrix multiplication is not commutative (order matters), which can be confusing at first.
Works the same way in 2-D and 3-D (just use 3 × 3 or 4 × 4 matrices).	Non-linear transformations (like bending or warping) require more advanced techniques.
Inverse matrices let you undo a transformation easily.	Numerical rounding errors can accumulate when many matrices are multiplied together.

Pros and cons of using matrices for geometric transformations

✦

KEY TAKEAWAY

Matrix transformations are like a universal remote control for geometry. One remote (matrix) can resize, rotate, or flip any shape you point it at. The limitation is that the remote only works for straight-line moves — if you need to bend or squish something, you'll need fancier tools from calculus and differential geometry.

SECTION 8

Connection to Advanced Theory — 3-D & Beyond

Everything you've learned about 2 × 2 matrices extends naturally to three dimensions. In 3-D graphics, a 4 × 4 matrix (using homogeneous coordinates) can handle rotation, scaling, reflection, and translation all in one multiplication. This is the foundation behind every 3-D video game engine and animation studio pipeline.

Concept	What You Learned (2-D)	Advanced Version (3-D+)
Rotation	2 × 2 matrix with sin/cos	3 × 3 matrix (rotate around x, y, or z axis) or quaternions
Translation	Vector addition	Embedded in 4 × 4 homogeneous matrix
Composition	Multiply two 2 × 2 matrices	Multiply chains of 4 × 4 matrices (model-view-projection pipeline)
Inverse	2 × 2 inverse undoes transformation	Used for camera transforms & physics simulations

From this lesson to advanced computer graphics

If you continue into courses on computer graphics, robotics, or physics simulations, you'll also encounter eigenvalues and eigenvectors. These tell you which directions a transformation stretches and by how much — a powerful concept that builds directly on the matrix ideas you practiced today.

SECTION 9

Practice Problems

Question 1: CONCEPTUAL

Which of the four main transformations (translation, reflection, rotation, scaling) changes the size of a figure but keeps all angles the same? Explain why.

Show answer

Scaling changes the size of a figure while preserving all of its angles. This happens because the scaling matrix multiplies every coordinate by the same factor (or proportional factors for each axis), so the ratios between sides remain constant. Angles depend on those ratios, not on absolute lengths, so they stay unchanged. Translation, reflection, and rotation are all rigid motions — they don't alter size or angles at all.

Question 2: BASIC CALCULATION

A point P = (4, −2) is scaled by sₓ = 3 and sᵧ = 3. Find the coordinates of the image P'.

Show answer

Multiply the scaling matrix by the vector: new x = 3 × 4 = 12; new y = 3 × (−2) = −6. So P' = (12, −6). The point moves three times farther from the origin in every direction, which makes sense because both scale factors equal 3.

Question 3: INTERMEDIATE

Reflect the point Q = (5, 3) over the y-axis, then rotate the result 90° counterclockwise. What are the final coordinates?

Show answer

First, reflect over the y-axis using the matrix [ −1 0 ] / [ 0 1 ]. This gives Q₁ = (−5, 3). Next, apply the 90° CCW rotation matrix [ 0 −1 ] / [ 1 0 ] to Q₁: new x = 0 × (−5) + (−1) × 3 = −3; new y = 1 × (−5) + 0 × 3 = −5. The final coordinates are Q' = (−3, −5). Notice how performing two transformations in sequence requires careful attention to order — reflecting first and then rotating produces a different result than the reverse order.

Question 4: APPLIED

A game designer has a spaceship sprite with a tip at (0, 5), a left wing at (−3, 0), and a right wing at (3, 0). She wants the ship to face right instead of up, which requires a 90° clockwise rotation. Find the three new vertex positions.

Show answer

A 90° clockwise rotation is the same as a 270° counterclockwise rotation. The matrix is [ 0 1 ] / [ −1 0 ]. Apply it to each vertex: tip (0, 5) → (5, 0); left wing (−3, 0) → (0, 3); right wing (3, 0) → (0, −3). The ship now points to the right along the positive x-axis. The tip is at (5, 0), the upper fin at (0, 3), and the lower fin at (0, −3). This is exactly how game engines re-orient sprites — they multiply every vertex by a single rotation matrix.

Question 5: CRITICAL THINKING

Prove that if you apply a 2 × 2 rotation matrix R(θ) twice — computing R(θ) × R(θ) — the result equals R(2θ). Hint: use the double-angle identities cos 2θ = cos²θ − sin²θ and sin 2θ = 2 sin θ cos θ.

Show answer

R(θ) × R(θ) multiplies [ cos θ −sin θ ] / [ sin θ cos θ ] by itself. The (1,1) entry is cos θ × cos θ + (−sin θ) × sin θ = cos²θ − sin²θ = cos 2θ. The (1,2) entry is cos θ × (−sin θ) + (−sin θ) × cos θ = −2 sin θ cos θ = −sin 2θ. The (2,1) entry is sin θ × cos θ + cos θ × sin θ = 2 sin θ cos θ = sin 2θ. The (2,2) entry is sin θ × (−sin θ) + cos θ × cos θ = cos²θ − sin²θ = cos 2θ. So the product matrix is [ cos 2θ −sin 2θ ] / [ sin 2θ cos 2θ ], which is exactly R(2θ). This confirms that rotating by θ twice is the same as rotating by 2θ — composition of rotations adds angles.

SUMMARY

Lesson Summary

Summary

In this lesson you learned that vectors represent points on the coordinate plane and that matrices encode geometric transformations. The four core transformations are translation (sliding via vector addition), reflection (flipping across a line), rotation (turning by an angle using sin and cos), and scaling (resizing by a factor). Each can be applied by a single matrix multiplication, and multiple transformations can be composed into one combined matrix.

Remember that order matters when composing transformations — rotating then scaling is not the same as scaling then rotating. These 2-D ideas extend naturally to 3-D computer graphics using larger matrices. Whether you're programming a game, designing a building, or analyzing data, the geometric power of linear algebra gives you precise, repeatable control over shapes and space.