Angles and Orthogonality

Angles describe how two directions relate.

The dot product connects algebra to angle:

u \cdot v = \|u\| \|v\| \cos(\theta)

If two nonzero vectors are perpendicular, the angle is 90 degrees, so cos(theta) = 0.

That means:

u \cdot v = 0

Perpendicular directions have zero dot product.

Orthogonality

Two nonzero vectors are orthogonal when their dot product is zero.

For example:

u = [2, -1]
v = [1, 2]

Then:

u \cdot v = 2\cdot1 + (-1)\cdot2 = 0

So $u$ and $v$ are orthogonal.

MATH-C04-T04-001Exercise: Check orthogonality

Let u = [2, -1] and v = [1, 2].

Compute u . v.

Check the dot product

Compute it first, then check your number.

HintProducts then sum

Multiply matching entries, then add.

SolutionZero dot product

Multiply matching entries and add:

u \cdot v = 2(1) + (-1)(2) = 2 - 2 = 0

The result is zero, so these nonzero vectors are orthogonal under the dot product.

Why Orthogonality Matters

Orthogonal directions do not overlap in the dot-product sense.

This matters when we want features, components, or directions to carry different information. Orthogonal axes are also easier to reason about because movement along one axis does not change movement along another.

Orthogonal does not mean unrelated in every possible sense. It means zero linear overlap under the dot product being used. Two real features can still interact later in a model, even if their current vectors are orthogonal.

MATH-C04-T04-002Exercise: Recognize zero overlap

Enter 1 if orthogonal directions have zero dot product.

Your answer

Compute it first, then check your number.

HintDot-product test

Orthogonality is checked with the dot product.

SolutionOrthogonality test

Enter 1. For nonzero vectors, orthogonality means the dot product is zero. In this course, that is the algebraic test for "no overlap" in the chosen dot-product geometry.

MATH-C04-T04-004Exercise: Avoid overclaiming orthogonality

Enter 1 if orthogonality means zero dot-product overlap, not a guarantee that two real-world features can never interact.

Your answer

Compute it first, then check your number.

HintMathematical claim

Keep the claim tied to the dot product.

SolutionPrecise interpretation

Enter 1. Orthogonality means zero overlap under the dot product. It does not prove that the represented concepts can never interact elsewhere.

Angle Versus Distance

Distance asks how far apart two points are.

Angle asks how similar two directions are.

In embedding spaces, both can be useful. Cosine similarity is often used when direction matters more than magnitude.

MATH-C04-T04-003Exercise: Choose angle or distance

Enter 1 if cosine similarity is mainly about direction, or 2 if it is mainly about point-to-point distance.

Your answer

Compute it first, then check your number.

HintCosine uses angle

Cosine similarity comes from the angle formula.

SolutionDirection comparison

Enter 1. Cosine similarity compares direction after accounting for vector lengths. Distance asks how far two points are from each other; cosine asks how aligned two nonzero directions are.

Before Moving On

Orthogonality is zero overlap under the dot product. It is one of the main ways geometry becomes computation.