The Dot Product: Multiply, Add, Done

What Is a Dot Product?

You have two lists of numbers. Same length. You multiply them in pairs, then add everything up. That's it. That's the whole thing.

$$\vec{a} \cdot \vec{b} = a_1 b_1 + a_2 b_2 + \cdots + a_n b_n$$

Two lists in, one number out. Always.

Your First Dot Product

Let's say you have these two lists (mathematicians call them "vectors"):

$$\vec{a} = \begin{bmatrix} 2 \\ 3 \end{bmatrix} \qquad \vec{b} = \begin{bmatrix} 4 \\ 5 \end{bmatrix}$$

Step 1: Multiply the first elements together:

$$2 \times 4 = 8$$

Step 2: Multiply the second elements together:

$$3 \times 5 = 15$$

Step 3: Add them up:

$$8 + 15 = \mathbf{23}$$

Done. The dot product of $\vec{a}$ and $\vec{b}$ is 23.

$$\begin{bmatrix} 2 \\ 3 \end{bmatrix} \cdot \begin{bmatrix} 4 \\ 5 \end{bmatrix} = (2 \times 4) + (3 \times 5) = 8 + 15 = 23$$

Now With 3 Numbers

Works the exact same way with longer lists. Here's a 3D dot product:

$$\begin{bmatrix} 1 \\ 3 \\ 5 \end{bmatrix} \cdot \begin{bmatrix} 2 \\ 4 \\ 6 \end{bmatrix} = (1 \times 2) + (3 \times 4) + (5 \times 6) = 2 + 12 + 30 = \mathbf{44}$$

100 numbers? Same thing. A million numbers? Same thing. Multiply pairs, add them all up.

The Angle Trick

Here's where it gets cool. The dot product secretly tells you the angle between two vectors:

$$\vec{a} \cdot \vec{b} = |\vec{a}| \cdot |\vec{b}| \cdot \cos(\theta)$$

You don't need to memorize that. Just remember what the result means:

Positive number: the two vectors point in roughly the same direction
Zero: they're perpendicular (90 degrees apart)
Negative number: they point in roughly opposite directions

This is the single most useful property of the dot product. Almost every real-world use comes from this.

Real-World Uses (Not Just Math Class)

3D Graphics: Lighting

Every lit pixel in every video game uses a dot product. When light hits a surface, the brightness depends on the angle between the light direction and the surface normal (the direction the surface faces).

$$\text{brightness} = \vec{L} \cdot \vec{N} = |\vec{L}||\vec{N}|\cos(\theta)$$

Light hitting the surface head-on? $\cos(0°) = 1$, full brightness. Light at a steep angle? $\cos(80°) \approx 0.17$, barely lit. Light behind the surface? Dot product is negative, no light at all.

This is called Lambertian shading and it runs billions of times per frame on your GPU. Every single one is a dot product.

Recommendation Engines: "How Similar Are You?"

Netflix, Spotify, Amazon: they all represent you as a list of numbers. Maybe your "taste vector" for movies is:

$$\text{you} = \begin{bmatrix} 0.9 \\ 0.1 \\ 0.7 \\ 0.2 \end{bmatrix} \quad \leftarrow \text{ action, romance, sci-fi, horror}$$

And a movie's "vibe vector" is:

$$\text{Interstellar} = \begin{bmatrix} 0.6 \\ 0.3 \\ 0.95 \\ 0.05 \end{bmatrix}$$

Dot product them:

$$(0.9 \times 0.6) + (0.1 \times 0.3) + (0.7 \times 0.95) + (0.2 \times 0.05)$$

$$= 0.54 + 0.03 + 0.665 + 0.01 = \mathbf{1.245}$$

High number = good match. The algorithm does this for every movie in the catalog and shows you the highest scores. That's literally it.

Search Engines: "How Relevant Is This Page?"

Google (and every search engine) turns your query and every web page into vectors of numbers. The dot product (or its normalized cousin, cosine similarity) tells them how relevant a page is to your search.

$$\text{cosine similarity} = \frac{\vec{a} \cdot \vec{b}}{|\vec{a}| \cdot |\vec{b}|}$$

This gives a score between $-1$ and $1$. Closer to 1 means the page is about what you searched for. It's a dot product with the lengths divided out so long documents don't automatically win.

Physics: Work Done by a Force

When you push a box across the floor at an angle, not all your force moves the box forward. Only the component in the direction of motion counts. That's a dot product.

$$W = \vec{F} \cdot \vec{d} = |\vec{F}||\vec{d}|\cos(\theta)$$

Push straight ($\theta = 0°$): all your force does work. Push at 90° (straight down into the floor): zero work done. The box doesn't move. The dot product is zero.

Audio: Noise Cancellation

Your AirPods record outside noise as a list of numbers (a signal vector). They generate an "anti-noise" signal. If the dot product of the noise and anti-noise is close to zero, they've cancelled each other out. Silence.

Game Dev: "Can the Enemy See Me?"

In stealth games, the AI checks if the player is in the enemy's field of view. Take the enemy's look direction and the direction to the player. Dot product them.

Result close to 1: player is right in front. Spotted.
Result close to 0: player is off to the side. Maybe peripheral vision.
Result negative: player is behind the enemy. Safe.

One dot product replaces a bunch of messy angle math.

The Pattern

Every single one of these uses boils down to the same question: "how much do these two lists of numbers point in the same direction?"

That's the dot product. Multiply pairs, add them up, get an answer. The simplest useful operation in all of math.