Dot Product

A farmer walks 5 km east carrying a heavy sack, then turns and walks 3 km northeast. At the end, she has covered 8 km of ground — but she is not 8 km from where she started. The total displacement depends on the angle between the two legs of her journey. And the amount of work she does against gravity while climbing a slope depends on the angle between her force and the direction of motion.

Angles between directions keep showing up. But vectors, as you have seen them so far, are arrows with magnitude and direction — you can add them, subtract them, scale them. None of those operations produce a single number that captures the relationship between two directions.

The dot product fills that gap. It takes two vectors and returns one number — a scalar — that encodes the angle between them. When the vectors point in the same direction, this number is large and positive. When they are perpendicular, it is exactly zero. When they point in opposite directions, it is large and negative. This single operation unlocks angle measurement, projection, and the notion of perpendicularity in any number of dimensions.

Building the idea from components

Suppose you have two vectors in the plane:

You already know how to add them and scale them. But what if you multiply them component by component and add the results?

This is just a number — not a vector. At first glance, it looks arbitrary. Why multiply matching components? Why add? The answer is that this particular combination has a geometric meaning that no other combination has: it is directly related to the angle between the two vectors.

Take a specific case. Let \vec{a} = 3\hat{i} + 4\hat{j} and \vec{b} = 4\hat{i} + 3\hat{j}. Both have magnitude 5 (check: \sqrt{9 + 16} = 5). The component-wise product-and-sum gives 3 \times 4 + 4 \times 3 = 24.

Now take \vec{b} = 0\hat{i} + 5\hat{j} instead — same magnitude, but pointing straight up. The product-and-sum gives 3 \times 0 + 4 \times 5 = 20. A smaller number, because the two vectors are further apart in direction.

And if \vec{b} = -4\hat{i} - 3\hat{j} — pointing in the opposite direction to the original — the product-and-sum gives 3 \times (-4) + 4 \times (-3) = -24. Negative, because the vectors oppose each other.

The product-and-sum is tracking the angle. This operation has a name.

The definition

There are two definitions here and they must agree. That the geometric form (|\vec{a}||\vec{b}|\cos\theta) equals the component form (a_1b_1 + a_2b_2 + a_3b_3) is not obvious — it requires a proof. The proof below uses the cosine rule from trigonometry, connecting the two forms cleanly.

Why the two definitions are equivalent

Place \vec{a} and \vec{b} tail-to-tail. The vector from the tip of \vec{b} to the tip of \vec{a} is \vec{a} - \vec{b}. These three vectors form a triangle, and the cosine rule gives:

Now expand the left side using components. If \vec{a} - \vec{b} = (a_1 - b_1)\hat{i} + (a_2 - b_2)\hat{j} + (a_3 - b_3)\hat{k}, then:

Compare this with the cosine rule expansion:

Cancel |\vec{a}|^2 + |\vec{b}|^2 from both sides:

The two definitions are the same thing.

Properties of the dot product

The dot product obeys several clean algebraic rules. Each one is worth knowing and worth proving.

1. Commutative: \vec{a} \cdot \vec{b} = \vec{b} \cdot \vec{a}

From the component form: a_1b_1 + a_2b_2 + a_3b_3 = b_1a_1 + b_2a_2 + b_3a_3, since multiplication of real numbers is commutative. Alternatively, from the geometric form: |\vec{a}||\vec{b}|\cos\theta = |\vec{b}||\vec{a}|\cos\theta.

2. Distributive over addition: \vec{a} \cdot (\vec{b} + \vec{c}) = \vec{a} \cdot \vec{b} + \vec{a} \cdot \vec{c}

Write \vec{b} + \vec{c} = (b_1 + c_1)\hat{i} + (b_2 + c_2)\hat{j} + (b_3 + c_3)\hat{k}. Then:

This is just the distributive law of real numbers, applied component by component.

3. Scalar multiplication: (k\vec{a}) \cdot \vec{b} = k(\vec{a} \cdot \vec{b})

(k\vec{a}) \cdot \vec{b} = (ka_1)b_1 + (ka_2)b_2 + (ka_3)b_3 = k(a_1b_1 + a_2b_2 + a_3b_3) = k(\vec{a} \cdot \vec{b}).

4. Dot product with itself: \vec{a} \cdot \vec{a} = |\vec{a}|^2

\vec{a} \cdot \vec{a} = a_1^2 + a_2^2 + a_3^2 = |\vec{a}|^2.

Geometrically: |\vec{a}||\vec{a}|\cos 0 = |\vec{a}|^2, since the angle between a vector and itself is 0, and \cos 0 = 1.

This property is especially useful — it lets you compute the magnitude of a vector as |\vec{a}| = \sqrt{\vec{a} \cdot \vec{a}}.

5. Dot product of perpendicular vectors: if \vec{a} \perp \vec{b}, then \vec{a} \cdot \vec{b} = 0

When \theta = \pi/2, \cos\theta = 0, so |\vec{a}||\vec{b}|\cos(\pi/2) = 0. The converse also holds (provided neither vector is zero): if \vec{a} \cdot \vec{b} = 0 and both vectors are nonzero, then \cos\theta = 0, so \theta = \pi/2 and the vectors are perpendicular.

6. Dot products of the standard unit vectors:

These follow directly from properties 4 and 5: each unit vector has magnitude 1, and any two distinct standard unit vectors are perpendicular.

Angle between two vectors

The geometric definition immediately gives you a formula for the angle between any two vectors. Rearranging \vec{a} \cdot \vec{b} = |\vec{a}||\vec{b}|\cos\theta:

This is remarkable. Before the dot product, if someone handed you two vectors in component form — say \vec{a} = 2\hat{i} + 3\hat{j} + \hat{k} and \vec{b} = \hat{i} - \hat{j} + 4\hat{k} — you had no clean way to find the angle between them. Now you do: compute the dot product, divide by the product of magnitudes, and take the inverse cosine. Geometry reduced to arithmetic.

The formula also tells you the qualitative story at a glance, without computing the angle itself:

• If \vec{a} \cdot \vec{b} > 0, the angle is acute (0 < \theta < \pi/2) — the vectors point in broadly the same direction.
• If \vec{a} \cdot \vec{b} = 0, the angle is exactly \pi/2 — the vectors are perpendicular.
• If \vec{a} \cdot \vec{b} < 0, the angle is obtuse (\pi/2 < \theta < \pi) — the vectors point in broadly opposite directions.

The sign of the dot product is a quick diagnostic. You can classify the angle as acute, right, or obtuse before ever touching a calculator.

Projection of one vector onto another

Here is the second major use of the dot product: measuring how much one vector lies along the direction of another.

Stand in sunlight with a stick held at an angle. The shadow the stick casts on the ground is shorter than the stick itself. The length of the shadow depends on the angle between the stick and the ground. That shadow-length is a projection.

The scalar projection (or component) of \vec{a} along \vec{b} is the signed length of this shadow:

The sign matters: if the projection is positive, \vec{a} has a component in the direction of \vec{b}. If negative, \vec{a} has a component in the opposite direction.

The vector projection of \vec{a} onto \vec{b} is not just the length but the actual vector along \vec{b}:

This formula says: take the scalar projection, divide by |\vec{b}| to get a scale factor, then multiply by \vec{b} itself. The result is a vector pointing along \vec{b} with the right length.

Why does this matter? Because projection decomposes \vec{a} into two perpendicular pieces:

The first piece lies along \vec{b}. The second piece — the remainder — is perpendicular to \vec{b}. Every vector can be split this way relative to any direction. This decomposition is the reason the dot product matters in physics: resolving forces into components along a slope and perpendicular to it, or splitting a velocity into radial and tangential parts, all use this projection.

Computing the dot product — first example

Projection — second example

Common confusions

• "The dot product is a vector." It is not. The dot product of two vectors is a scalar — a single number. That is why it is called the scalar product. The operation that produces a vector from two vectors is the cross product, which is a different operation entirely.

• "If \vec{a} \cdot \vec{b} = 0, one of the vectors is zero." Not necessarily. It could mean the vectors are perpendicular. \vec{a} \cdot \vec{b} = 0 has two possible explanations: either at least one vector is \vec{0}, or the vectors are at a right angle to each other. The perpendicularity case is far more common and far more useful.

• "The angle \theta can be any angle from 0 to 2\pi." The angle in the dot product formula is always between 0 and \pi (inclusive). There is no notion of "clockwise versus anticlockwise" here — the dot product treats the two vectors symmetrically and gives the smaller of the two possible angles.

• "\vec{a} \cdot \vec{b} \cdot \vec{c} makes sense." It does not. \vec{a} \cdot \vec{b} is a scalar, and you cannot take the dot product of a scalar with a vector. The dot product is defined only between two vectors. If you see three vectors multiplied, you are likely looking at a triple product, which is a different operation.

• "Projection is always positive." The scalar projection \frac{\vec{a} \cdot \vec{b}}{|\vec{b}|} can be negative — it is negative when the angle between \vec{a} and \vec{b} is obtuse. A negative projection means \vec{a} points partly opposite to \vec{b}.

Going deeper

The dot product as a bilinear form

The properties proved above — commutativity, distributivity over addition, and compatibility with scalar multiplication — are not just a list of useful facts. Together, they say that the dot product is a bilinear form: it is linear in each argument separately. Linearity in the first argument means:

and commutativity gives you linearity in the second argument for free. This bilinear structure is what makes the dot product so algebraically well-behaved: you can expand dot products of sums the same way you expand products of sums of numbers.

For example, (\vec{a} + \vec{b}) \cdot (\vec{a} + \vec{b}) = \vec{a} \cdot \vec{a} + 2(\vec{a} \cdot \vec{b}) + \vec{b} \cdot \vec{b}, which gives:

This is the vector version of the algebraic identity (p + q)^2 = p^2 + 2pq + q^2, and it is how the law of cosines emerges from the dot product.

The Cauchy–Schwarz inequality

From \vec{a} \cdot \vec{b} = |\vec{a}||\vec{b}|\cos\theta and the fact that -1 \leq \cos\theta \leq 1, you immediately get:

This is the Cauchy–Schwarz inequality. Equality holds exactly when \theta = 0 or \theta = \pi — that is, when \vec{a} and \vec{b} are parallel (pointing in the same or opposite directions). In component form, it says:

This purely algebraic inequality — which looks like it would need pages to prove — follows in one line from the geometric interpretation of the dot product. The geometry gives you algebra for free.

Why the angle is between 0 and π

When you place two vectors tail-to-tail, there are two angles formed: one on each side, summing to 2\pi. The convention 0 \leq \theta \leq \pi picks the smaller one. This is not arbitrary — it is the only choice that makes \cos\theta uniquely determine \theta in the range [0, \pi], since \cos is strictly decreasing on this interval. Any other range would create ambiguity: is the angle 60° or 300°? The dot product does not distinguish between them (cosine is the same for both), so the convention picks the one that is well-defined.

The triangle inequality, revisited

You already know the triangle inequality: |\vec{a} + \vec{b}| \leq |\vec{a}| + |\vec{b}|. The dot product gives a clean proof. Start from |\vec{a} + \vec{b}|^2 = |\vec{a}|^2 + 2(\vec{a} \cdot \vec{b}) + |\vec{b}|^2. By Cauchy–Schwarz, \vec{a} \cdot \vec{b} \leq |\vec{a}||\vec{b}|, so:

Taking square roots (both sides are non-negative) gives |\vec{a} + \vec{b}| \leq |\vec{a}| + |\vec{b}|. The inequality is tight (equality holds) exactly when \vec{a} \cdot \vec{b} = |\vec{a}||\vec{b}|, meaning \cos\theta = 1, meaning the two vectors point in the same direction. Geometrically: the third side of a triangle is shorter than the sum of the other two, unless the triangle is degenerate (all three vertices on a line).

Dot product in higher dimensions

Nothing in the component definition a_1b_1 + a_2b_2 + \cdots + a_nb_n requires you to stop at three dimensions. You can define the dot product of two n-dimensional vectors the same way, and all the properties — commutativity, distributivity, the Cauchy–Schwarz inequality — carry over unchanged. The geometric interpretation still holds: \cos\theta = \frac{\vec{a} \cdot \vec{b}}{|\vec{a}||\vec{b}|} defines the angle between two vectors in \mathbb{R}^n, even when you cannot visualise it. This is how angles are measured in machine learning, data science, and quantum mechanics — through the dot product, not through a protractor.

Where this leads next

The dot product is the first of two fundamental ways to multiply vectors. The next steps:

• Applications of Dot Product — work done by a force, the cosine rule derived from dot products, and resolving vectors along perpendicular directions.
• Cross Product — the other vector multiplication, which produces a vector instead of a scalar. It measures how much two vectors are not aligned — the complement of what the dot product measures.
• Components and Direction — the component language used throughout this article, with direction cosines and unit vectors in any direction.
• Vectors Introduction — if the basic language of vectors, types, and notation needs a refresher.
• Scalar Triple Product — where the dot product meets the cross product to compute volumes.