Applications of Dot Product

A rickshaw-puller hauls his rickshaw along a flat road. He pulls with a force of 200 newtons, but his arms are angled upward at 30° to the ground. Not all of that 200 N is moving the rickshaw forward — some of it is wasted lifting upward against the road. How much of his effort actually goes into moving the rickshaw horizontally?

The answer is 200 \cos 30° = 100\sqrt{3} \approx 173 N. The cosine of the angle filters out the "useless" vertical part and keeps only the component along the direction of motion. And the total work done over a displacement — force times the displacement in the direction of the force — is exactly a dot product.

This is the pattern. Wherever you need to extract the component of one vector along another, the dot product is doing the work. This article collects the four most important places this idea appears.

Application 1: Work done by a force

In physics, work is defined as the energy transferred when a force moves an object through a displacement. If the force and the displacement point in the same direction, the work is simply force times distance. But when they point in different directions — when you pull upward but the object moves horizontally — you only count the part of the force that acts along the displacement.

The definition is:

where \vec{F} is the force and \vec{d} is the displacement. This is the dot product — nothing more. The \cos\theta automatically extracts the right component.

Three cases make the formula clear:

• \theta = 0 (force and displacement in the same direction): W = |\vec{F}||\vec{d}|. All of the force contributes. Pushing a box along the floor in the direction you push.

• \theta = 90° (force perpendicular to displacement): W = 0. None of the force contributes. Carrying a suitcase horizontally — gravity pulls down but you move sideways, so gravity does zero work on the suitcase.

• \theta = 180° (force opposes displacement): W = -|\vec{F}||\vec{d}|. The work is negative — energy is taken out of the moving object. Friction does negative work when you slide to a stop.

In component form, if \vec{F} = F_x\hat{i} + F_y\hat{j} + F_z\hat{k} and \vec{d} = d_x\hat{i} + d_y\hat{j} + d_z\hat{k}, then:

This is the reason the dot product is taught in physics alongside vectors. It is not a mathematical curiosity — it is the definition of work.

A note on units: if the force is in newtons and the displacement in metres, the work is in joules. The dot product handles the unit arithmetic naturally — newtons times metres gives joules, and the \cos\theta is dimensionless. The formula works in any consistent set of units without modification.

Application 2: Deriving the cosine rule

The cosine rule (also called the law of cosines) says that for a triangle with sides a, b, c and the angle C opposite to side c:

Most textbooks prove this using coordinate geometry or by dropping a perpendicular. The dot product gives a proof in three lines.

The proof. Let the three vertices be A, B, C. Define \vec{a} = \overrightarrow{CA} and \vec{b} = \overrightarrow{CB}. Then |\vec{a}| = b (the side opposite B), |\vec{b}| = a (the side opposite A), and the angle between \vec{a} and \vec{b} is C.

The third side of the triangle is \vec{a} - \vec{b} = \overrightarrow{BA}, and its length is c. Compute c^2:

Expand using bilinearity (distributive property):

That is the cosine rule. The entire proof rested on one idea: expand the squared magnitude using the dot product.

Notice something: the cosine rule was used to prove that the two definitions of the dot product (geometric and component) agree. Now the dot product is being used to derive the cosine rule. This is not circular. The equivalence proof uses the cosine rule in a triangle of known side-lengths (given by components). Once the equivalence is established, the dot product's algebraic properties — especially distributivity — give the cosine rule for any triangle, in any dimension, with no coordinates needed.

As a bonus, the same expansion technique gives the cosine rule's companion. Expanding |\vec{a} + \vec{b}|^2 instead of |\vec{a} - \vec{b}|^2:

This is the law for the other diagonal of the parallelogram formed by \vec{a} and \vec{b}. Adding the two expansions:

This is the parallelogram law: the sum of the squares of the diagonals equals twice the sum of the squares of the sides. One line of dot-product algebra gave you a classical geometric identity.

Application 3: Resolving a vector along perpendicular directions

In physics and engineering, you often need to break a vector into two parts: one along a chosen direction and one perpendicular to it. This is called resolving the vector.

You already met the projection formula: the component of \vec{a} along \vec{b} is

The perpendicular component is whatever is left over:

Check that this is actually perpendicular to \vec{b}:

The perpendicular component has zero dot product with \vec{b}, confirming it is perpendicular. The decomposition \vec{a} = \vec{a}_{\parallel} + \vec{a}_{\perp} is exact — no information is lost.

This is the general version of what you do in physics every time you set up axes. When you resolve a force into horizontal and vertical components, you are projecting the force vector onto \hat{i} and \hat{j}. When you resolve a velocity along and perpendicular to a slope, you are projecting onto a unit vector along the slope. The dot product gives you the tool to do this along any direction, not just the coordinate axes.

Application 4: Condition for perpendicularity

Two nonzero vectors \vec{a} and \vec{b} are perpendicular (at right angles) if and only if

This follows directly from \vec{a} \cdot \vec{b} = |\vec{a}||\vec{b}|\cos\theta: since |\vec{a}| and |\vec{b}| are both positive (the vectors are nonzero), the product is zero exactly when \cos\theta = 0, which happens exactly when \theta = \pi/2.

In component form, \vec{a} \perp \vec{b} if and only if a_1b_1 + a_2b_2 + a_3b_3 = 0.

This test is astonishingly useful. Without the dot product, checking whether two vectors are perpendicular requires you to find the angle between them — which means inverse-cosine, which means a calculator. With the dot product, you compute one number and check if it is zero. No trigonometry needed.

Quick example. Are \vec{p} = 3\hat{i} - 2\hat{j} + \hat{k} and \vec{q} = 2\hat{i} + \hat{j} + 4\hat{k} perpendicular? Compute: 3(2) + (-2)(1) + 1(4) = 6 - 2 + 4 = 8 \neq 0. Not perpendicular.

Are \vec{p} = 3\hat{i} - 2\hat{j} + \hat{k} and \vec{r} = 2\hat{i} + \hat{j} - 4\hat{k} perpendicular? Compute: 3(2) + (-2)(1) + 1(-4) = 6 - 2 - 4 = 0. Yes, perpendicular.

The second pair passed the test — zero dot product, so they form a right angle. This kind of computation comes up constantly in coordinate geometry and physics: checking if lines are perpendicular, verifying that a vector lies in a plane, confirming that a force has no component in a certain direction.

The perpendicularity condition also has a neat algebraic form. For two vectors \vec{a} = a_1\hat{i} + a_2\hat{j} and \vec{b} = b_1\hat{i} + b_2\hat{j} in 2D, \vec{a} \perp \vec{b} means a_1 b_1 + a_2 b_2 = 0, which rearranges to \frac{a_2}{a_1} \cdot \frac{b_2}{b_1} = -1 (when the denominators are nonzero). If you recognise a_2/a_1 as the slope of \vec{a} and b_2/b_1 as the slope of \vec{b}, this is the familiar condition from coordinate geometry: two lines are perpendicular if and only if the product of their slopes is -1. The dot product reveals this as a special case of a general vector condition — the slopes-multiply-to--1 rule is the 2D version of \vec{a} \cdot \vec{b} = 0.

Computing work done — first example

Resolving a vector — second example

Common confusions

• "Work can't be negative." It absolutely can. When the force opposes the displacement (angle greater than 90°), \cos\theta is negative, and the work is negative. Negative work means energy is being taken out of the object — friction slowing you down does negative work on you.

• "Perpendicular vectors have no relationship." They have a very specific relationship: their dot product is zero. Saying two vectors are perpendicular is a strong geometric statement, not the absence of a statement. The condition \vec{a} \cdot \vec{b} = 0 is one of the most-used equations in all of vector geometry.

• "The cosine rule is a trigonometry result, not a vector result." It is both. The trigonometric proof and the vector proof give the same formula. But the vector proof is shorter, works in any number of dimensions, and reveals why the cosine appears: it comes from the dot product, which naturally produces cosines of angles.

• "You can only resolve along \hat{i} and \hat{j}." The projection formula works with any direction, not just the coordinate axes. You can project a velocity onto the direction of a slope, a force onto the direction of a wire, or a vector onto any unit vector you choose. The axes \hat{i} and \hat{j} are just the most common choice.

• "If \vec{a} \cdot \vec{b} = 0, then \vec{a} = \vec{0} or \vec{b} = \vec{0}." Not true. This is the "zero product property" of real numbers, but it does not hold for the dot product. \vec{a} \cdot \vec{b} = 0 most often means the vectors are perpendicular, not that either is zero.

Going deeper

Orthogonal decomposition as a general principle

The resolution of \vec{a} into \vec{a}_{\parallel} and \vec{a}_{\perp} is an instance of a much deeper idea: any vector space with an inner product (the dot product is one such inner product) supports orthogonal decomposition. Given any subspace W (in the simplest case, a line through the origin), every vector \vec{v} in the full space splits uniquely as

where \vec{v}_W lies in W and \vec{v}_{W^\perp} is perpendicular to every vector in W. The formula is the same projection formula you used above, and the perpendicularity check is the same dot-product-equals-zero test.

This principle scales up. In higher dimensions, you can project onto a plane, onto a three-dimensional subspace, onto any subspace at all. The formula always involves dot products, and the perpendicular remainder always has zero dot product with the subspace. This is the foundation of least-squares fitting, Fourier series, and the Gram–Schmidt process — all of which you will meet if you continue into linear algebra.

The polarization identity

Here is a surprising relationship. The dot product can be recovered from magnitudes alone:

This follows from expanding |\vec{a} + \vec{b}|^2 = |\vec{a}|^2 + 2(\vec{a} \cdot \vec{b}) + |\vec{b}|^2 and solving for \vec{a} \cdot \vec{b}.

The identity says: if you know the length of \vec{a}, the length of \vec{b}, and the length of \vec{a} + \vec{b}, you can compute the dot product without knowing any components or any angles. Three length measurements suffice. This is the polarization identity, and it is theoretically important because it shows that the inner product (the dot product) is completely determined by the norm (the length function). Any notion of length that satisfies the parallelogram law automatically gives you a notion of angle, through exactly this formula.

The cosine rule gives the dot product its geometric meaning

It is worth pausing to see the logical flow:

1. You define the dot product as a_1b_1 + a_2b_2 + a_3b_3 — a purely algebraic object.
2. You prove the cosine rule using the dot product's algebraic properties.
3. You read off the geometric meaning: \vec{a} \cdot \vec{b} = |\vec{a}||\vec{b}|\cos\theta.

This order — algebra first, geometry second — is the opposite of how the definition is usually presented (geometry first: "the dot product is |\vec{a}||\vec{b}|\cos\theta"). Both orderings are logically valid. The algebraic-first approach has the advantage of generalizing to higher dimensions, where you cannot draw pictures but can still compute a_1b_1 + a_2b_2 + \cdots + a_nb_n. The geometric-first approach has the advantage of making the physical meaning immediate. The two are equivalent, and which you prefer is a matter of taste.

Verifying the perpendicularity condition with direction cosines

If two vectors have direction cosines (l_1, m_1, n_1) and (l_2, m_2, n_2), then their unit vectors are \hat{a} = l_1\hat{i} + m_1\hat{j} + n_1\hat{k} and \hat{b} = l_2\hat{i} + m_2\hat{j} + n_2\hat{k}. The angle between the vectors satisfies:

The perpendicularity condition \theta = \pi/2 reduces to:

This is the condition for perpendicularity of two lines in three-dimensional geometry, written in terms of their direction cosines. It appears frequently in the study of straight lines in 3D, and it is nothing more than the dot product of unit vectors set equal to zero.

Where this leads next

The dot product and its applications are one half of the story of vector multiplication. The other half:

• Dot Product — the definition, properties, and proofs that underlie everything in this article.
• Cross Product — the vector product, which produces a vector perpendicular to both inputs and measures area rather than alignment.
• Trigonometric Identities — the identities used alongside the dot product, especially \cos(\alpha - \beta) and its relationship to component-wise multiplication.
• Coordinate Geometry Basics — the coordinate framework in which these component formulas live.
• Distance Formulas — using |\vec{a} - \vec{b}| (which involves the dot product) to compute distances in 2D and 3D.