Cross Product

Hold a door by the handle and push. The door swings open — it rotates on its hinges. Now push on the hinge itself with the same force. Nothing happens. The door does not move.

The force was the same both times. The difference was where you pushed — how far from the hinge, and at what angle. The turning effect of a force depends not on the force alone, but on the force and the position together, combined in a way that cares about the angle between them.

The dot product cannot capture this. The dot product of force and position gives a number related to how much they align — but turning is about how much they don't align. A force perfectly aligned with the door (pushing along it) produces no rotation at all. A force perpendicular to the door (pushing at a right angle) produces maximum rotation. The turning effect is largest when the vectors are most different in direction.

There is an operation that does exactly this. It takes two vectors, measures how much they fail to align, and produces a result that points along the axis of rotation. It is called the cross product — and unlike the dot product, its result is a vector, not a scalar.

The definition

Two things are specified: a magnitude (|\vec{a}||\vec{b}|\sin\theta) and a direction (\hat{n}, perpendicular to both, chosen by the right-hand rule). The result is a full vector — magnitude and direction.

Compare with the dot product. The dot product uses \cos\theta and gives a scalar. The cross product uses \sin\theta and gives a vector. Where the dot product measures alignment (maximum when vectors are parallel), the cross product measures non-alignment (maximum when vectors are perpendicular, zero when they are parallel).

The component formula

The geometric definition is elegant but hard to compute with directly. You need a component formula. If \vec{a} = a_1\hat{i} + a_2\hat{j} + a_3\hat{k} and \vec{b} = b_1\hat{i} + b_2\hat{j} + b_3\hat{k}, then:

This formula looks intimidating until you see where it comes from. The key is the cross products of the unit vectors.

Cross products of unit vectors

Start from the definition. The unit vectors \hat{i}, \hat{j}, \hat{k} are mutually perpendicular and each has magnitude 1. So for \hat{i} and \hat{j}: the angle between them is \pi/2, giving |\hat{i}||\hat{j}|\sin(\pi/2) = 1. The result is a unit vector perpendicular to both \hat{i} and \hat{j} — and by the right-hand rule (curl fingers from \hat{i} toward \hat{j}), that direction is \hat{k}.

These cycle: i \to j \to k \to i. Going the other way reverses the sign (the right-hand rule gives the opposite direction):

And any vector crossed with itself gives zero (since \sin 0 = 0):

Deriving the component formula

Now expand \vec{a} \times \vec{b} using distributivity (which will be proved below):

Distribute to get nine terms:

Three terms vanish (\hat{i} \times \hat{i}, \hat{j} \times \hat{j}, \hat{k} \times \hat{k} are all \vec{0}). The remaining six, using the unit vector products above:

Collecting by component:

This is the component formula.

The determinant shortcut

The component formula has a compact form as a 3 \times 3 determinant:

Expand along the first row:

This is the same formula, but the determinant layout makes it easier to remember. Each component is a 2 \times 2 determinant of the remaining entries: the \hat{i}-component comes from the j and k entries, the \hat{j}-component (with a minus sign) from the i and k entries, and the \hat{k}-component from the i and j entries.

Properties of the cross product

The cross product has clean algebraic rules, but they differ from the dot product's rules in one critical way: the cross product is not commutative.

1. Anti-commutative: \vec{a} \times \vec{b} = -(\vec{b} \times \vec{a})

Swapping the order reverses the direction. From the definition: curling your right hand from \vec{b} to \vec{a} instead of from \vec{a} to \vec{b} points your thumb in the opposite direction. The magnitude |\vec{a}||\vec{b}|\sin\theta stays the same.

From the component formula: swapping \vec{a} and \vec{b} swaps the rows in the determinant, which changes the sign of every component.

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

Expand using the component formula. If \vec{b} + \vec{c} = (b_1 + c_1)\hat{i} + (b_2 + c_2)\hat{j} + (b_3 + c_3)\hat{k}, then the \hat{i}-component of \vec{a} \times (\vec{b} + \vec{c}) is:

This is the \hat{i}-component of \vec{a} \times \vec{b} plus the \hat{i}-component of \vec{a} \times \vec{c}. The same argument applies to the \hat{j} and \hat{k} components.

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

(k\vec{a}) \times \vec{b}: the \hat{i}-component is (ka_2)b_3 - (ka_3)b_2 = k(a_2 b_3 - a_3 b_2). Each component picks up a factor of k, so (k\vec{a}) \times \vec{b} = k(\vec{a} \times \vec{b}).

4. Cross product with itself: \vec{a} \times \vec{a} = \vec{0}

\sin 0 = 0, so |\vec{a}||\vec{a}|\sin 0 = 0. The cross product of any vector with itself is the zero vector.

From the component formula: a_2 a_3 - a_3 a_2 = 0, and similarly for the other components.

5. Not associative: \vec{a} \times (\vec{b} \times \vec{c}) \neq (\vec{a} \times \vec{b}) \times \vec{c} in general

Here is a specific counterexample. Take \vec{a} = \hat{i}, \vec{b} = \hat{j}, \vec{c} = \hat{j}.

So \vec{0} \neq -\hat{i}. The cross product is not associative. Parentheses matter.

Geometric interpretation: area of a parallelogram

The magnitude of the cross product has a beautiful geometric meaning. If \vec{a} and \vec{b} are two sides of a parallelogram, its area is:

Why? Draw the parallelogram with sides \vec{a} and \vec{b}. Its base is |\vec{a}|, and its height is |\vec{b}|\sin\theta — the perpendicular distance from the tip of \vec{b} to the line of \vec{a}. Base times height is |\vec{a}||\vec{b}|\sin\theta, which is exactly the magnitude of the cross product.

For a triangle with two sides \vec{a} and \vec{b}, the area is half the parallelogram:

This gives you a coordinate formula for the area of a triangle in 3D space — something that is surprisingly hard to compute by other methods. You just need the cross product of two side vectors, then take half the magnitude.

A useful corollary: if \vec{a} \times \vec{b} = \vec{0} (and neither vector is zero), then \sin\theta = 0, so \theta = 0 or \theta = \pi — the vectors are parallel (or anti-parallel). The cross product vanishes exactly when the two vectors lie along the same line. Zero area, zero parallelogram, zero non-alignment.

Physical interpretation: moment of a force (torque)

Back to the door. You push at a point whose position from the hinge is \vec{r}, with a force \vec{F}. The moment (or torque) about the hinge is:

The magnitude is |\vec{r}||\vec{F}|\sin\theta — the force times the perpendicular distance from the hinge to the line of force. This is exactly the "turning effect" that your intuition demands: it is zero when you push along the door (parallel, \sin\theta = 0) and maximum when you push perpendicular to the door (\sin\theta = 1).

The direction of \vec{\tau} is along the axis of rotation — the hinge — given by the right-hand rule. If you curl your fingers from \vec{r} toward \vec{F}, your thumb points along the axis around which the door rotates. This is how torque becomes a vector, not just a number: the direction encodes which axis the rotation is around.

The same cross product describes angular momentum (\vec{L} = \vec{r} \times \vec{p}, where \vec{p} is momentum) and the magnetic force on a moving charge (\vec{F} = q\vec{v} \times \vec{B}). In each case, the physical quantity depends on the angle between two vectors and points along the axis of rotation or the perpendicular direction. The cross product is the natural mathematical language for all of these.

Computing the cross product — first example

Torque — second example

Common confusions

• "The cross product is commutative." It is anti-commutative: \vec{a} \times \vec{b} = -(\vec{b} \times \vec{a}). Swapping the order reverses the direction. This is the single most common source of sign errors in cross product calculations. If your answer has the wrong sign, check whether you swapped the order.

• "The cross product works in 2D." Not really. The cross product is defined for vectors in three dimensions. In 2D, you can extend vectors by adding a zero \hat{k}-component (e.g., \vec{a} = a_1\hat{i} + a_2\hat{j} + 0\hat{k}), and then \vec{a} \times \vec{b} will have only a \hat{k}-component: (a_1 b_2 - a_2 b_1)\hat{k}. This is the "2D cross product" that gives the signed area. But the full cross product is inherently three-dimensional.

• "If \vec{a} \times \vec{b} = \vec{0}, one of the vectors must be zero." Not necessarily. \vec{a} \times \vec{b} = \vec{0} when the vectors are parallel (or anti-parallel), since \sin\theta = 0 for \theta = 0 or \pi. The zero cross product is the test for parallelism, just as the zero dot product is the test for perpendicularity.

• "The cross product is associative." It is not. \vec{a} \times (\vec{b} \times \vec{c}) \neq (\vec{a} \times \vec{b}) \times \vec{c} in general, as shown in the properties section. Always respect the parentheses.

• "|\vec{a} \times \vec{b}|^2 + (\vec{a} \cdot \vec{b})^2 = |\vec{a}|^2|\vec{b}|^2." This one is actually true — it is not a confusion but a fact that students often doubt. Since |\vec{a} \times \vec{b}| = |\vec{a}||\vec{b}|\sin\theta and \vec{a} \cdot \vec{b} = |\vec{a}||\vec{b}|\cos\theta, squaring and adding gives |\vec{a}|^2|\vec{b}|^2(\sin^2\theta + \cos^2\theta) = |\vec{a}|^2|\vec{b}|^2. The dot product and cross product together account for the full magnitude — one measures alignment, the other measures non-alignment, and Pythagoras unites them.

Going deeper

The Lagrange identity

The fact noted in the confusions section — |\vec{a} \times \vec{b}|^2 + (\vec{a} \cdot \vec{b})^2 = |\vec{a}|^2|\vec{b}|^2 — is the Lagrange identity for vectors. It says that the dot product and cross product together carry all the information about two vectors: you can recover both the angle and the magnitudes from them.

In component form, this becomes:

This is a purely algebraic identity — it can be verified by expanding both sides — but the vector interpretation makes it almost obvious: it is just \sin^2\theta + \cos^2\theta = 1 multiplied by |\vec{a}|^2|\vec{b}|^2.

Oriented area and the sign of the cross product

The cross product does not just give the area — it gives the oriented area. The direction of \vec{a} \times \vec{b} tells you which side of the parallelogram you are looking at. If you walk around the parallelogram from \vec{a} to \vec{b} (counterclockwise as seen from the direction of \vec{a} \times \vec{b}), the interior is on your left.

This notion of oriented area is essential in physics (for surface integrals, where you need to know which side of a surface a flux is going through) and in computer graphics (for determining whether a triangle faces toward or away from the camera). The cross product provides this information naturally.

Why three dimensions?

The cross product of two vectors produces a vector perpendicular to both. In three dimensions, given any two linearly independent vectors, there is exactly one direction (up to sign) perpendicular to both. In two dimensions, there is no direction perpendicular to two independent vectors within the plane. In four or more dimensions, there are infinitely many directions perpendicular to two given vectors.

This is why the cross product, as defined here, works only in three dimensions: it is the unique dimension where "the perpendicular direction to two vectors" is a well-defined line. The generalization to higher dimensions is the exterior product (or wedge product), which produces not a vector but a different kind of object called a bivector. The exterior product exists in any number of dimensions, but it requires more sophisticated algebraic machinery. For the purposes of physics and JEE mathematics, the three-dimensional cross product is sufficient.

The scalar triple product, previewed

If you take the cross product of \vec{b} and \vec{c}, and then take the dot product of the result with \vec{a}, you get a scalar:

This is the scalar triple product. Its absolute value gives the volume of the parallelepiped (the 3D analogue of a parallelogram) formed by the three vectors. Just as the cross product's magnitude gives area in 2D, the scalar triple product gives volume in 3D. If the scalar triple product is zero, the three vectors are coplanar — they all lie in the same plane, and the parallelepiped has zero volume.

Where this leads next

You now have both vector products: the dot product (measuring alignment) and the cross product (measuring non-alignment). Together, they handle most of the vector algebra in physics and coordinate geometry. The next destinations:

• Scalar Triple Product — the operation \vec{a} \cdot (\vec{b} \times \vec{c}) that gives volumes and tests coplanarity.
• Dot Product — the scalar product, the complementary operation to the cross product.
• Area and Collinearity — using cross products to compute areas of triangles and to check whether three points lie on a line.
• Vectors Introduction — the foundational article on what vectors are, if any of the language here needs revisiting.
• Determinants Introduction — the determinant formalism used in the cross product formula, treated in its own right.