Calculus › Section L · Vectors and the geometry of space › Topic L.4
The cross product
The dot product turns two vectors into a number; the cross product turns them into a third vector, one standing perpendicular to both. That single construction unlocks areas, torque and volumes.
A vector perpendicular to two vectors
Many problems ask for a direction orthogonal to two given ones: the axis a wheel spins about, the normal to a tilted surface, the direction of a wrench's twist. The cross product \(\mathbf{u} \times \mathbf{v}\) is the answer to that request, and it exists only for vectors in three dimensions, where there is room to stand perpendicular to a plane. It is defined by a formula that looks forbidding at first and turns out to be a simple pattern:
Each component subtracts one diagonal product from another, cycling through the coordinate positions: 2 with 3, then 3 with 1, then 1 with 2.
The determinant shortcut
The pattern is worth memorising as a \(3 \times 3\) determinant. Write the standard unit vectors across the top row, the components of \(\mathbf{u}\) in the second row and those of \(\mathbf{v}\) in the third:
Expanding along the top row gives exactly the component formula: \(\mathbf{i}(u_2v_3 - u_3v_2) - \mathbf{j}(u_1v_3 - u_3v_1) + \mathbf{k}(u_1v_2 - u_2v_1)\). The middle sign is a minus, which is the classic trap. Note that the rows must be in order, \(\mathbf{u}\) above \(\mathbf{v}\), because the cross product is anticommutative.
Worked example: computing a cross product
Compute \(\langle 1, 2, 0\rangle \times \langle 0, 1, 3\rangle\) and verify the result is perpendicular to both inputs. First component: \(2 \cdot 3 - 0 \cdot 1 = 6\). Second: \(0 \cdot 0 - 1 \cdot 3 = -3\). Third: \(1 \cdot 1 - 2 \cdot 0 = 1\). So \(\mathbf{u} \times \mathbf{v} = \langle 6, -3, 1\rangle\). Now check: \(\langle 6, -3, 1\rangle \cdot \langle 1, 2, 0\rangle = 6 - 6 + 0 = 0\), and \(\langle 6, -3, 1\rangle \cdot \langle 0, 1, 3\rangle = 0 - 3 + 3 = 0\). Both dot products vanish, so the cross product really is orthogonal to each input. This two-line check should become a habit: it catches most arithmetic slips instantly.
Direction and order
The cross product's direction follows the right-hand rule: curl the fingers of your right hand from \(\mathbf{u}\) toward \(\mathbf{v}\) and your thumb points along \(\mathbf{u} \times \mathbf{v}\). Swapping the order swaps the thumb: \(\mathbf{v} \times \mathbf{u} = -(\mathbf{u} \times \mathbf{v})\), the anticommutative law, visible in the determinant because the two lower rows traded places. Two more algebra facts are immediate from the formula: \(\mathbf{u} \times \mathbf{u} = \mathbf{0}\), since every diagonal difference cancels, and the cross product distributes over addition. The standard unit vectors cycle among themselves: \(\mathbf{i} \times \mathbf{j} = \mathbf{k}\), \(\mathbf{j} \times \mathbf{k} = \mathbf{i}\) and \(\mathbf{k} \times \mathbf{i} = \mathbf{j}\), with reversed products giving the negatives.
Length as area
Geometry gives the cross product's magnitude a meaning:
where \(\theta\) is the angle between the vectors. Compare with the dot product, which carries a cosine: the cross product is largest when the vectors are perpendicular and zero when they are parallel. The value is exactly the area of the parallelogram that \(\mathbf{u}\) and \(\mathbf{v}\) span, base times perpendicular height, and half of it is the area of the triangle with \(\mathbf{u}\) and \(\mathbf{v}\) as two sides. This is the standard way to compute areas of figures floating anywhere in space, where no base-times-height formula is directly visible.
Worked example: area of a triangle in space
Find the area of the triangle with vertices \(P(0, 0, 0)\), \(Q(2, 3, 0)\) and \(R(1, 4, 0)\). Two edge vectors from \(P\) are \(\mathbf{u} = \overrightarrow{PQ} = \langle 2, 3, 0\rangle\) and \(\mathbf{v} = \overrightarrow{PR} = \langle 1, 4, 0\rangle\). Their cross product is \(\langle 3 \cdot 0 - 0 \cdot 4,\ 0 \cdot 1 - 2 \cdot 0,\ 2 \cdot 4 - 3 \cdot 1 \rangle = \langle 0, 0, 5\rangle\), with magnitude 5. The parallelogram area is 5, so the triangle's area is \(\frac{5}{2} = 2.5\). Because the triangle lies flat in the \(xy\)-plane, an independent check is available: the usual shoelace formula on \((0,0)\), \((2,3)\), \((1,4)\) gives \(\frac{1}{2}|2 \cdot 4 - 3 \cdot 1| = \frac{5}{2}\). The two methods agree.
Torque: the cross product in action
Torque, the twisting effect of a force, is the physical reason the cross product was invented. A force \(\mathbf{F}\) applied at a point with position vector \(\mathbf{r}\) from the pivot produces torque
Only the force perpendicular to the lever arm contributes, which is the sine in \(\|\mathbf{r}\|\|\mathbf{F}\|\sin\theta\), and the twist happens about an axis perpendicular to both the arm and the force, which is the direction of the product. Pushing a door at its edge, straight at the hinges, produces no twist; pushing perpendicular to the door at the handle produces the most. A 20-newton push perpendicular to a 0.5-metre arm gives a torque of magnitude 10 newton-metres about the hinge axis.
The triple scalar product
Combining both products, \(\mathbf{u} \cdot (\mathbf{v} \times \mathbf{w})\), is called the triple scalar product: cross two vectors, then dot with a third, producing a number. Its absolute value is the volume of the parallelepiped, the skewed box, that the three vectors span. The reason is pleasing: \(\mathbf{v} \times \mathbf{w}\) is a vector whose length is the base parallelogram's area and whose direction is perpendicular to the base, so dotting with \(\mathbf{u}\) multiplies that area by the box's perpendicular height. For \(\mathbf{u} = \langle 1, 0, 0\rangle\), \(\mathbf{v} = \langle 0, 2, 0\rangle\), \(\mathbf{w} = \langle 0, 0, 3\rangle\): the cross product of \(\mathbf{v}\) and \(\mathbf{w}\) is \(\langle 2 \cdot 3 - 0 \cdot 0,\ 0 \cdot 0 - 0 \cdot 3,\ 0 \cdot 0 - 2 \cdot 0\rangle = \langle 6, 0, 0\rangle\), and dotting with \(\mathbf{u}\) gives \(6\): the volume of a 1-by-2-by-3 box, as expected.
Common mistakes to avoid
The classic error is swapping the order, which flips the answer's sign: \(\mathbf{v} \times \mathbf{u}\) is not \(\mathbf{u} \times \mathbf{v}\). The second is forgetting the minus sign on the middle component in the determinant expansion. The third is confusing area with triangle area: \(\|\mathbf{u} \times \mathbf{v}\|\) is the parallelogram's area, and the triangle needs the extra half. The fourth is trying to cross two-dimensional vectors: the cross product needs three components, so embed a plane vector as \(\langle v_1, v_2, 0\rangle\) first.
Key takeaways
The cross product \(\mathbf{u} \times \mathbf{v} = \langle u_2v_3 - u_3v_2, u_3v_1 - u_1v_3, u_1v_2 - u_2v_1\rangle\) is a vector perpendicular to both inputs, with direction from the right-hand rule and length \(\|\mathbf{u}\|\|\mathbf{v}\|\sin\theta\), the parallelogram area. It is anticommutative, distributive, and zero for parallel vectors. Triangle areas take half; torque is \(\mathbf{r} \times \mathbf{F}\); and the triple scalar product \(\mathbf{u} \cdot (\mathbf{v} \times \mathbf{w})\) measures the volume of the spanned parallelepiped. Always verify a computed cross product by dotting it with both inputs.
Flashcards
Flashcard
u × v = ⟨u₂v₃ − u₃v₂, u₃v₁ − u₁v₃, u₁v₂ − u₂v₁⟩
Test your understanding
Question 1 of 5
What is ⟨1, 2, 0⟩ × ⟨0, 1, 3⟩?
Question 2 of 5
What is i × j?
Question 3 of 5
The triangle with vertices (0, 0, 0), (2, 3, 0) and (1, 4, 0) has area:
Question 4 of 5
If u and v are parallel nonzero vectors, what is u × v?
Question 5 of 5
What is the volume of the parallelepiped spanned by ⟨1, 0, 0⟩, ⟨0, 2, 0⟩ and ⟨0, 0, 3⟩?
Standards: Beyond AP
