Cross products in the light of linear transformations | Chapter 11, Essence of linear algebra

The 3D cross product isn’t just a memorized formula—it’s the dual vector of a specific linear transformation built from two vectors v and w. Once that connection is made, the usual geometric claims about v × w (perpendicular direction, right-hand rule, and magnitude equal to parallelogram area) stop feeling like “just believe it” facts and instead follow from how determinants encode signed volume.

The starting point is the familiar determinant trick used to compute v × w: form a 3×3 matrix whose second and third columns are the coordinates of v and w, while the first column is the symbolic basis placeholders i-hat, j-hat, and k-hat. Expanding the determinant yields three coefficients, which are interpreted as the coordinates of the resulting vector. Rather than treating this as a purely computational stunt, the reasoning reframes it as follows: fix v and w, and let the first column be an arbitrary vector x = (x, y, z). Then the determinant becomes a function f(x) that maps 3D vectors to a single number—specifically, the signed volume of the parallelepiped spanned by x, v, and w.

Because determinants are linear in their columns, this function f(x) is linear in x. Linear maps from 3D to the number line have a special representation via duality: there exists a vector p such that f(x) equals the dot product p · x for every x. So the cross product emerges as the identity of that dual vector. In other words, v × w is the unique vector p satisfying p · x = det([x v w]) for all x.

The geometry then locks in the same conclusion. A dot product p · x can be read as the length of x projected onto the direction of p, multiplied by the length of p. Meanwhile, the determinant det([x v w]) is the signed volume, which can be decomposed into “base area times height”: the base is the area of the parallelogram spanned by v and w, and the height is the component of x perpendicular to that parallelogram. That means the determinant equals (area of v–w parallelogram) × (signed perpendicular component of x). The only vector p that reproduces this dot product for every x must be perpendicular to both v and w, with magnitude equal to the parallelogram’s area. The sign matches the orientation encoded by the right-hand rule.

So the determinant computation and the geometric properties are two descriptions of the same dual vector. The cross product’s direction, magnitude, and orientation aren’t separate facts—they’re the inevitable consequence of representing a determinant-based linear functional as a dot product with a perpendicular vector. The payoff is a cleaner mental model for cross products that sets up the next major linear algebra idea: change of basis.