The determinant | Chapter 6, Essence of linear algebra

Determinants turn the messy question of “how much does a linear transformation stretch space?” into a single number: the factor by which areas (in 2D) or volumes (in 3D) change. In two dimensions, that number is the area-scaling factor for any region—positive for ordinary scaling, negative when the transformation flips orientation, and zero when everything collapses into a lower dimension.

A concrete example makes the idea tangible. The matrix with columns (3,0) and (0,2) stretches the i-hat direction by 3 and the j-hat direction by 2. The unit square spanned by i-hat and j-hat becomes a 2 by 3 rectangle, so its area grows from 1 to 6. Because linear transformations send grid lines to parallel, evenly spaced grid lines, every tiny grid square experiences the same area scaling. Any shape can be approximated by many small grid squares, so the entire “blob” scales by the same factor. That universal area factor is the determinant.

Shears show why the determinant is not just about “stretching.” A shear matrix like one with columns (1,1) and (0,1) slants the unit square into a parallelogram without changing its base or height, leaving its area unchanged. Even though the shape looks different, the determinant stays at 1 because the area scale factor is 1.

The determinant also detects when a transformation destroys dimensionality. If the determinant is 0, the transformation squashes all of space into a line (or, in the extreme, a point), making the area or volume of any region collapse to zero. In 2D, this corresponds to the matrix columns becoming linearly dependent—exactly the situation where the transformation can’t span the plane.

Negative determinants add the missing piece: orientation. A transformation can flip a coordinate system—like turning a sheet of paper over—so the determinant becomes negative. The absolute value still gives the magnitude of area scaling. The intuition comes from continuously moving i-hat toward j-hat: as they align, the determinant approaches 0; pushing past alignment makes the determinant continue into negative values, reflecting the flip.

In three dimensions, the same logic applies to volumes. The determinant gives the factor by which a unit cube becomes a parallelipiped, and a determinant of 0 means the cube collapses into something with zero volume (a plane, a line, or a point). Orientation in 3D is tracked with the right-hand rule: if the transformed i-hat and j-hat still produce k-hat using the right hand, the determinant is positive; if only the left hand works, it’s negative.

Finally, the transcript connects meaning to computation. For a 2×2 matrix [[a,b],[c,d]], the determinant is ad − bc, with intuition tied to how the unit square’s image depends on stretching along axes versus diagonal effects. It also notes a key algebraic property: det(AB) = det(A)det(B), so determinants multiply under matrix multiplication—an insight that will matter when linear transformations are composed to solve systems of equations later.