Determinant
One number that says what a matrix does to area — and, by its sign, whether the plane got turned over on the way.
The Matrix
columns of the matrix are where i and j land
What The Unit Square Becomes
—The faint square is the original. The shaded parallelogram is its image.
The Number
Where It Came From
(2.0 × 1.5) − (1.0 × 1.0) = 2.00
Every area in the plane is multiplied by 2.00 — not just this square.
The Determinant
For a square matrix, the determinant records signed area or volume scaling and whether the transformation is reversible.
What it tells you—and what it does not
A square matrix is invertible exactly when its determinant is nonzero. But determinant magnitude is not a general measure of numerical stability: a matrix can have a nonzero determinant and still be ill-conditioned. For computations, use a condition number or a suitable factorisation instead of relying only on det(A). See rank, matrix transformations, and QR decomposition.
Compute it in two dimensions
For A=[[a,b],[c,d]], det(A)=ad−bc. The two terms compare the oriented area spanned by the columns. With A=[[2,1],[1,1]], the determinant is 1: the transformed unit square has the same area, though its shape is sheared. With [[1,2],[2,4]], the second column is twice the first and the determinant is zero; the square collapses to a line. A negative determinant means the map reverses orientation as well as scaling area by its absolute value.
For larger matrices, determinant can be defined through signed sums or computed by elimination, but it is not a practical measure of numerical quality by itself. A very small determinant can result from ordinary scaling in many dimensions; a nonzero determinant can coexist with an ill-conditioned system. In applications, use rank or a condition estimate to reason about near-dependence. The determinant answers invertibility exactly for a square matrix, while rank also describes rectangular maps and projections describe maps that deliberately remove components.
Try this above
Watch the transformed grid while the determinant changes sign or reaches zero. At zero, check that the grid has lost a dimension.
For a 2×2 matrix, the inverse (when it exists) is (1/(ad−bc))[[d,−b],[−c,a]]. The determinant in the denominator shows why a value of zero prevents an inverse. If the determinant is small, inverse entries may become large, but this depends on scaling and is not by itself a condition number. For example, multiplying an otherwise well-behaved matrix by a small scalar changes its determinant while preserving its shape. Use singular values or a condition estimate to assess sensitivity.
In higher dimensions, the determinant's sign records whether orientation is preserved, and its absolute value is the volume scale of the unit cube. If a transformation doubles one axis and triples another without changing the remaining axes, its determinant magnitude is six. If one output direction is a linear combination of others, the transformed volume is zero. Geometric intuition is clearest for square maps; determinant is not defined as an ordinary scalar for rectangular matrices.
Sources and next steps
- MIT OpenCourseWare: Linear Algebra explains determinants, volume, and invertibility.
- Continue with identity, inverse, and transpose to see how invertible maps are undone.
Overview
Read the determinant geometrically
A 2 × 2 matrix maps the unit square to a parallelogram. The absolute determinant is the area scale factor; its sign records whether orientation is preserved or flipped. In three dimensions, the absolute determinant scales volume. A determinant of zero means the transformation collapses at least one dimension.
For A = [[2, 1], [1, 1]], det(A) = 2·1 − 1·1 = 1, so areas are preserved, although the shape can still shear. For [[1, 2], [2, 4]], the determinant is zero: its columns are parallel and the plane collapses to a line.