Linear Algebra › Section C · Determinants › Topic C.3

Cramer's rule, volume, and transformations

Determinants give a formula for the solution of a square system, a formula for the inverse, and a precise meaning in geometry: they measure how a matrix scales area and volume and whether it flips orientation.

From a test to formulas and geometry

So far the determinant has been a yes-or-no test: non-zero means invertible. This topic gets more out of it. First, Cramer's rule writes each unknown of a square system \(A\mathbf{x} = \mathbf{b}\) as a ratio of two determinants. Second, the same idea produces a formula for every entry of \(A^{-1}\). Third, the absolute value of a determinant is an area or a volume, and that turns every linear transformation into a scaling machine for regions. The rules from "Properties of determinants", especially \(\det(AB) = \det A \det B\), do most of the work.

Cramer's rule

For an n x n matrix \(A\) and a vector \(\mathbf{b}\), write \(A_i(\mathbf{b})\) for the matrix \(A\) with column \(i\) replaced by \(\mathbf{b}\). Cramer's rule, published by Gabriel Cramer in 1750, says: if \(A\) is invertible, the unique solution of \(A\mathbf{x} = \mathbf{b}\) has entries

x_i = \frac{\det A_i(\mathbf{b})}{\det A}, \quad i = 1, 2, \ldots, n

Each unknown needs its own numerator, so an n x n system needs \(n + 1\) determinants. The rule says nothing when \(\det A = 0\); then the system has either no solution or infinitely many.

Worked example: Cramer's rule for two unknowns

Solve \(4x_1 - x_2 = 6\) and \(2x_1 + 3x_2 = -4\). 1. The coefficient matrix has rows (4, -1) and (2, 3), so \(\det A = 12 + 2 = 14\). It is not zero, so the rule applies. 2. Replace column 1 by \(\mathbf{b} = (6, -4)\): rows (6, -1) and (-4, 3), with determinant \(18 - 4 = 14\). So \(x_1 = 14/14 = 1\). 3. Replace column 2 by \(\mathbf{b}\): rows (4, 6) and (2, -4), with determinant \(-16 - 12 = -28\). So \(x_2 = -28/14 = -2\). 4. Check: \(4(1) - (-2) = 6\) and \(2(1) + 3(-2) = -4\). Both equations hold.

Why Cramer's rule works

Let \(I_i(\mathbf{x})\) be the identity matrix with column \(i\) replaced by the solution \(\mathbf{x}\). Multiplying, \(A \, I_i(\mathbf{x})\) has columns \(A\mathbf{e}_1, \ldots, A\mathbf{x}, \ldots, A\mathbf{e}_n\), which is exactly \(A_i(\mathbf{b})\), because \(A\mathbf{x} = \mathbf{b}\). Taking determinants, \(\det A \cdot \det I_i(\mathbf{x}) = \det A_i(\mathbf{b})\). Expanding \(I_i(\mathbf{x})\) along row \(i\) shows its determinant is just \(x_i\). Divide by \(\det A\) and the rule appears. Cramer's rule is valuable in theory and when entries contain a parameter, but for numerical systems row reduction is far faster.

A formula for the inverse

Applying Cramer's rule to each column of \(AA^{-1} = I\) gives a formula for the whole inverse. Form the matrix of cofactors \(C_{ij}\) and transpose it; the result is the adjugate, \(\operatorname{adj} A\). Then

A^{-1} = \frac{1}{\det A}\operatorname{adj} A

so the \((i, j)\) entry of \(A^{-1}\) is \(C_{ji}/\det A\). Note the swapped subscripts. For a 2 x 2 matrix this gives the familiar rule: swap \(a\) and \(d\), negate \(b\) and \(c\), and divide by \(ad - bc\). For the matrix with rows (2, -1, 3), (1, 4, 0), (-2, 5, 1), whose determinant is 48, the (3, 1) entry of the inverse is \(C_{13}/48 = 13/48\).

Determinants as area and volume

If \(A\) is a 2 x 2 matrix, the parallelogram with sides along its two columns has area \(|\det A|\). If \(A\) is 3 x 3, the parallelepiped (a slanted box) with edges along its three columns has volume \(|\det A|\). The reason is that column replacements slide the shape without changing its area or volume, until the columns are perpendicular and the shape is a rectangle or box. A determinant of zero means the columns lie on a line or in a plane, so the shape is flat. Two edge vectors from any vertex give the same area, so a parallelogram need not start at the origin.

Worked example: a triangle and a box

1. Find the area of the triangle with vertices (-1, 2), (3, 1) and (1, 5). Edge vectors from (-1, 2) are (4, -1) and (2, 3). The parallelogram they span has area \(|4 \cdot 3 - (-1) \cdot 2| = 14\), and the triangle is half of it: 7 square units. 2. Find the volume of the parallelepiped with edges (1, 0, 2), (3, 1, 0) and (0, 2, 1). Put them in the columns of

M = \begin{bmatrix} 1 & 3 & 0 \\ 0 & 1 & 2 \\ 2 & 0 & 1 \end{bmatrix}

Expanding along row 1,

\det M = 1(1 - 0) - 3(0 - 4) + 0 = 13

The volume is 13 cubic units.

How a transformation scales regions

Let \(T(\mathbf{x}) = A\mathbf{x}\) for a 2 x 2 matrix \(A\). The unit square maps to the parallelogram spanned by the columns of \(A\), so its area is multiplied by \(|\det A|\). Because any region can be filled with tiny squares, every region with finite area is scaled by the same factor:

\text{area of } T(S) = |\det A| \cdot \text{area of } S

The same holds for volume in three dimensions. For example, with columns (2, 1) and (1, -1), \(\det A = -2 - 1 = -3\), so a region of area 5 maps to one of area 15. Composing transformations multiplies the factors, as \(\det(AB) = \det A \det B\) predicts.

Orientation and where determinants reappear

The sign of \(\det A\) records orientation. A positive determinant keeps a counterclockwise pair of vectors counterclockwise; a negative one reverses it, as a mirror does. Rotations have determinant 1 and reflections -1, and both preserve area. Determinants reappear widely. In calculus, the Jacobian determinant scales small areas under a change of variables, which is why polar coordinates use \(r \, dr \, d\theta\). The cross product in \(\mathbb{R}^3\) is a formal determinant with \(\mathbf{i}, \mathbf{j}, \mathbf{k}\) in the top row. Three points in the plane lie on a line exactly when the 3 x 3 determinant with rows \((x_k, y_k, 1)\) is 0.

Common mistakes to avoid

In Cramer's rule, replace the column belonging to the unknown, not a row, and keep \(\det A\) as the denominator every time. Never use the rule when \(\det A = 0\). In the inverse formula, the \((i, j)\) entry uses \(C_{ji}\), not \(C_{ij}\): forgetting the transpose is the usual error. An area or volume is \(|\det A|\); a negative determinant is not a negative area, only a reversed orientation. A triangle is half the parallelogram, and its edge vectors must start from the same vertex.

Key takeaways

For invertible \(A\), Cramer's rule gives \(x_i = \det A_i(\mathbf{b}) / \det A\), where \(A_i(\mathbf{b})\) has column \(i\) replaced by \(\mathbf{b}\). The inverse is \(\frac{1}{\det A}\operatorname{adj} A\), whose \((i, j)\) entry is \(C_{ji}/\det A\). Both are elegant but slow for large systems. \(|\det A|\) is the area or volume spanned by the columns, and \(\mathbf{x} \mapsto A\mathbf{x}\) multiplies every area or volume by \(|\det A|\). The sign records orientation: positive keeps it, negative reverses it.

Flashcards

Flashcard

Aᵢ(b)

A with column i replaced by b

The numerator matrix in Cramer's rule.

Test your understanding

Question 1 of 5

Using Cramer's rule on 2x₁ + x₂ = 7 and x₁ - 3x₂ = 0, what is x₁?

Question 2 of 5

For an invertible 3 x 3 matrix A, which expression gives the (1, 2) entry of A⁻¹?

Question 3 of 5

What is the area of the parallelogram with sides along the vectors (5, 2) and (1, 3)?

Question 4 of 5

T(x) = Ax with A having columns (3, 1) and (2, 4). A region S has area 6. What is the area of T(S)?

Question 5 of 5

A 2 x 2 matrix A has det A = -2. What does T(x) = Ax do to a region of area 4?

← Previous topicNext topic →