Linear Algebra › Section C · Determinants › Topic C.2

Properties of determinants

Cofactor expansion defines the determinant, but its properties are what make it useful. Knowing how row operations, transposes and products change a determinant lets you compute it quickly and prove that it decides invertibility.

Why properties matter

The previous topic, "Determinants and cofactor expansion", ended with a problem: full expansion of an n x n determinant needs n! terms. The way out uses an idea you already know from "Row reduction and echelon forms". Row operations turn any square matrix into an echelon form, which is triangular, and the determinant of a triangular matrix is just the product of its diagonal. If we know exactly how each row operation changes the determinant, we can reduce first and multiply the diagonal afterwards. The same rules also explain why a zero determinant means a matrix has no inverse.

How row operations change the determinant

Let \(A\) be a square matrix and \(B\) the result of one row operation on it. 1. Replacement: adding a multiple of one row to another leaves the determinant unchanged, so \(\det B = \det A\). 2. Interchange: swapping two rows changes the sign, so \(\det B = -\det A\). 3. Scaling: multiplying one row by \(k\) multiplies the determinant by \(k\), so \(\det B = k \det A\). A useful consequence: if a matrix has two equal rows, swapping them changes nothing yet must flip the sign, so its determinant is 0. A row of zeros also gives 0, by expanding along that row.

Why the rules hold

Each row operation is the same as multiplying on the left by an elementary matrix \(E\), which is the identity with that one operation applied. A replacement matrix is triangular with 1s on the diagonal, so its determinant is 1. A scaling matrix is diagonal with one entry \(k\), so its determinant is \(k\). An interchange matrix has determinant -1, which you can check directly in the 2 x 2 case: the matrix with rows (0, 1) and (1, 0) gives \(0 - 1 = -1\). In every case \(\det(EA) = (\det E)(\det A)\), which is exactly the three rules.

Computing a determinant by row reduction

Row reduce \(A\) to an echelon form \(U\) using only replacements and interchanges, and let \(r\) be the number of interchanges. Replacements change nothing and each swap flips the sign, so

\det A = (-1)^r \, u_{11}u_{22}\cdots u_{nn}

where the \(u_{ii}\) are the diagonal entries of \(U\). If you also scale a row by \(k\) along the way, divide the final answer by \(k\) to undo it. This method takes about \(\frac{2}{3}n^3\) arithmetic operations, so a 25 x 25 determinant needs roughly ten thousand steps instead of about \(1.6 \times 10^{25}\).

Worked example: reduce, then multiply

Find the determinant of

A = \begin{bmatrix} 0 & 2 & 4 \\ 1 & 3 & -1 \\ 2 & 8 & 6 \end{bmatrix}

1. The top-left entry is 0, so swap rows 1 and 2. That is one interchange, \(r = 1\). The rows are now (1, 3, -1), (0, 2, 4), (2, 8, 6). 2. Replace row 3 by row 3 minus 2 times row 1, giving (0, 2, 8). 3. Replace row 3 by row 3 minus row 2, giving (0, 0, 4). 4. The echelon form is triangular with diagonal 1, 2, 4, so \(\det A = (-1)^1(1 \cdot 2 \cdot 4) = -8\). 5. Check by cofactor expansion along row 1:

0 - 2(6 + 2) + 4(8 - 6) = -16 + 8 = -8

The invertibility test

A square matrix \(A\) is invertible if and only if \(\det A \ne 0\). Here is why. Reduce \(A\) to echelon form \(U\) with replacements and swaps; then \(\det A\) is \(\pm\) the product of the diagonal of \(U\). If \(A\) is invertible, it has a pivot in every column, so every diagonal entry of \(U\) is non-zero and so is the product. If \(A\) is not invertible, \(U\) has fewer than n pivots, its last row is zero, and the product is 0. This adds one more condition to the list in "The invertible matrix theorem". In particular, if the columns or rows of \(A\) are linearly dependent, \(\det A = 0\).

Transposes and column operations

A matrix and its transpose have the same determinant: \(\det A^T = \det A\). For 2 x 2 matrices both give \(ad - bc\); in general, expanding \(A^T\) along a row is the same as expanding \(A\) down the matching column. The practical payoff is that every row rule has a column twin. Swapping two columns flips the sign, scaling a column by \(k\) multiplies the determinant by \(k\), adding a multiple of one column to another changes nothing, and two equal columns force a zero determinant.

Products, inverses and scalar multiples

Determinants respect multiplication: for n x n matrices, \(\det(AB) = (\det A)(\det B)\). Since \(AA^{-1} = I\) and \(\det I = 1\), it follows that \(\det(A^{-1}) = 1/\det A\), and \(\det(A^k) = (\det A)^k\). Scaling the whole matrix is different from scaling one row: \(cA\) scales all n rows, so \(\det(cA) = c^n \det A\). Determinants do not respect addition. With \(A = B = I_2\), \(\det(A + B) = \det(2I_2) = 4\), while \(\det A + \det B = 2\).

Linear in one row at a time

Although \(\det(A + B)\) is not \(\det A + \det B\), the determinant is linear in any single row when the other rows are held fixed. If row \(i\) is a sum \(\mathbf{u} + \mathbf{v}\), the determinant splits into the determinant with \(\mathbf{u}\) in row \(i\) plus the one with \(\mathbf{v}\) in row \(i\); a scalar in that row factors out. This follows from expanding along row \(i\), since its cofactors do not involve row \(i\) itself. In fact this row-by-row linearity, together with the sign change under swaps and \(\det I = 1\), pins down the determinant completely.

Worked example: properties without entries

Let \(A\) and \(B\) be 3 x 3 matrices with \(\det A = 4\) and \(\det B = -2\). 1. \(\det(AB) = 4 \cdot (-2) = -8\). 2. \(\det(2A) = 2^3 \cdot 4 = 32\), not 8. 3. \(\det(A^{-1}) = \frac{1}{4}\). 4. \(\det(A^T B^{-1}) = 4 \cdot \left(-\frac{1}{2}\right) = -2\). 5. \(\det(B^3) = (-2)^3 = -8\). 6. Swapping two rows of \(A\) gives a matrix with determinant -4. Each answer came from a rule, not from any entries.

Common mistakes to avoid

Forgetting the sign change for each row swap is the commonest error; count swaps as you go. The second is writing \(\det(cA) = c \det A\); the factor is \(c^n\) because every row is scaled. The third is assuming \(\det(A + B) = \det A + \det B\), which fails even for the identity. The fourth is scaling a row to make a pivot 1 and then forgetting to account for that factor at the end. Finally, a determinant of 0 means "not invertible"; it never means the matrix is the zero matrix.

Key takeaways

Row replacement leaves the determinant unchanged, a row swap changes its sign, and scaling a row by \(k\) multiplies it by \(k\). So \(\det A = (-1)^r\) times the product of the diagonal of an echelon form, found in about \(\frac{2}{3}n^3\) operations. A is invertible exactly when \(\det A \ne 0\). \(\det A^T = \det A\), so column rules mirror row rules. \(\det(AB) = \det A \det B\), \(\det(A^{-1}) = 1/\det A\) and \(\det(cA) = c^n\det A\), but the determinant is not additive. It is linear in each row separately.

Flashcards

Flashcard

Row replacement

Determinant unchanged

Add a multiple of one row to another.

Test your understanding

Question 1 of 5

A 3 x 3 matrix A has det A = 5. Swap two of its rows, then multiply one row by 3. What is the new determinant?

Question 2 of 5

A is a 3 x 3 matrix with det A = -2. What is det(3A)?

Question 3 of 5

A and B are n x n with det A = 4 and det B = -3. What is det(A⁻¹Bᵀ)?

Question 4 of 5

Which row operation never changes the determinant of a square matrix?

Question 5 of 5

Row reducing A with one row swap and some replacements gives an echelon form with diagonal 2, 3 and -1. What is det A?

← Previous topicNext topic →