Mathwords logoMathwords

Diagonally Dominant Matrix — Definition, Formula & Example

A diagonally dominant matrix is a square matrix in which each diagonal entry, ignoring sign, is at least as large as all the other entries in its row added together, also ignoring sign. If the diagonal entry is strictly larger in every row, the matrix is strictly diagonally dominant.

An n×nn \times n matrix A=[aij]A = [a_{ij}] is (row) diagonally dominant if ∣aii∣≥∑j≠i∣aij∣|a_{ii}| \ge \sum_{j \ne i} |a_{ij}| for every i=1,…,ni = 1, \dots, n. It is strictly diagonally dominant if the inequality is strict for every ii. Column diagonal dominance is defined the same way using sums down columns instead of across rows, so AA is column diagonally dominant exactly when A⊤A^\top is row diagonally dominant.

Key Formula

∣aii∣  ≥  ∑j=1j≠in∣aij∣for all i=1,…,n|a_{ii}| \;\ge\; \sum_{\substack{j=1 \\ j \ne i}}^{n} |a_{ij}| \quad \text{for all } i = 1, \dots, n
Where:
  • aiia_{ii} = Diagonal entry in row i
  • aija_{ij} = Entry in row i, column j, where j is not equal to i
  • nn = Number of rows and columns of the square matrix

Worked Example

Problem: Decide whether A=[10−2318−42−59]A = \begin{bmatrix} 10 & -2 & 3 \\ 1 & 8 & -4 \\ 2 & -5 & 9 \end{bmatrix} is strictly diagonally dominant. Then state what this tells you about whether AA is invertible.
Step 1: For each row, add the absolute values of the off-diagonal entries.
Row 1: ∣−2∣+∣3∣=5,Row 2: ∣1∣+∣−4∣=5,Row 3: ∣2∣+∣−5∣=7\text{Row 1: } |-2| + |3| = 5, \quad \text{Row 2: } |1| + |-4| = 5, \quad \text{Row 3: } |2| + |-5| = 7
Step 2: Compare each diagonal entry, in absolute value, with its row sum. All three inequalities are strict, so AA is strictly diagonally dominant.
10>5,8>5,9>710 > 5, \qquad 8 > 5, \qquad 9 > 7
Step 3: The Levy–Desplanques theorem says every strictly diagonally dominant matrix is invertible. You can confirm this with a cofactor expansion along row 1. The determinant is nonzero, as the theorem predicts.
det⁡A=10(72−20)+2(9+8)+3(−5−16)=520+34−63=491\det A = 10(72 - 20) + 2(9 + 8) + 3(-5 - 16) = 520 + 34 - 63 = 491
Answer: AA is strictly diagonally dominant, so it is invertible. Its determinant is 491≠0491 \ne 0, which agrees with the theorem.

Why It Matters

In a numerical analysis course, strict diagonal dominance of the coefficient matrix guarantees that the Jacobi and Gauss–Seidel iterative methods converge for every starting guess. It also guarantees that Gaussian elimination can run without row swaps. Engineers use these facts when solving large systems from heat-flow and circuit models, where diagonal dominance often holds naturally.

Common Mistakes

Mistake: Assuming any diagonally dominant matrix is invertible, even when some rows have equality instead of a strict inequality.
Correction: Weak dominance alone does not guarantee invertibility. For example, [1111]\begin{bmatrix} 1 & 1 \\ 1 & 1 \end{bmatrix} satisfies ∣1∣≥∣1∣|1| \ge |1| in both rows, but its determinant is 00. Check that every row has a strict inequality before you apply the theorem.

Related Terms

  • Matrix — The grid of numbers this property describes
  • Coefficient Matrix — The matrix tested for dominance before iterative solving
  • Cramer's Rule — Requires the nonzero determinant that strict dominance guarantees
  • Column of a Matrix — Column sums define the column version of dominance
  • Augmented Matrix — Row-reduced without swaps when its coefficient matrix is strictly dominant
  • Square Matrix — Diagonal dominance is defined only for this kind of matrix