Visual Tools
Calculators
Tables
Mathematical Keyboard
Converters
Other Tools

How to Identify the Type of a Matrix

Most matrix types are defined by one equation: AT=AA^T = A, ATA=IA^TA = I, A2=AA^2 = A. Test the cheap properties first (shape, zeros, transpose), then the ones that need a product.

Recognize the task

"Is this matrix symmetric, orthogonal or idempotent?", "classify the matrix", or a step that only works for one type, such as the shortcut for triangular determinants. Typical inputs: (21013−10−14)\begin{pmatrix} 2 & 1 & 0 \\ 1 & 3 & -1 \\ 0 & -1 & 4 \end{pmatrix}, a permutation matrix, a projection. Test: is it square? Almost every type requires it.
Every type is defined in the types of matrices lesson; its summary table lists the defining equations.

Steps

StepOn A=(21013−10−14)A = \begin{pmatrix} 2 & 1 & 0 \\ 1 & 3 & -1 \\ 0 & -1 & 4 \end{pmatrix}
1Check the shape. Every type below needs a square (n×nn \times n) matrix.3×33 \times 3: square
2Look at the zero pattern. Zeros off the diagonal: diagonal. Zeros below it: upper triangular. Zeros above it: lower triangular.zeros only at (1,3)(1,3) and (3,1)(3,1): not diagonal or triangular
3Compare A with its transpose. AT=AA^T = A: symmetric. AT=−AA^T = -A: skew-symmetric.Transpose ↗a12=a21=1a_{12} = a_{21} = 1, a13=a31=0a_{13} = a_{31} = 0, a23=a32=−1a_{23} = a_{32} = -1: symmetric
4Test the product equations. ATA=IA^TA = I: orthogonal. A2=AA^2 = A: idempotent. A2=IA^2 = I: involutory. Ak=0A^k = 0: nilpotent.column 1 has length 5\sqrt{5}: not orthogonal; (A2)11=5(A^2)_{11} = 5: not idempotent or involutory; trace 9≠09 \neq 0: not nilpotent
5Check whether it is singular. det⁡(A)=0\det(A) = 0 means singular; otherwise the matrix is invertible.det⁡(A)=18≠0\det(A) = 18 \neq 0: symmetric and invertible

Worked examples

Example 1 · skew-symmetric

Aᵀ = −A
K=(02−1−2031−30)K = \begin{pmatrix} 0 & 2 & -1 \\ -2 & 0 & 3 \\ 1 & -3 & 0 \end{pmatrix}
zero diagonal, mirrored entries change sign
KT=(0−2120−3−130)=−KK^T = \begin{pmatrix} 0 & -2 & 1 \\ 2 & 0 & -3 \\ -1 & 3 & 0 \end{pmatrix} = -K
skew-symmetric, and det⁡(K)=0\det(K) = 0
odd size forces determinant 00

Example 2 · a permutation matrix

orthogonal
P=(010001100)P = \begin{pmatrix} 0 & 1 & 0 \\ 0 & 0 & 1 \\ 1 & 0 & 0 \end{pmatrix}
one 11 in each row and column
PTP=IP^TP = I
the columns are orthonormal
orthogonal, with P−1=PTP^{-1} = P^T and det⁡(P)=1\det(P) = 1

Example 3 · a projection

A² = A
E=(1100)E = \begin{pmatrix} 1 & 1 \\ 0 & 0 \end{pmatrix}: E2=(1100)=EE^2 = \begin{pmatrix} 1 & 1 \\ 0 & 0 \end{pmatrix} = E
idempotent, upper triangular, singular; tr⁡(E)=1=rank⁡(E)\operatorname{tr}(E) = 1 = \operatorname{rank}(E)
for idempotents, trace equals rank
Generate every type in the matrix types visualizer, and test orthogonality in the orthogonal matrices visualizer.

See it happen

Each type is a symmetry or a habit of the matrix. Symmetric matrices are mirror images across the diagonal. Orthogonal matrices are rotations and reflections; they keep lengths, which is what QTQ=IQ^TQ = I says. Idempotent matrices are projections: projecting twice changes nothing. Nilpotent matrices push vectors down a chain until they vanish.

Common mistakes

Calling a non-square matrix symmetric

✗ (122100)\begin{pmatrix} 1 & 2 \\ 2 & 1 \\ 0 & 0 \end{pmatrix} is symmetric
✓ its transpose is 2×32 \times 3, so AT≠AA^T \neq A; symmetric matrices are square

Orthogonal columns are not enough

✗ (2003)\begin{pmatrix} 2 & 0 \\ 0 & 3 \end{pmatrix} is orthogonal
✓ QTQ=(4009)≠IQ^TQ = \begin{pmatrix} 4 & 0 \\ 0 & 9 \end{pmatrix} \neq I: the columns must also have length 11

Confusing idempotent and involutory

✗ (0110)2=I\begin{pmatrix} 0 & 1 \\ 1 & 0 \end{pmatrix}^2 = I, so it is idempotent
✓ A2=IA^2 = I is involutory; idempotent means A2=AA^2 = A

A skew-symmetric matrix with a nonzero diagonal

✗ (12−21)\begin{pmatrix} 1 & 2 \\ -2 & 1 \end{pmatrix} is skew-symmetric
✓ aii=−aiia_{ii} = -a_{ii} forces zeros on the diagonal