Canonical forms of matrices
Canonical forms of matrices are standard representatives for equivalence or similarity classes of matrices. A canonical form provides a unique, simplified representation that reveals key structural properties.
Important canonical forms include:
- Rank canonical form: $\begin{pmatrix} I_k & 0 \\ 0 & 0 \end{pmatrix}$ for matrix equivalence
- Row echelon form / RREF: for row equivalence
- Jordan canonical form: for similarity (every square matrix over $\mathbb{C}$ is similar to a Jordan matrix)
- Rational canonical form: for similarity (works over any field)
- Hessenberg form: nearly upper triangular, used as intermediate step
- Diagonal form: when a matrix is diagonalizable
- Schur form: upper triangular via unitary similarity
- SVD: $U\Sigma V^*$ via two unitary transformations
Each canonical form is adapted to a particular equivalence relation and reveals different aspects of the matrix structure.