An eigen decomposition is a matrix decomposition of a square matrix in terms of its eigenvalues
and eigenvectors. It is also called "matrix
diagonalization." However, this moniker is less than optimal, since the
process being described is really the decomposition of a matrix into a product of
three other matrices, only one of which is diagonal, and also because all other standard
types of matrix decomposition use the term
"decomposition" in their names, e.g., Cholesky
decomposition, Hessenberg decomposition,
and so on. As a result, the decomposition of a matrix into matrices composed of its
eigenvectors and eigenvalues is called eigen decomposition in this work.
Assume
is a
matrix over a field
and has a vector basis of
eigenvectors
,
, ...,
, with corresponding eigenvalues
,
, ...,
. The eigenvalues
need not be distinct. The existence of such a vector
basis, rather than merely knowing the eigenvalues,
is the condition for
to be a diagonalizable
matrix over
(Axler 2024, p. 165). The eigenvectors can be
denoted
|
(1)
|
Define the matrices composed of eigenvectors
|
(2)
| |||
|
(3)
|
and eigenvalues
|
(4)
|
where
is a diagonal matrix. Then
|
(5)
| |||
|
(6)
| |||
|
(7)
| |||
|
(8)
| |||
|
(9)
| |||
|
(10)
|
giving the amazing decomposition of into a similarity
transformation involving
and
,
|
(11)
|
The fact that this decomposition is possible for a square matrix
whenever
is nonsingular is known in this work as the
eigen decomposition theorem.
Furthermore, squaring both sides of equation (11) gives
|
(12)
| |||
|
(13)
| |||
|
(14)
|
By induction, it follows that for general positive integer powers,
|
(15)
|
If
is a nonsingular matrix, its inverse is
|
(16)
| |||
|
(17)
|
where the inverse of the diagonal matrix is trivially given by
|
(18)
|
Under this nonsingular matrix hypothesis, equation (15) also holds for negative .
For real matrices or complex matrices, a further result follows from the definition of the matrix exponential
|
(19)
| |||
|
(20)
| |||
|
(21)
| |||
|
(22)
|
This is true since
is a diagonal matrix and
|
(23)
| |||
|
(24)
| |||
|
(25)
| |||
|
(26)
|
so
can be found using
.