Let
Find the spectral decomposition of .
The spectral decomposition of is a way to write as
such that
Intuitively, 's are projection matrices whose ranges are orthogonal to each other and span the whole space.
In practice, we may first find the eigenvalues and an orthonormal basis of . Then
where are the distinct eigenvalues of and is the sum of for all .
According to Find an orthonormal basis composed of eigenvectors, we already have , , . Thus, we may set , and with .
The corresponding eigenvectors are , , , and . As and both correspond to the eigenvalue , we will add them together to form a projection matrix onto . In summary, we have
where
and
which is the spectral decomposition of .
This note can be found at Course website > Learning resources.