10.3.7 Theorem ( Spectral Decomposition ):
Let A be a n × n real symmetric matrix . Then there exist real numbers
and real matrices   such that
the  following holds :
(i)    Each
is an eigenvalue of A.
(ii)   Each is an orthogonal projection matrix, of rank
1, i.e.,
        is real symmetric, = P and rank
( ) = 1.

                                                                                                                                               

Proof of Theorem 10.3.7

     Back

By theorem 10.3.4, A has n-real eigenvalues, , and an orthonormal basis 
, of  , where . Further, if
           
then
           
Define
            
Then, each is a real symmetric matrix with
          
That follows from the fact that P is orthogonal.
Clearly, rank ( ) = 1. Finally,