I have a matrix A
A = [ 124.6,95.3,42.7 ; 95.3,55.33,2.74 ; 42.7,2.74,33.33 ]
The eigenvalues and vectors:
[V,D] =
To compute all dot products between columns of V
:
M = squeeze(sum(bsxfun(@times, conj(V), permute(V, [1 3 2]))));
The columns of V
(eigenvectors) will be orthogonal if the above M
is diagonal.
To show orthogonality
>> V'*V-eye(size(V))
ans =
1.0e-15 *
0.2220 0.1110 0.2498
0.1110 -0.4441 0.1388
0.2498 0.1388 0.4441
To show that the definition of the eigendecomposition is satisfied,
>> A*V - V*D
ans =
1.0e-13 *
0.4086 0.0400 0.8527
0.3908 0.0355 0.5684
0.1954 0.0355 0
The results won't be exactly zero, because digital computers don't do exact math, but you can see that they're pretty close.