Determine the square of the length of each vector using dot product
T T T
LSq1 x1 x1 11 LSq2 x2 x2 11 LSq3 x3 x3 21
1 0.636 0.197
X X 0.636 0.855
T
1
0.197 0.855 1
Why does this not work? Because the example has n distinct eigenvalues and therefore its
eigenvectors form a basis. However, only a symmetric matrix provides an orthonormal basis.
Since the matrix in question is not symmetric, the eigenvectors are not orthogonal.
T but
xn1 xn1 1
7.3 0.2 3.7
A = 11.5 1.0 5.5 and T not 0 and
xn1 xn2 0.6364
17.7 1.8 9.3 T not 0 etc.
xn1 xn3 0.1974
So keep this in mind - symmetric matrices form an orthonormal basis so it is easy to find the inverse
of X (it is simply XT if each eigenvector is normalized). But this only works if the matrix is symmetric,
which is NOT the only way to get n distinct eigenvalues (and thus form an eigen basis).