For a given square matrix, a
nonzero
vector that changes length, but not direction, when multiplied by the matrix. The computation of
Principal Components
for a set of
Variable
s uses the eigenvectors of the variables'
Covariance
or
Correlation
matrix.