
Singular value decomposition 21
4 Eige nvecto rs (as rows )
5 -0.6 4092380 -0.71641836 0.27561100
6 -0.2 2689155 -0.16619583 -0 .95963 492
7 0 .73330 547 -0.67758666 -0.0 560304 08
8 Orth ogonal ity
9 1.000000 0 0.0 000000 0 5.0306981 e -017
10 0.0 0000000 1.000 0000 4.16 33363 e -017
11 5.0306981 e -017 4.1633 363 e -017 1.0000 000
Notice that, due to rounding errors, U
⊤
U has finite — but very small — off-
diagonal e lements. The eigenvalues are all positive, so the covariance matrix
calculated by the program is positive definite.
In Python, the eigenvalues a nd e igenvectors of real symmetric (or complex
Hermitian) matric es can be calculated with numpy. ...