Hi

I'm generating eigen vectors using eigen solver and compared results with a Jacobi implementation on VBA.

I have noticed the Eigen values are the same but the eigenvectors differ by negative sign.

I have tried to change RowMajor, ColMajor parameters when constructing the matrixes to see if this is affecting the sign but it's not worked.

I'm quite new to PCA, so not sure if the negative sign is important or can be ignored. Alternatively what am i missing?

Eric

indeed, the sign does not matter, likewise be aware that when multiple eigenvalues are the same, then their respective eigenvectors are not uniquely defined.