deepdream of
          a sidewalk
Show Question
Math and science::Algebra

Eigenvector decomposition. Equivalent forms.

The following is true, for a diagonizable matrix A:

S1AS=Λ

The above is equivalent to the more commonly seen:

{{A=SΛS1

And another equivalent is:

AS=SΛ

Adding brackets to emphasize a multiplication order can help make the expressions more intuitive:

S1(A(S))=Λ

involves taking A times it's eigenvectors, which will produce scaled versions of these eigenvectors, and then expressing these scaled columns in the basis of the eigenvectors, which of course will just be a single number (an eigenvalue) for each column.

The expression:

AS=SΛ

Is simply Av=λv, but for all eigenvectors at once. As Λ is a diagonal matrix, we can say:

AS=SΛ=ΛS

which is slighly closer to the form Av=λv.