Today I am reading an article: Eigenvalues and sums of Hermitian matrices there is an exercise copy from that article:

Exercise 1 Suppose that the eigenvalues $\lambda_1(A)>\cdots>\lambda_n(A)$ of an $n\times n$ Hermitian matrix are distinct. Show that the associated eigenbasis $u_1(A),\cdots,u_n(A)$ is unique up to rotating each individual eigenvector $u_j(A)$ by a complex phase $e^{i\theta_j}$. In particular, the spectral projections $P_j(A):=u_j(A)^*u_j(A)$ are unique. what happens when there is eigenvalue multiplicity?

Assume $A$ is the matrix of self-adjoint operator $\mathcal{A}$ under some property basis of $V_{\Bbb{C}}$.
I don't know why eigenbasis $u_1(A),\cdots,u_n(A)$ is unique up to rotating. If we multiply a complex number $c_k \not=0$ respectively, then $c_1u_1(A),\cdots,c_nu_n(A)$ is also a basis of $V_{\Bbb{C}}$ consist of $\mathcal{A}$'s eigenvectors,under this basis, $\mathcal{A}$ correspond to a diagonal matrix consist of eigenvalues: $\text{diag}\{\lambda_1,\cdots,\lambda_n\}$. I think it is unique up to multiply a complex number $c_k$ respectively. what's wrong?


Since this is a self-adjoint operator, there is an orthonormal basis of eigenvectors, and that's what they're talking about. If you multiply basis elements by arbitrary nonzero complex numbers you still have a basis of eigenvectors, but it's no longer orthonormal.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.