<aside> ♻️ Eigenvalues and eigenvectors: For a linear operator $\hat A$ the eigenvalue equation is defined as

$$ \hat A |u \rang = \lambda | u \rang $$

where $\lambda$ and $|u\rang$ are the eigenvalues and eigenvectors associated to $\hat A$ repsectively, where $|u\rang \ne |0\rang$

</aside>

πŸ—’οΈ Note: each distinct eigenvalue gives an eigenvector. If two eigenvectors are the same then it is a degeneracy

Eigenvalues and eigenvectors of Hermitian operators

πŸ§”β€β™€οΈ Theorem: Let $\hat A$ be a Hermitian operators ($\hat A^\dagger=\hat A$). The eigenvalues for this operator are real and its eigenvectors are orthogonal

πŸ’« Proof: The eigenvalue equation for an operator is given by

$$ \hat A|u_i\rang =\lambda _i |u_i \rang $$

Consider the matrix element of $\hat A$ with respect to two eigenvectors

$$ \lang u_j |\hat A |u_k \rang = \lambda _k \lang u_j|u_k \rang $$

Using Hermitian property

$$ \lang u_j |\hat A|u_k\rang =\lang u_j |\hat A^\dagger |u_k \rang = \overline {\lang u_k |\hat A|u_j \rang} = \overline \lambda _j \lang u_j |u_k \rang $$

Equating the two expressions we obtain

$$ (\lambda _k-\overline \lambda_j)\lang u_j|u_k \rang =0 $$

Eigenvalues and eigenvectors of unitary operators

πŸ§”β€β™€οΈ Theorem: A unitary operator $\hat U$ which satisfies $\hat U \hat U^\dagger = \hat {\bold 1}$ has eigenvalue equation

$$ \hat U |u_j \rang = \lambda _j |u_j \rang $$

which satisfy the following equations

  1. $|\lambda_j|=1$ with $\lambda_j =e^{i\theta_j}$ where $\theta_j\in \R$
  2. If $\lambda_j \ne \overline \lambda_j$ then $\lang u_k | u_j \rang =0$
  3. The eigenvalues of $\hat U$ can be chosen to be an orthonormal basis for $V^N$

πŸ’« Proof: the eigenvalue equation of $\hat U$ and $\hat U^\dagger$ are

$$ \hat U|u_j \rang = \lambda _j |u_j \rang \qquad \lang u_k | \hat U^\dagger = \overline\lambda _k \lang u_k | $$

Combining these equations we can write

$$ \begin{aligned} \hat U |u_j \rang &= \lambda j |u_j \rang \\ \lang u_k|\underbrace{\hat U^\dagger\hat U }{\hat{\bold 1}}|u_j \rang &= \lambda j \underbrace{\lang u_k| \hat U^\dagger}{\overline\lambda _k \lang u_k |} |u_j \rang \\ \lang u_k|u_j \rang &=\lambda_j \overline\lambda_k\lang u_k|u_j \rang \end{aligned} $$

  1. For $\lambda_j\overline \lambda_j=1$ we have $|\lambda_j|=({\lambda_j\overline \lambda_j})^{1/2}=\sqrt 1 =1$ thus $\lambda_j=e^{i\theta_j}$ where $\theta_j\in \R$
  2. For $\lambda_j \ne \overline\lambda_k$ then $\lang u_k | u_j \rang =0$
  3. If $\lang u_k | u_j \rang =0$ then $u_k$ and $u_j$ are orthogonal so $\lambda_j$ and $\lambda_k$ can be used to be an orthonormal basis

πŸ—’οΈ Note: statement 3 does not hold when $N\to \infin$ because some infinite vectors can’t be normalised

Spectral representation and diagonalising operators

If $\hat A$ is Hermitian or Unitary, then the set of its eigenvectors $\{|u_j\rang \}^N_{j=1}$ are an orthonormal basis, therefore we can write a completeness relation $\hat {\bold 1}=\sum^N_{j=1} |u_j \rang \lang u_j |$.

πŸ’Ž Conclusion: this is known as a spectral representation of $\hat A$

πŸ’ƒ Example: diagonalisation of the Pauli operators