Let $A \in \mathbb{C}^{n\times n}$ be Hermitian Matrix. Prove all eigenvalues of $A$ are real and that eigenvectors corresponding to different eigenvalues are orthogonal
Aryan Yadav, AIR 31, 2025
let $\lambda$ be an Eigen value of A & $x$ be corresponding Eigen vector
$\Rightarrow Ax = \lambda x$ —— ①
let $A^\theta$ denote conjugate transpose
Multiply ① by $x^\theta$
$\Rightarrow x^\theta A x = \lambda x^\theta x$ - ②
Take conjugate transpose of both sides
$\Rightarrow x^\theta A^\theta (x^\theta)^\theta = \bar{\lambda}\, x^\theta (x^\theta)^\theta$
$A^\theta = A$ (Hermitian) & $(x^\theta)^\theta = x$
$\Rightarrow x^\theta A x = \bar{\lambda}\, x^\theta x$ -③
② - ③ $\Rightarrow (\lambda - \bar{\lambda})\, x^\theta x = 0$
But if $x^\theta x$ (norm) $= 0 \Rightarrow x = 0$
But $x \neq 0$ as it's Eigen vector
Hence $\lambda - \bar{\lambda} = 0$
$\Rightarrow \lambda \in \mathbb{R}$
let $\lambda_1, \lambda_2$ be 2 distinct Eigen-values of A & $x_1, x_2$ be corresponding Eigen vector
$A x_1 = \lambda_1 x_1$ -① $A x_2 = \lambda_2 x_2$ -②
Take conjugate transpose of ①
$\Rightarrow x_1^\theta A^\theta = \bar{\lambda_1} x_1^\theta$
$\Rightarrow x_1^\theta A = \lambda_1 x_1^\theta$ ($A = A^\theta$, $\lambda_1 \in \mathbb{R}$ shown before)
Post multiply by $x_2$
$\Rightarrow x_1^\theta A x_2 = \lambda_1 x_1^\theta x_2$
$\Rightarrow x_1^\theta (\lambda_2 x_2) = \lambda_1 x_1^\theta x_2$ (From ②)
$\Rightarrow (\lambda_2 - \lambda_1)\, x_1^\theta x_2 = 0$
But $\lambda_1 \neq \lambda_2$ (distinct)
$\Rightarrow x_1^\theta x_2 = 0$
$\Rightarrow$ Orthogonal