Lean, informalized

Nielsen & Chuang · Chapter 2, Introduction to quantum mechanics · p. 79

Exercise 2.49 - The polar decomposition of a normal operator in the outer product representation

A normal operator factors into its moduli and phases.

Hover over a blue or green term in the text, or a node in the graph, to find it in the other. Select it to read its declaration both in Lean and in words.

The Exercise

An operator AA on a finite-dimensional complex Hilbert space is normal when it commutes with its adjoint, AA†=A†AAA^\dagger = A^\dagger A. The exercise asks for the polar decomposition of a normal operator in the outer product representation. The polar decomposition writes AA as a unitary UU times a positive operator JJ, and the exercise wants both factors as sums of outer products |i⟩⟨i||i\rangle\langle i| over an orthonormal basis {|i⟩}\{|i\rangle\} in which AA itself reads ∑iλi|i⟩⟨i|\sum_i \lambda_i |i\rangle\langle i|.

Diagonal Operators

Let EE be a Hilbert space of finite dimension dd, over the complex or the real numbers. The diagonal operator DD with entries wiw_i in an orthonormal basis {|i⟩}\{|i\rangle\} of EE is the operator on EE that sums the projectors onto the basis vectors, each weighted by its entry: D=∑iwi|i⟩⟨i|.D = \sum_i w_i\, |i\rangle\langle i| .(1)✦

The phase ph⁡(z)\operatorname{ph}(z) of a number zz is zz divided by its modulus, and 11 when z=0z = 0: ph⁡(z)={z/|z|,z≠0,1,z=0.\operatorname{ph}(z) = \begin{cases} z/|z|, & z \neq 0, \\ 1, & z = 0 . \end{cases}(2)✦

The right singular vectors |v0⟩,…,|vd−1⟩|v_0\rangle, \dots, |v_{d-1}\rangle of an operator TT on EE are an orthonormal basis of EE made of eigenvectors of the Hermitian operator T†TT^\dagger T.✦ The absolute value |T||T| of TT is the operator on EE diagonal in that basis, with the singular values s0,…,sd−1s_0, \dots, s_{d-1} of TT as its entries: |T|=∑i=0d−1si|vi⟩⟨vi|.|T| = \sum_{i=0}^{d-1} s_i\, |v_i\rangle\langle v_i| .(3)✦

The Claim

Theorem 1. Let AA be a normal operator on a complex Hilbert space EE of finite dimension dd. Then there are an orthonormal basis |0⟩,…,|d−1⟩|0\rangle, \dots, |d-1\rangle of EE and complex numbers λ0,…,λd−1\lambda_0, \dots, \lambda_{d-1} with A=∑i=0d−1λi|i⟩⟨i|.A = \sum_{i=0}^{d-1} \lambda_i\, |i\rangle\langle i| .(4)✦ Let now A=∑iλi|i⟩⟨i|A = \sum_i \lambda_i |i\rangle\langle i| be diagonal in an orthonormal basis {|i⟩}\{|i\rangle\} of a finite-dimensional real or complex Hilbert space, and let J=∑i|λi||i⟩⟨i|J = \sum_i |\lambda_i|\, |i\rangle\langle i| and U=∑iph⁡(λi)|i⟩⟨i|U = \sum_i \operatorname{ph}(\lambda_i)\, |i\rangle\langle i|. Then AA factors in both orders, A=JU,A = JU ,(5)✦ A=UJ,A = UJ ,(6)✦ and JJ is the absolute value of AA: J=|A|.J = |A| .(7)✦

For a normal AA, the basis and the numbers λi\lambda_i of (4) turn (5) and (6) into the decomposition the exercise asks for.

What Is Checked

The claim is stated over the diagonal operator (1) in an orthonormal basis, the phase (2) of a number, and the absolute value (3) of an operator.

For any orthonormal basis and any numbers λi\lambda_i, the factor of moduli is positive and the factor of phases is unitary: J≥0,J \ge 0 ,(8)✦ U†U=UU†=I.U^\dagger U = U U^\dagger = I .(9)✦