Skip to content
Tutorials Center
My learning

Topics

Lesson 7 of 39 · Hilbert Space and Dirac Notation

Hermitian Operators and the Spectral Theorem

Why observables are Hermitian: real eigenvalues, orthogonal eigenvectors, degeneracy, unitary operators and complete sets of commuting observables.

Measurement results are real numbers, and the possible results of measuring an observable are the eigenvalues of an operator. For this to make sense, the operator needs real eigenvalues and enough eigenvectors to expand any state. Hermitian operators have exactly these properties, which is why they represent observables.

This lesson proves the key facts, introduces unitary operators (which represent symmetries and time evolution), and explains how commuting observables label degenerate levels.

Hermitian operators

An operator A is Hermitian (self-adjoint, in finite dimensions) when A† = A. In a basis this means Aₙₘ = Aₘₙ*: the matrix equals its conjugate transpose.

Three properties follow immediately.

  1. Real eigenvalues. If A|a⟩ = a|a⟩ then ⟨a|A|a⟩ = a⟨a|a⟩. Taking the complex conjugate and using A† = A gives a* = a.
  2. Orthogonal eigenvectors. If A|a⟩ = a|a⟩ and A|b⟩ = b|b⟩ with a ≠ b, then (a − b)⟨a|b⟩ = 0, so ⟨a|b⟩ = 0.
  3. Real expectation values. ⟨ψ|A|ψ⟩* = ⟨ψ|A†|ψ⟩ = ⟨ψ|A|ψ⟩.

The spectral theorem states that a Hermitian operator on a finite-dimensional Hilbert space has an orthonormal basis of eigenvectors, so A = Σ aₙ|n⟩⟨n|. (With a repeated eigenvalue the projector becomes a sum over the eigenspace.)

Python Properties of a Hermitian matrix

Press Run (or Ctrl + Enter) to see the result here.

Program input · what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

The last line shows a general fact: an expectation value is a weighted average of eigenvalues, so it lies between the smallest and largest eigenvalue.

In infinite dimensions, "Hermitian" is not quite enough. An operator such as momentum on a half-line satisfies ⟨φ|Aψ⟩ = ⟨Aφ|ψ⟩ formally but fails to have a full set of eigenvectors. The proper condition is self-adjointness, which also fixes the allowed domain of the operator. For the systems in this track the distinction does not change any calculation, but it matters in rigorous treatments of boundary conditions.

Unitary operators

An operator U is unitary if U†U = UU† = 1. Unitary operators preserve inner products: ⟨Uφ|Uψ⟩ = ⟨φ|ψ⟩. They represent changes of basis, symmetry transformations (rotations, translations) and time evolution exp(−iHt/ħ).

A key link: if A is Hermitian then U = e^{iA} is unitary, and every unitary can be written this way. Hermitian operators are the generators of unitary transformations. Momentum generates translations, angular momentum generates rotations and the Hamiltonian generates time evolution.

Eigenvalues of a unitary operator all have magnitude 1: they are phases e^{iθ}.

Degeneracy and commuting observables

Sometimes an eigenvalue is repeated. Then specifying the eigenvalue does not fix the state: you must say which vector inside the eigenspace. The remedy is to find another observable B that commutes with A.

If [A, B] = 0, you can choose a basis in which both are diagonal. States are then labeled by pairs (a, b). A complete set of commuting observables (CSCO) is a set whose joint eigenvalues label every basis state uniquely. For hydrogen (ignoring spin) the set {H, L², L_z} labels n, ℓ, m.

Python Lifting degeneracy with a commuting operator

Press Run (or Ctrl + Enter) to see the result here.

Program input · what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

The two states with S_z = 0 are distinguished by the swap eigenvalue: +1 (symmetric, the triplet state) and −1 (antisymmetric, the singlet). This is a first glimpse of the exchange symmetry that organizes Chapter 6.

Functions and unitary change of basis

If {|n⟩} and {|m'⟩} are two orthonormal bases, the matrix Uₘₙ = ⟨m'|n⟩ is unitary. Matrices of an operator in the two bases are related by A' = U A U†. Eigenvalues, traces and determinants do not change: they are properties of the operator, not of the basis.

Python Basis change leaves the spectrum unchanged

Press Run (or Ctrl + Enter) to see the result here.

Program input · what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

Check your understanding

Quick check · 4 questions

of 4 answered You scored / 4 · perfect!

  1. 1. Why are eigenvalues of a Hermitian operator real?

    The conjugate of ⟨a|A|a⟩ equals itself when A is Hermitian, so the eigenvalue equals its own conjugate.
  2. 2. Eigenvectors of a Hermitian operator belonging to different eigenvalues are

    From (a − b)⟨a|b⟩ = 0 and a ≠ b, the inner product must vanish.
  3. 3. All eigenvalues of a unitary operator have the form

    Unitarity preserves norms, so each eigenvalue has magnitude 1.
  4. 4. If two observables commute, what can you do?

    Commuting Hermitian operators can be diagonalized together, so both quantities can have definite values.

Summary

  • Hermitian operators have real eigenvalues, orthogonal eigenvectors and a spectral decomposition.
  • Unitary operators preserve inner products and represent symmetries and dynamics; e^{iA} is unitary for Hermitian A.
  • Degeneracy is resolved by commuting observables; a CSCO labels every state uniquely.
  • Eigenvalues, trace and determinant are basis independent.

Next: Commutators and the Uncertainty Principle