Skip to content
Tutorials Center
My learning

Topics

Lesson 10 of 39 · Time-Independent Perturbation Theory

Nondegenerate Perturbation Theory

Derive Rayleigh–Schrödinger perturbation theory for non-degenerate levels, apply it to a two-level system and a shifted oscillator, and test it in code.

Suppose you can solve H₀|n⁰⟩ = Eₙ⁰|n⁰⟩ exactly, and you want the levels of H = H₀ + λV, where λ is a small number that bookkeeps the strength of the perturbation V. Rayleigh–Schrödinger perturbation theory expands the energy and the state in powers of λ. The first term costs one integral; the second costs a sum over all other states. This lesson derives the formulas and tests them against exact answers.

Setting up the expansion

Write

Text
Eₙ = E⁰ₙ + λE¹ₙ + λ²E²ₙ + ...
|n⟩ = |n⁰⟩ + λ|n¹⟩ + λ²|n²⟩ + ...

and insert them in (H₀ + λV)|n⟩ = Eₙ|n⟩. Collect equal powers of λ.

  • Order λ⁰: H₀|n⁰⟩ = E⁰ₙ|n⁰⟩ (the solved problem).
  • Order λ¹: H₀|n¹⟩ + V|n⁰⟩ = E⁰ₙ|n¹⟩ + E¹ₙ|n⁰⟩.

Take the inner product of the order-λ¹ equation with ⟨n⁰|. The terms with |n¹⟩ cancel because H₀ is Hermitian, leaving the first-order energy. Taking the inner product with ⟨m⁰| (m ≠ n) gives the coefficient of |m⁰⟩ in |n¹⟩. We use the convention ⟨n⁰|n^k⟩ = 0 for k ≥ 1, which fixes the normalization to first order.

The working formulas

With Vₘₙ = ⟨m⁰|V|n⁰⟩:

Text
E¹ₙ = Vₙₙ
|n¹⟩ = Σ_{m≠n} [Vₘₙ / (E⁰ₙ − E⁰ₘ)] |m⁰⟩
E²ₙ = Σ_{m≠n} |Vₘₙ|² / (E⁰ₙ − E⁰ₘ)

Several facts follow directly.

  1. First order is an expectation value. The energy shift is the average of V in the unperturbed state.
  2. Second order lowers the ground state. For the ground state, every denominator E⁰₀ − E⁰ₘ is negative, so E²₀ ≤ 0 always.
  3. Level repulsion. In the second-order sum, a state m above n pushes n down, and a state below pushes it up. Neighbouring levels repel.
  4. Validity. The expansion needs |Vₘₙ| ≪ |E⁰ₙ − E⁰ₘ|. When the denominator becomes small, as for near-degenerate levels, the series fails. That is the topic of the next lessons.

Perturbation recipe

Solve H₀ Known states and energies
Write V As matrix elements Vₘₙ
Sum the series E¹, E², then states
1 / 2

Test case 1: a two-level system

Let H₀ = diag(−Δ, +Δ) and V = σₓ (off-diagonal, Vₘₙ = 1). The exact energies are ±√(Δ² + λ²). The first-order shift is zero because Vₙₙ = 0. The second-order shift for the lower level is |V₁₀|²/(−Δ − Δ) = −1/(2Δ) times λ², and the exact expansion −Δ√(1 + λ²/Δ²) ≈ −Δ − λ²/(2Δ) agrees.

Python Two-level system: exact versus second order

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 error shrinks like λ⁴, the order of the first term we left out. For λ comparable to Δ, perturbation theory is no longer a good guide.

Test case 2: an oscillator in a linear field

Take H₀ = p²/2 + x²/2 (units ħ = m = ω = 1) and V = x. This problem is exactly solvable by completing the square: H = p²/2 + (x + λ)²/2 − λ²/2, so every level shifts by exactly −λ²/2 and nothing else changes. Perturbation theory must reproduce that.

The matrix elements of x connect n only to n ± 1: ⟨n+1|x|n⟩ = √((n+1)/2) and ⟨n−1|x|n⟩ = √(n/2). Then E¹ₙ = 0 and

Text
E²ₙ = (n/2)/(+1) + ((n+1)/2)/(−1) = −1/2

Exactly right, and independent of n. The next code applies the general formula to arbitrary matrices and checks both this case and a quartic perturbation.

Python A general perturbation-theory routine

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.

For the quartic case the second-order energy is negative for n = 0, and the combined first-plus-second order result is considerably closer to the exact shift than first order alone.

When does the series work?

A necessary condition is that the matrix element is small compared with the level spacing. For a nearly harmonic anharmonic oscillator, x⁴ couples n to n ± 2, n ± 4 with matrix elements that grow like n², while the spacing stays ħω. So perturbation theory is best for low levels and fails for high ones.

Moreover, the perturbation series for many problems (the anharmonic oscillator included) is asymptotic, not convergent: adding more terms improves the answer up to some point, and then makes it worse. That does not make the low orders wrong; it limits how accurately you can use them.

A perturbation can be "small" in one sense and not in another. A term that is small at the center of an atom can be large at infinity (for example, λx⁴ grows without bound). Always check that the perturbed Hamiltonian is still bounded below and that your states fall off fast enough.

Check your understanding

Quick check · 4 questions

of 4 answered You scored / 4 · perfect!

  1. 1. The first-order energy shift of level n is

    To first order the energy shift is the expectation value of the perturbation in the unperturbed state.
  2. 2. The second-order shift of the ground state is

    Every denominator E⁰₀ − E⁰ₘ is negative for excited m, and the numerators are non-negative.
  3. 3. For the oscillator with perturbation λx (units ħ = m = ω = 1), the exact energy shift of every level is

    Completing the square gives (x + λ)²/2 − λ²/2, so every level is lowered by λ²/2.
  4. 4. Perturbation theory is expected to fail when

    The small parameter is |Vₘₙ|/|Eₙ − Eₘ|; when it is not small the series is unreliable.

Summary

  • Expand E and |n⟩ in powers of λ and match orders.
  • E¹ₙ = Vₙₙ; E²ₙ = Σ_{m≠n} |Vₘₙ|²/(E⁰ₙ − E⁰ₘ); the first-order state mixes in other states with weights Vₘₙ/(E⁰ₙ − E⁰ₘ).
  • Ground-state second-order shifts are negative; neighbouring levels repel.
  • It works when matrix elements are small compared with level spacings; the series is often only asymptotic.

Next: The Anharmonic Oscillator