Skip to content
Tutorials Center
My learning

Topics

Lesson 11 of 39 · Time-Independent Perturbation Theory

The Anharmonic Oscillator

Apply perturbation theory to cubic and quartic anharmonic oscillators, compare with exact diagonalization, and see why the series is only asymptotic.

Real molecular vibrations are not perfectly harmonic. Bonds stretch more easily than they compress, and they eventually break. The simplest model of such behavior adds a cubic or quartic term to the oscillator potential. The anharmonic oscillator is also the standard laboratory in which perturbation theory is tested, because it is easy to solve numerically to high accuracy.

In this lesson you will compute first- and second-order corrections analytically, verify them numerically, and meet an important fact: the perturbation series diverges.

The model

We work in units ħ = m = ω = 1, so H₀ = p²/2 + x²/2 with energies n + ½. Two perturbations are standard:

Text
quartic:  V = λ x⁴        cubic:  V = λ x³

Position in terms of ladder operators is x = (a + a†)/√2. Powers of x expand into products of a and a†, whose matrix elements follow from a|n⟩ = √n |n−1⟩ and a†|n⟩ = √(n+1) |n+1⟩.

Quartic perturbation

The first-order shift requires ⟨n|x⁴|n⟩. Expanding (a + a†)⁴ and keeping the terms with equal numbers of a and a† gives

Text
⟨n|x⁴|n⟩ = (3/4)(2n² + 2n + 1)

So E¹ₙ = (3/4)(2n² + 2n + 1) λ. For the ground state this is (3/4)λ. Notice the quadratic growth in n: highly excited states are shifted much more.

The second-order calculation sums over m = n ± 2, n ± 4. The result is

Text
E²ₙ = −(λ²/8)(34n³ + 51n² + 59n + 21)

For n = 0 this gives −(21/8)λ².

Cubic perturbation

A cubic term is odd in x, so ⟨n|x³|n⟩ = 0 and the first-order shift vanishes. The correction first appears at second order, from m = n ± 1, n ± 3:

Text
E²ₙ = −(λ²/8)(30n² + 30n + 11)

For n = 0, −(11/8)λ². The sign is negative for all n, even though the cubic potential is unbounded below at large negative x (the system is actually unstable and its levels have only a finite lifetime, a subtlety we return to below).

Python Checking the analytic corrections numerically

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 small residual differences are higher-order terms (third order for the quartic case is of size λ³). The basis N = 120 is large enough that truncation does not matter for these low levels. For the cubic potential the "exact" numbers refer to the truncated matrix, which is a regularized version of an unstable problem; for small λ low levels are practically stable and well described.

The cubic oscillator has no true bound states: for large negative x the potential x³ plunges to minus infinity and the particle can tunnel out. The levels you compute are resonances with lifetimes that are exponentially long for small λ. Perturbation theory cannot see this: the series only captures the real part of the energy, and it too is asymptotic.

Why the series diverges

For the quartic oscillator, the known series for the ground-state energy begins

Text
E₀(λ) = 1/2 + (3/4)λ − (21/8)λ² + (333/16)λ³ − (30885/128)λ⁴ + ...

The coefficients grow factorially, roughly like n!, with alternating signs. No matter how small λ is, the series diverges for large enough order. The reason is physical: for negative λ the Hamiltonian is unbounded below, so E(λ) has a singularity at λ = 0 and the radius of convergence is zero. For small positive λ, summing the series up to the smallest term gives an excellent approximation, with an error about the size of that term. This is called an asymptotic series.

Python Partial sums of the quartic ground-state series

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 λ = 0.01 each extra order helps. For λ = 0.3 the terms stop shrinking after the first few and the partial sums wander away from the exact value. That is the asymptotic behavior in action. Specialized resummation techniques (Padé, Borel) can recover accurate results from divergent series, but they are beyond this track.

How to use anharmonic results

For a diatomic molecule, a Morse-type or cubic-plus-quartic potential gives vibrational energies

Text
E(n) ≈ ħω(n + ½) − ħω xₑ (n + ½)² + ...

so level spacings shrink as n rises, a feature seen in measured vibrational spectra. The anharmonic correction is negative-quadratic in (n + ½), consistent with the cubic second-order result above, which is negative and quadratic in n.

Check your understanding

Quick check · 4 questions

of 4 answered You scored / 4 · perfect!

  1. 1. Why is the first-order shift from a cubic perturbation λx³ equal to zero?

    The oscillator probability density is symmetric, so the expectation of an odd function vanishes.
  2. 2. The first-order energy shift of the ground state for V = λx⁴ (ħ = m = ω = 1) is

    From (3/4)(2n² + 2n + 1) with n = 0, the shift is 3λ/4.
  3. 3. The perturbation series for the quartic oscillator is asymptotic because

    The energy as a function of λ is singular at λ = 0, so the radius of convergence is zero even though truncated sums are accurate for small positive λ.
  4. 4. In a real molecule, anharmonicity makes the spacing between vibrational levels

    The negative quadratic term in (n + ½) reduces the spacing at higher levels.

Summary

  • For V = λx⁴: E¹ₙ = (3/4)(2n² + 2n + 1)λ, E²ₙ = −(λ²/8)(34n³ + 51n² + 59n + 21).
  • For V = λx³: first order vanishes; E²ₙ = −(λ²/8)(30n² + 30n + 11).
  • Numerical diagonalization in the oscillator basis confirms these formulas.
  • The series is asymptotic: use it for small λ, stop at the smallest term.
  • Anharmonicity makes vibrational level spacings shrink at higher n.

Next: Degenerate Perturbation Theory