Skip to content
Tutorials Center
My learning

Topics

Lesson 16 of 39 · Time-Dependent Perturbation and Transitions

Time-Dependent Perturbation Theory

Derive first-order transition amplitudes for a system driven by a weak time-dependent perturbation, study constant and pulsed cases, and test them numerically.

The time-independent methods of Chapter 3 shift energy levels. But many experiments ask a different question: if a system starts in state |i⟩ and a weak perturbation acts for a while, what is the probability of finding it in state |f⟩ afterwards? An atom absorbing light, a nucleus hit by a passing particle, a spin tipped by a magnetic pulse: all are transition problems. Time-dependent perturbation theory solves them as a series in the strength of the perturbation.

Setup

Let H(t) = H₀ + V(t) where H₀|n⟩ = Eₙ|n⟩ is known. Expand the state in the H₀ eigenbasis:

Text
|ψ(t)⟩ = Σₙ cₙ(t) e^{−iEₙt/ħ} |n⟩

(the phase factors remove the trivial evolution; cₙ are the interaction-picture amplitudes of Lesson 1 of this chapter). Substituting into the Schrödinger equation gives a set of coupled equations

Text
iħ dc_m/dt = Σₙ V_{mn}(t) e^{iω_{mn}t} cₙ(t),   ω_{mn} = (Eₘ − Eₙ)/ħ

These are exact. Solve them iteratively starting from c_i(0) = 1, all others zero.

First-order amplitude

To lowest order, c_i ≈ 1 on the right-hand side and c_f for f ≠ i grows:

Text
c_f^{(1)}(t) = −(i/ħ) ∫₀ᵗ V_{fi}(t') e^{iω_{fi}t'} dt'

and the transition probability is P_{i→f}(t) = |c_f^{(1)}(t)|². Everything depends on the Fourier transform of the matrix element V_{fi}(t) at the transition frequency ω_{fi}: the perturbation can cause the transition only if it contains frequency components at ω_{fi}.

First-order theory is valid only while P ≪ 1. It does not conserve probability (the probability of staying in |i⟩ is still taken as 1). If the computed P approaches 1, the approximation has failed and you need a higher order or an exact method, as in Rabi oscillation (next lesson).

Case 1: a constant perturbation switched on at t = 0

Let V(t) = V for t ≥ 0. Then V_{fi} is constant and the integral is elementary:

Text
c_f^{(1)}(t) = −(V_{fi}/ħ) · (e^{iω_{fi}t} − 1)/ω_{fi}
P_{i→f}(t) = (|V_{fi}|²/ħ²) · 4 sin²(ω_{fi}t/2) / ω_{fi}²

The probability oscillates in time with the frequency ω_{fi}. Its maximum is 4|V_{fi}|²/(ħω_{fi})², small if the energy difference is much larger than the coupling. A constant perturbation can cause appreciable transitions only if ω_{fi} is small, that is, if the two levels are nearly degenerate. For ω_{fi} = 0, P = |V_{fi}|²t²/ħ², growing quadratically (until first order breaks down).

The next experiment compares this formula with an exact solution of a two-level system.

Python Constant perturbation: first order versus exact

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.

At short times the agreement is within a fraction of a percent. At later times the exact result drifts from the first-order curve (about 25 percent off by t = 6 here) because the coupling slightly shifts the level energies, an effect that first order cannot see. For larger couplings the departure sets in sooner.

Case 2: a pulse

Let V(t) = V₀ e^{−t²/τ²} (a Gaussian pulse centered at t = 0) acting from t = −∞ to +∞. The Fourier transform is ∫ e^{−t²/τ²} e^{iωt} dt = τ√π e^{−ω²τ²/4}, so

Text
P_{i→f} = (π |V_{fi}|² τ² / ħ²) · exp(−ω_{fi}²τ²/2)

Two regimes follow:

  • Slow (adiabatic) pulse, ω_{fi}τ ≫ 1: the probability is exponentially small. A very gradual perturbation does not change the quantum state, apart from letting it follow the instantaneous eigenstate (Lesson 5 returns to this).
  • Fast (sudden) pulse, ω_{fi}τ ≪ 1: P ≈ π|V_{fi}|²τ²/ħ², which is the overlap picture: a very short kick transfers amplitude proportional to its total area.
Python Gaussian pulse: first-order formula versus exact numerics

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.

Notice how quickly the probability falls once ωτ exceeds about 2 to 3: a smooth, slow pulse has almost no frequency content at ω.

Higher orders

The second-order amplitude involves intermediate states:

Text
c_f^{(2)}(t) = (−i/ħ)² Σₙ ∫₀ᵗ dt₁ ∫₀^{t₁} dt₂ V_{fn}(t₁)e^{iω_{fn}t₁} V_{ni}(t₂)e^{iω_{ni}t₂}

The system goes i → n → f in two steps. The intermediate state n need not conserve energy (it is "virtual"), but the overall transition, observed for long times, will. This is how two-photon processes are described.

Check your understanding

Quick check · 4 questions

of 4 answered You scored / 4 · perfect!

  1. 1. The first-order transition amplitude depends on

    c_f^{(1)} = −(i/ħ)∫V_{fi}(t)e^{iω_{fi}t}dt is the Fourier transform of the matrix element at the transition frequency.
  2. 2. For a constant perturbation, the maximum first-order transition probability is

    From P = 4|V|² sin²(ωt/2)/(ħω)², the maximum of sin² is 1.
  3. 3. A very slow Gaussian pulse (ω_{fi}τ ≫ 1) causes

    The factor exp(−ω²τ²/2) suppresses transitions: the system adiabatically follows the perturbation.
  4. 4. When is first-order perturbation theory not reliable?

    The approximation assumes the initial-state amplitude stays near 1, which fails when P approaches 1.

Summary

  • Expand |ψ⟩ in H₀ eigenstates; the amplitudes obey iħ ċ_m = Σ V_{mn} e^{iω_{mn}t} cₙ.
  • First order: c_f = −(i/ħ)∫V_{fi}(t)e^{iω_{fi}t}dt, P = |c_f|², valid only for P ≪ 1.
  • A constant perturbation gives P = 4|V_{fi}|² sin²(ω_{fi}t/2)/(ħω_{fi})².
  • A smooth Gaussian pulse has transition probability ∝ exp(−ω²τ²/2).
  • Transitions need frequency content of the perturbation at ω_{fi}.

Next: Rabi Oscillations and the Two-Level System