31
Part One · Quantum mechanics

The Schrödinger equation

Over the Christmas holiday of 1925, in a villa in the Swiss Alps, Erwin Schrödinger guessed the equation that bears his name. He did not derive it from anything deeper; there is nothing deeper to derive it from. But the guess was not a wild one. Given de Broglie’s waves and two facts about energy and momentum, it is almost the only possible equation, and this chapter shows why.

The plate is the workhorse of the rest of Part One: an exact solver for a wave packet in one dimension, in any potential you choose.

Why the guess had to be right

Start with a free particle as a plane wave, ψ=ei(kx−ωt)\psi = e^{i(kx-\omega t)}, with de Broglie’s p=ℏkp = \hbar k and Planck’s E=ℏωE = \hbar\omega. A time derivative pulls down −iω-i\omega; two space derivatives pull down −k2-k^2. So iℏ ∂tψ=Eψi\hbar\,\partial_t\psi = E\psi and −ℏ2∂x2ψ=p2ψ-\hbar^2\partial_x^2\psi = p^2\psi. The classical rule E=p2/2m+VE = p^2/2m + V then becomes a wave equation:

iℏ∂ψ∂t=−ℏ22m∂2ψ∂x2+V(x) ψi\hbar\frac{\partial\psi}{\partial t} = -\frac{\hbar^2}{2m}\frac{\partial^2\psi}{\partial x^2} + V(x)\,\psi

It must be first order in time, so that the present state fixes the future, and linear, so that waves superpose as the double slit demands. The ii is unavoidable: without it, the equation would be the heat equation of Chapter 5, and wavefunctions would decay instead of oscillating. Established

The solver

Pick a landscape: , , , or . Then .

The amber area is ∣ψ∣2|\psi|^2; the pale line is the real part, showing the oscillation inside. Watch it reflect from the wall in a burst of interference, slosh in the valley, split at the barrier. This is quantum dynamics, exactly, in units where ℏ=m=1\hbar = m = 1.

DeeperThe solver uses the split-step Fourier method: over a short time Δt\Delta t, it applies half a step of the potential as a phase e−iVΔt/2ℏe^{-iV\Delta t/2\hbar} in position space, a full step of kinetic energy as e−iℏk2Δt/2me^{-i\hbar k^2\Delta t/2m} in momentum space, and another half step of potential. Each factor is unitary, so the norm is kept exactly; the error from splitting is of order Δt3\Delta t^3 per step, by BCH (Chapter 29). It reproduces the exact spreading of a free Gaussian to four decimal places.

States that do not change

Try a solution where space and time separate: ψ(x,t)=ϕ(x) e−iEt/ℏ\psi(x,t) = \phi(x)\,e^{-iEt/\hbar}. Substituting gives the time-independent Schrödinger equation, an eigenvalue problem:

−ℏ22mϕ′′+Vϕ=E ϕ-\frac{\hbar^2}{2m}\phi'' + V\phi = E\,\phi

Its solutions are the stationary states. Their phase turns, but ∣ψ∣2|\psi|^2 never changes: nothing moves. The plate puts the lowest state of the parabolic valley in; watch the real part shimmer while the amber area stays perfectly still. Atoms in their ground state are like this, which is why they do not radiate.

Beats between two states

Now mix the two lowest stationary states equally. Each turns at its own rate, E1/ℏE_1/\hbar and E2/ℏE_2/\hbar, so their sum keeps changing: the probability sloshes back and forth at the difference frequency (E2−E1)/h(E_2 - E_1)/h. Motion in quantum mechanics is always interference between states of different energy. .

Probability is never lost

The total probability ∫∣ψ∣2dx\int|\psi|^2dx stays exactly one for all time. The readout shows it to six decimal places while the packet moves. This is unitarity: the evolution ψ(t)=e−iHt/ℏψ(0)\psi(t) = e^{-iHt/\hbar}\psi(0) is a rotation in Hilbert space, because HH is Hermitian (Chapter 29).

The classical limit, built in

Paul Ehrenfest showed in 1927 that averages obey Newton-like laws: ddt⟨x⟩=⟨p⟩/m\frac{d}{dt}\langle x\rangle = \langle p\rangle/m and ddt⟨p⟩=−⟨V′(x)⟩\frac{d}{dt}\langle p\rangle = -\langle V'(x)\rangle. In the parabolic valley, V′V' is linear, so ⟨V′(x)⟩=V′(⟨x⟩)\langle V'(x)\rangle = V'(\langle x\rangle) and the average position follows the classical path exactly. The plate traces ⟨x⟩\langle x\rangle against time, in amber, over the classical cosine. They are the same curve. In other potentials they agree only while the packet is narrow compared with how fast the force changes.

DeeperUse ddt⟨A⟩=iℏ⟨[H,A]⟩\frac{d}{dt}\langle A\rangle = \frac{i}{\hbar}\langle[H, A]\rangle. With H=p2/2m+VH = p^2/2m + V: [H,x]=−iℏp/m[H, x] = -i\hbar p/m and [H,p]=iℏV′(x)[H, p] = i\hbar V'(x).

Two pictures of the same physics

So far the state changes and the operators stay fixed: the Schrödinger picture. You can instead freeze the state and let the operators evolve, A^(t)=eiHt/ℏA^ e−iHt/ℏ\hat A(t) = e^{iHt/\hbar}\hat A\,e^{-iHt/\hbar}, which obey dA^dt=iℏ[H,A^]\frac{d\hat A}{dt} = \frac{i}{\hbar}[H,\hat A]. This is the Heisenberg picture, and it is Hamilton’s equation of Chapter 13 with Dirac’s replacement. Every prediction is identical in the two pictures; you use whichever makes the calculation easier.

What you now have

The Schrödinger equation and why it must look the way it does, stationary states, beats, unitarity, Ehrenfest’s theorem, the two pictures, and an exact solver you will use again and again.

· · ·

Three questions to carry with you.

Why does the Schrödinger equation need i?tap
Without it the equation is a diffusion equation and wavefunctions would decay; with it they oscillate and probability is conserved.
Why does nothing move in a stationary state?tap
Its time dependence is a single global phase e^(−iEt/ħ), which leaves |ψ|² unchanged.
When does ⟨x⟩ follow the classical path exactly?tap
When the force is linear in x (constant force or harmonic potential); otherwise only approximately, for narrow packets.

And one to work out. Two states differ in energy by 2 eV. What is the period of the beating, in femtoseconds? Use h=4.136×10−15h = 4.136\times10^{-15} eV s.

T =fs

Sources: E. Schrödinger, Ann. Phys. 79, 361 (1926). P. Ehrenfest, Z. Phys. 45, 455 (1927). D. J. Tannor, Introduction to Quantum Mechanics: A Time-Dependent Perspective. Solver checked in code: free Gaussian width 2.6926 at t = 5 (exact 2.6926); norm 1.000000; harmonic half-period ⟨x⟩ = −3.000.

Next · Chapter 32
Free particle and wavepackets

Leave an electron alone and it spreads out. How long until it's the size of a room?

Turn the page