30
Part One · Quantum mechanics

The uncertainty principle

The uncertainty principle is often explained as clumsiness: to see an electron you must hit it with light, and the light kicks it. That story is not wrong, but it misses the point. The real statement is stronger and stranger. A particle with both a sharp position and a sharp momentum does not exist, before anyone measures anything.

It follows in a few lines from the commutator [x^,p^]=iℏ[\hat x,\hat p] = i\hbar of Chapter 29, and from the Fourier trade-off of Chapter 6. The plate shows a wavefunction in position on the left and the same state in momentum on the right, with their spreads.

Squeeze one, the other spreads

Make the packet narrower or wider: width 1.00. Its momentum distribution does the opposite. Choose a shape: , or . The readout shows Δx Δp\Delta x\,\Delta p in units of ℏ\hbar.

For the bell curve the product is exactly 12\tfrac12, whatever the width. The triangle gives about 0.55; two separated bumps give much more. Nothing ever goes below one half.

The Robertson relation

For any two observables A^\hat A and B^\hat B and any state, Howard Robertson proved in 1929 that

ΔA ΔB≥12∣⟨[A^,B^]⟩∣\Delta A\,\Delta B \ge \tfrac12\big|\langle[\hat A,\hat B]\rangle\big|

With [x^,p^]=iℏ[\hat x,\hat p] = i\hbar this gives Heisenberg’s Δx Δp≥ℏ/2\Delta x\,\Delta p \ge \hbar/2. The proof is Cauchy–Schwarz (Chapter 9) applied to the vectors (A^−⟨A⟩)∣ψ⟩(\hat A - \langle A\rangle)|\psi\rangle and (B^−⟨B⟩)∣ψ⟩(\hat B - \langle B\rangle)|\psi\rangle. Established

DeeperLet ∣f⟩=(A^−⟨A⟩)∣ψ⟩|f\rangle = (\hat A - \langle A\rangle)|\psi\rangle and ∣g⟩=(B^−⟨B⟩)∣ψ⟩|g\rangle = (\hat B - \langle B\rangle)|\psi\rangle. Then ΔA2ΔB2=⟨f∣f⟩⟨g∣g⟩≥∣⟨f∣g⟩∣2≥(Im⟨f∣g⟩)2=14∣⟨[A^,B^]⟩∣2\Delta A^2\Delta B^2 = \langle f|f\rangle\langle g|g\rangle \ge |\langle f|g\rangle|^2 \ge (\mathrm{Im}\langle f|g\rangle)^2 = \tfrac14|\langle[\hat A,\hat B]\rangle|^2. Keeping the real part as well gives Schrödinger’s stronger form, with an extra covariance term. Equality needs ∣g⟩∝i∣f⟩|g\rangle \propto i|f\rangle, which for xx and pp is a first-order equation whose solution is a Gaussian.

What uncertainty is, and is not

The Robertson relation is about preparation: prepare many particles in the same state, measure position on some and momentum on others, and the two spreads obey the bound. No single measurement disturbs anything in that statement.

A separate question is how much measuring xx on one particle disturbs its pp. Heisenberg’s original microscope argument suggested the product of measurement error and disturbance obeys the same bound. In 2003 Masanao Ozawa argued it need not, and proposed a corrected relation. In 2012 Jacqueline Erhart and colleagues measured neutron spins and found products below Heisenberg’s naive bound, consistent with Ozawa’s. Others, notably Paul Busch, Pekka Lahti and Reinhard Werner, define error and disturbance differently and prove a Heisenberg-like bound does hold. Which definitions are the right ones is still argued. Debated

Why atoms do not collapse

Here is the answer to the crisis of Chapter 16, in rough form. Squeeze an electron into a region of size rr around a proton and its momentum spread is at least about ℏ/r\hbar/r, costing kinetic energy ℏ2/2mr2\hbar^2/2m r^2. The electrical energy is −e2/4πε0r-e^2/4\pi\varepsilon_0 r. Shrinking rr lowers the electrical energy but raises the kinetic energy faster. Set the size to 2.00 Bohr radii and watch the total.

The minimum sits at exactly one Bohr radius, with energy −13.6-13.6 eV. The estimate is rough, but here it lands on the exact answer. Matter is stable because confinement costs energy.

Energy and time

There is also a relation ΔE Δt≳ℏ/2\Delta E\,\Delta t \gtrsim \hbar/2, but it is a different kind of statement: time is not an operator in quantum mechanics, so Robertson’s proof does not apply. The precise version, due to Leonid Mandelstam and Igor Tamm in 1945, says Δt\Delta t is how long it takes some observable to change by about one of its own spreads. A state with a sharp energy changes slowly.

In practice: an excited atom that lives for a time τ\tau emits light with a frequency spread of about 1/2πτ1/2\pi\tau. The yellow sodium line comes from a state living 16 ns, giving a natural width of about 10 MHz. Established

What you now have

The Robertson and Schrödinger relations, Gaussians as minimum-uncertainty states, the difference between preparation uncertainty and measurement disturbance, the stability of atoms, and the meaning of energy–time uncertainty.

· · ·

Three questions to carry with you.

Which wave packet has the smallest ΔxΔp?tap
The Gaussian, with ΔxΔp = ħ/2 exactly.
Is Δx Δp ≥ ħ/2 about measurement disturbance?tap
No; it is about the spreads of a prepared state. Error–disturbance relations are a separate, debated question.
How does uncertainty keep atoms from collapsing?tap
Confining the electron to a small region raises its kinetic energy as ħ²/2mr², which beats the electrical energy gain at small r.

And one to work out. An excited state lives 1 ns. Roughly what is its natural linewidth 1/2πτ1/2\pi\tau, in MHz?

Δf ≈MHz

Sources: W. Heisenberg, Z. Phys. 43, 172 (1927). H. P. Robertson, Phys. Rev. 34, 163 (1929). L. Mandelstam and I. Tamm, J. Phys. USSR 9, 249 (1945). M. Ozawa, Phys. Rev. A 67, 042105 (2003). J. Erhart et al., Nat. Phys. 8, 185 (2012). P. Busch, P. Lahti and R. F. Werner, Phys. Rev. Lett. 111, 160405 (2013). Checked by code: ΔxΔp = 0.500 (Gaussian), 0.548 (triangle), 2.06 (two bumps); sodium D linewidth 9.8 MHz.

Next · Chapter 31
The Schrödinger equation

Schrödinger guessed the equation. Here's why the guess had to be right.

Turn the page