Uncertainty Relations
The uncertainty principle is the quantitative statement that non-commuting observables cannot both have sharp values in the same state. It is not a statement about measurement disturbance or experimental imperfection; it is a theorem about the spread of the probability distributions assigned by a state through the Born rule. This page derives the general Robertson–Schrödinger inequality, specializes it to position and momentum, discusses the (frequently misstated) energy–time relation, and identifies the minimum-uncertainty states.
1. Variance of an Observable
For an observable (Hermitian) and a normalized state , define the uncertainty as the standard deviation of its measurement outcomes:
Writing , we have . The uncertainty vanishes iff is an eigenstate of .
2. The Robertson–Schrödinger Inequality
Let and . The Cauchy–Schwarz inequality gives
Split the operator product into its Hermitian and anti-Hermitian parts,
where is the anticommutator. The first term has real expectation, the second (times ) real expectation, so the two contributions add in quadrature:
This is the Schrödinger form. Dropping the (nonnegative) anticommutator term gives the weaker, more familiar Robertson relation:
3. Position–Momentum
Using the canonical commutator , the Robertson relation gives the Heisenberg uncertainty principle:
Because and are a Fourier-transform pair, this is exactly the classical bandwidth theorem: a function and its transform cannot both be arbitrarily narrow. Localizing a particle in space forces a broad momentum distribution, and vice versa.
4. Minimum-Uncertainty (Gaussian) States
Equality in Cauchy–Schwarz requires , i.e. , and equality in §2 additionally requires the anticommutator term to vanish, forcing pure imaginary. Solving the resulting first-order ODE in the position representation gives a Gaussian wavepacket,
the unique states saturating . These are precisely the coherent states of the harmonic oscillator, the "most classical" quantum states.
5. The Energy–Time Relation
The relation
is not an instance of §2, because time is a parameter in non-relativistic QM, not an operator (there is no Hermitian conjugate to ). Its correct reading (Mandelstam–Tamm): for any observable , define the characteristic evolution time — the time for to change by one standard deviation. Then the Robertson relation applied to and yields
So is the timescale on which the state changes appreciably, not an uncertainty in a measured time. Consequences: stationary states () never evolve; short-lived states (small ) have broad energy widths — the origin of the natural linewidth of the decay processes treated by time-dependent perturbation theory.
See also
- Wave Mechanics — Fourier duality behind .
- Harmonic Oscillator — coherent states as minimum-uncertainty states.
- Postulates — the Born rule that defines the distributions being bounded.