Skip to content
Tutorials Center
My learning

Topics

Lesson 9 of 39 · Hilbert Space and Dirac Notation

Position and Momentum Representations

Connect abstract kets to wave functions: continuous bases, delta-function normalization, the Fourier link between x and p, and the FFT in practice.

The same quantum state can be written as a function of position, ψ(x), or as a function of momentum, φ(p). They are two representations of one abstract ket |ψ⟩, related by a Fourier transform. Choosing the right representation can turn a hard differential equation into an algebraic one.

This lesson builds the continuous bases |x⟩ and |p⟩ from Dirac notation, explains why they need delta-function normalization, derives the Fourier relation, and then uses the fast Fourier transform to move wave functions between representations on a computer.

Continuous eigenbases

The position operator x̂ has eigenvectors |x⟩ with x̂|x⟩ = x|x⟩, one for every real x. They are not normalizable. Instead:

Text
⟨x|x'⟩ = δ(x − x')            (orthonormality)
∫ |x⟩⟨x| dx = 1                (completeness)

Then the wave function is the component of the ket along |x⟩: ψ(x) = ⟨x|ψ⟩, and |ψ(x)|² dx is the probability of finding the particle in [x, x+dx]. Normalization of the state reads ∫|ψ(x)|² dx = 1.

The same applies to momentum: p̂|p⟩ = p|p⟩, ⟨p|p'⟩ = δ(p − p'), ∫|p⟩⟨p| dp = 1, and φ(p) = ⟨p|ψ⟩.

States like |x⟩ and |p⟩ are called generalized eigenvectors. They are not in the Hilbert space L², so no physical particle is ever in one. They are tools: any real state is a normalizable superposition such as a wave packet.

The commutation relation [x̂, p̂] = iħ is realized in the position representation by p̂ = −iħ d/dx. Solving −iħ dχ/dx = pχ gives the momentum eigenfunction

Text
⟨x|p⟩ = e^{ipx/ħ} / √(2πħ)

(the factor makes ⟨p|p'⟩ = δ(p − p')). Insert ∫|p⟩⟨p|dp to find the Fourier pair

Text
φ(p) = (1/√(2πħ)) ∫ ψ(x) e^{−ipx/ħ} dx
ψ(x) = (1/√(2πħ)) ∫ φ(p) e^{+ipx/ħ} dp

A key property is that ∫|φ|² dp = ∫|ψ|² dx (Plancherel). Narrow in x means wide in p, which is the uncertainty principle seen as a property of Fourier transforms.

In momentum space the roles swap: x̂ = iħ d/dp, and the kinetic energy p²/2m is just multiplication. That is why the free particle is trivial in momentum space and why split-step simulations alternate between the two.

A Gaussian packet and its momentum distribution

A Gaussian ψ(x) ∝ exp(−x²/(4σ²)) has position spread σ and momentum spread ħ/(2σ): it saturates Δx Δp = ħ/2. The next example computes this numerically with the FFT, in units with ħ = 1. The momentum values on an FFT grid are p = 2πħ k / (NΔx) for integer k, ordered with numpy.fft.fftfreq.

Python Gaussian wave packet: position and momentum spreads

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 Δx · Δp is the same for every σ and equals ħ/2, the minimum allowed. Squeezing the packet in position broadens it in momentum.

Computing with both representations

The operator p̂²/2m is diagonal in momentum space; V(x) is diagonal in position space. To act with p̂² on ψ(x) numerically, you transform to momentum space, multiply by p², and transform back. This is spectrally accurate for smooth, periodic functions, much better than the finite-difference formula of Lesson 3 in Chapter 1.

Python Kinetic energy via FFT versus finite differences

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 FFT result matches to near machine precision on this grid, while the three-point difference carries an error of order Δx².

The generalized view

Every observable has its own representation: energy (the basis that diagonalizes H), angular momentum, spin. A state is one abstract vector; its components in different bases are different descriptions, linked by unitary matrices (or integral kernels such as ⟨x|p⟩). Nothing physical changes when you change the representation.

One state, many views

Position ψ(x) = ⟨x|ψ⟩
Momentum φ(p) = ⟨p|ψ⟩
Energy cₙ = ⟨n|ψ⟩
1 / 2

Check your understanding

Quick check · 4 questions

of 4 answered You scored / 4 · perfect!

  1. 1. The wave function ψ(x) is best described as

    The ket is representation independent; ψ(x) = ⟨x|ψ⟩ is its position-space components.
  2. 2. Position eigenvectors |x⟩ are normalized as

    They are continuum states with delta-function orthonormality, not square-integrable vectors.
  3. 3. The momentum eigenfunction in position space is proportional to

    Solving −iħ dχ/dx = pχ gives a plane wave.
  4. 4. A Gaussian wave packet that is made narrower in position becomes

    For a Gaussian the product Δx Δp stays at ħ/2, so squeezing one broadens the other.

Summary

  • ψ(x) = ⟨x|ψ⟩ and φ(p) = ⟨p|ψ⟩ are components of one ket in two continuous bases.
  • Continuum states are normalized with delta functions and are not physical states by themselves.
  • ⟨x|p⟩ = e^{ipx/ħ}/√(2πħ), so the two representations form a Fourier pair.
  • Kinetic energy is diagonal in momentum space; potential energy is diagonal in position space.
  • The FFT converts between them on a computer and gives high accuracy for smooth states.

Next: Nondegenerate Perturbation Theory