Skip to content

Operators, Commutators, Dirac Notation and Uncertainty

CSIR NET quantum mechanics: Dirac notation, Hermitian operators, measurement, commutators, generalised uncertainty, position and momentum representations.

  • 18 min read
  • Updated
On this page · 16 sections
  1. Prerequisites
  2. The Hook: A Polariser as a Measurement
  3. States as Kets
  4. Operators
  5. The Measurement Postulates
  6. Commutators
  7. The Uncertainty Principle
  8. Coordinate and Momentum Representations
  9. Time Evolution: Schrödinger and Heisenberg Pictures
  10. Memory Tricks
  11. Don’t Confuse
  12. Common Mistakes
  13. Practice
  14. Quick Revision Table
  15. Key Takeaways
  16. What’s Next

Prerequisites

The Hook: A Polariser as a Measurement

States as Kets

A quantum state is a vector \( \lvert\psi\rangle \) in a complex Hilbert space. Its dual, the bra \( \langle\psi\rvert \), is the conjugate transpose. The inner product \( \langle\phi\vert\psi\rangle \) is a complex number with

$$\langle\phi\vert\psi\rangle = \langle\psi\vert\phi\rangle^*, \qquad \langle\psi\vert\psi\rangle \geq 0 \ (\text{equality only for the zero vector}).$$

Physical states are normalised, \( \langle\psi\vert\psi\rangle = 1 \), and two kets differing only by an overall phase represent the same state.

Bases and completeness

An orthonormal basis \( \{\lvert n\rangle\} \) satisfies \( \langle m\vert n\rangle = \delta_{mn} \) and the completeness relation

$$\boxed{\sum_n \lvert n\rangle\langle n\rvert = \hat 1}$$

Inserting this identity anywhere is the single most useful move in the subject. It gives the expansion \( \lvert\psi\rangle = \sum_n c_n\lvert n\rangle \) with \( c_n = \langle n\vert\psi\rangle \), and \( \langle\psi\vert\psi\rangle = \sum\lvert c_n\rvert^2 \).

For a continuous basis such as position, sums become integrals and Kronecker deltas become Dirac deltas:

$$\langle x\vert x'\rangle = \delta(x - x'), \qquad \int \lvert x\rangle\langle x\rvert\,dx = \hat 1, \qquad \psi(x) = \langle x\vert\psi\rangle.$$

The wavefunction is just the set of components of the ket in the position basis.

Operators

A linear operator maps kets to kets. In a basis it is a matrix, \( A_{mn} = \langle m\rvert\hat A\lvert n\rangle \). The adjoint \( \hat A^\dagger \) is defined by \( \langle\phi\rvert\hat A^\dagger\lvert\psi\rangle = \langle\psi\rvert\hat A\lvert\phi\rangle^* \); as a matrix, it is the conjugate transpose.

TypeDefinitionKey property
Hermitian\( \hat A^\dagger = \hat A \)real eigenvalues; eigenvectors of distinct eigenvalues orthogonal; complete basis
Anti-Hermitian\( \hat A^\dagger = -\hat A \)purely imaginary eigenvalues; \( i\hat A \) is Hermitian
Unitary\( \hat U^\dagger\hat U = \hat 1 \)eigenvalues of modulus 1; preserves inner products
Projector\( \hat P^2 = \hat P = \hat P^\dagger \)eigenvalues 0 and 1

Useful rules: \( (\hat A\hat B)^\dagger = \hat B^\dagger\hat A^\dagger \); \( (c\hat A)^\dagger = c^*\hat A^\dagger \); \( (\lvert\phi\rangle\langle\psi\rvert)^\dagger = \lvert\psi\rangle\langle\phi\rvert \). If \( \hat H \) is Hermitian, \( e^{i\hat H} \) is unitary. The product of two Hermitian operators is Hermitian only if they commute.

Observables are represented by Hermitian operators. Position \( \hat x \), momentum \( \hat p \), the Hamiltonian, angular momentum and spin components are all Hermitian. The operator \( d/dx \) is anti-Hermitian, which is why momentum needs the factor \( -i\hbar \).

Trace and determinant

The trace (sum of eigenvalues) and determinant (product of eigenvalues) are basis-independent, and \( \text{Tr}(\hat A\hat B) = \text{Tr}(\hat B\hat A) \). Since a commutator has zero trace, no finite-dimensional matrices can satisfy \( [\hat x,\hat p] = i\hbar\hat 1 \): position and momentum need an infinite-dimensional space.

The Measurement Postulates

For an observable \( \hat A \) with eigenvalues \( a_n \) and orthonormal eigenvectors \( \lvert a_n\rangle \):

  1. A measurement gives one of the eigenvalues \( a_n \).
  2. In state \( \lvert\psi\rangle \), the probability of \( a_n \) is \( \lvert\langle a_n\vert\psi\rangle\rvert^2 \) (for a degenerate eigenvalue, sum over an orthonormal basis of its eigenspace).
  3. Immediately after obtaining \( a_n \), the state is the normalised projection of \( \lvert\psi\rangle \) onto that eigenspace.
  4. The expectation value is \( \langle A\rangle = \langle\psi\rvert\hat A\lvert\psi\rangle = \sum_n a_n\lvert\langle a_n\vert\psi\rangle\rvert^2 \).

Commutators

$$[\hat A,\hat B] = \hat A\hat B - \hat B\hat A$$

Identities

$$[\hat A,\hat B\hat C] = [\hat A,\hat B]\hat C + \hat B[\hat A,\hat C], \qquad [\hat A\hat B,\hat C] = \hat A[\hat B,\hat C] + [\hat A,\hat C]\hat B$$$$[\hat A,\hat B]^\dagger = [\hat B^\dagger,\hat A^\dagger], \qquad [\hat A,[\hat B,\hat C]] + [\hat B,[\hat C,\hat A]] + [\hat C,[\hat A,\hat B]] = 0$$

The commutator of two Hermitian operators is anti-Hermitian, so it can always be written \( i\hat C \) with \( \hat C \) Hermitian.

The canonical commutator and its consequences

$$\boxed{[\hat x,\hat p] = i\hbar}$$

Prove it by acting on a test function: \( [\hat x,\hat p]f = -i\hbar x f' + i\hbar(xf)' = i\hbar f \). From it and the product rule:

CommutatorValue
\( [\hat x,\hat p^n] \)\( i\hbar n\hat p^{n-1} \)
\( [\hat p,\hat x^n] \)\( -i\hbar n\hat x^{n-1} \)
\( [\hat x,f(\hat p)] \)\( i\hbar f'(\hat p) \)
\( [\hat p,g(\hat x)] \)\( -i\hbar g'(\hat x) \)
\( [\hat x,\hat H] \), \( \hat H = \hat p^2/2m + V \)\( i\hbar\hat p/m \)
\( [\hat p,\hat H] \)\( -i\hbar V'(\hat x) \)
\( [\hat x_i,\hat p_j] \)\( i\hbar\delta_{ij} \)

In three dimensions different Cartesian components commute: \( [\hat x,\hat p_y] = 0 \).

Compatible observables

Two Hermitian operators commute if and only if they have a complete set of common eigenvectors. Then both quantities can be sharp at once, and measuring one does not disturb the other. When an eigenvalue of one is degenerate, its eigenvectors need not be eigenvectors of the other, but a suitable combination within the degenerate subspace always is. This is the reason we label hydrogen states by \( n, l, m \): \( \hat H \), \( \hat L^2 \) and \( \hat L_z \) all commute.

The Uncertainty Principle

Robertson’s generalised relation

For any two observables and any state,

$$\boxed{\sigma_A\,\sigma_B \geq \frac{1}{2}\left\lvert\langle[\hat A,\hat B]\rangle\right\rvert}$$

Derivation in four lines. Let \( \lvert f\rangle = (\hat A - \langle A\rangle)\lvert\psi\rangle \) and \( \lvert g\rangle = (\hat B - \langle B\rangle)\lvert\psi\rangle \), so \( \sigma_A^2 = \langle f\vert f\rangle \), \( \sigma_B^2 = \langle g\vert g\rangle \). Schwarz: \( \sigma_A^2\sigma_B^2 \geq \lvert\langle f\vert g\rangle\rvert^2 \geq (\text{Im}\langle f\vert g\rangle)^2 \). And \( \langle f\vert g\rangle - \langle g\vert f\rangle = \langle[\hat A,\hat B]\rangle \). So \( \text{Im}\langle f\vert g\rangle = \frac{1}{2i}\langle[\hat A,\hat B]\rangle \), and taking the square root gives the result.

With \( [\hat x,\hat p] = i\hbar \):

$$\boxed{\sigma_x\,\sigma_p \geq \frac{\hbar}{2}}$$

The right side of Robertson’s relation depends on the state. For two spin components, \( \sigma_{S_x}\sigma_{S_y} \geq \frac{\hbar}{2}\lvert\langle S_z\rangle\rvert \), which can be zero.

Minimum-uncertainty states

Equality needs \( \lvert g\rangle = ia\lvert f\rangle \) with \( a \) real. For position and momentum this gives a first-order ODE whose solution is a Gaussian:

$$\psi(x) \propto \exp\left[-\frac{(x - x_0)^2}{4\sigma^2} + \frac{ip_0x}{\hbar}\right]$$

The harmonic-oscillator ground state is one such state. Other states sit above the bound. For the infinite well ground state, \( \sigma_x\sigma_p = \frac{\hbar}{2}\sqrt{\frac{\pi^2}{3} - 2} \approx 0.568\hbar \).

Energy–time relation

$$\Delta t\,\Delta E \gtrsim \frac{\hbar}{2}$$

Time is not an operator, so this is not a Robertson relation. Its precise form: if \( \Delta t \) is the time for \( \langle Q\rangle \) to change by one standard deviation, \( \Delta t = \sigma_Q/\lvert d\langle Q\rangle/dt\rvert \), then \( \Delta E\,\Delta t \geq \hbar/2 \). It explains natural linewidths, \( \Gamma \approx \hbar/\tau \) for a state of lifetime \( \tau \), and the inverse relation between the width of a resonance and the lifetime of the particle.

Coordinate and Momentum Representations

The momentum eigenstates in the position basis are plane waves, normalised to a delta function:

$$\boxed{\langle x\vert p\rangle = \frac{1}{\sqrt{2\pi\hbar}}e^{ipx/\hbar}}$$

The momentum-space wavefunction \( \phi(p) = \langle p\vert\psi\rangle \) is therefore the Fourier transform:

$$\phi(p) = \frac{1}{\sqrt{2\pi\hbar}}\int e^{-ipx/\hbar}\psi(x)\,dx, \qquad \psi(x) = \frac{1}{\sqrt{2\pi\hbar}}\int e^{ipx/\hbar}\phi(p)\,dp$$

and \( \lvert\phi(p)\rvert^2dp \) is the probability of finding momentum in \( dp \). Parseval’s theorem guarantees normalisation is preserved.

OperatorPosition representationMomentum representation
\( \hat x \)\( x \)\( i\hbar\,\partial/\partial p \)
\( \hat p \)\( -i\hbar\,\partial/\partial x \)\( p \)
Eigenfunction of \( \hat x \) (value \( x_0 \))\( \delta(x - x_0) \)\( e^{-ipx_0/\hbar}/\sqrt{2\pi\hbar} \)
Eigenfunction of \( \hat p \) (value \( p_0 \))\( e^{ip_0x/\hbar}/\sqrt{2\pi\hbar} \)\( \delta(p - p_0) \)

Two Fourier facts do most of the work: a Gaussian transforms into a Gaussian with reciprocal width, and translation in one space is a phase in the other. Shifting \( \psi(x) \to \psi(x - a) \) multiplies \( \phi(p) \) by \( e^{-ipa/\hbar} \); the translation operator is \( e^{-i\hat pa/\hbar} \).

Time Evolution: Schrödinger and Heisenberg Pictures

With a time-independent Hamiltonian, \( \lvert\psi(t)\rangle = \hat U(t)\lvert\psi(0)\rangle \), with \( \hat U = e^{-i\hat Ht/\hbar} \) unitary. In the Heisenberg picture states are fixed and operators carry the time dependence, \( \hat A_H(t) = \hat U^\dagger\hat A\hat U \), obeying

$$\boxed{\frac{d\hat A_H}{dt} = \frac{i}{\hbar}[\hat H,\hat A_H] + \left(\frac{\partial\hat A}{\partial t}\right)_H}$$

Taking expectation values gives the same law in either picture:

$$\frac{d\langle A\rangle}{dt} = \frac{i}{\hbar}\langle[\hat H,\hat A]\rangle + \left\langle\frac{\partial\hat A}{\partial t}\right\rangle$$

So an observable with no explicit time dependence that commutes with \( \hat H \) is conserved. Ehrenfest’s theorem is this equation with \( \hat A = \hat x \) and \( \hat A = \hat p \).

Applied to \( \hat A = \hat x\hat p \) in a stationary state, where every \( d\langle A\rangle/dt \) vanishes, it gives the virial theorem:

$$2\langle T\rangle = \left\langle x\frac{dV}{dx}\right\rangle$$

For \( V \propto x^2 \), \( \langle T\rangle = \langle V\rangle \); for a Coulomb potential (in 3D, \( \langle\mathbf r\cdot\nabla V\rangle \)), \( 2\langle T\rangle = -\langle V\rangle \), so \( \langle T\rangle = -E \).

Parity

The parity operator \( \hat\Pi\psi(x) = \psi(-x) \) is Hermitian and unitary with eigenvalues \( \pm1 \), and \( \hat\Pi\hat x\hat\Pi = -\hat x \), \( \hat\Pi\hat p\hat\Pi = -\hat p \). If \( V(x) = V(-x) \), \( [\hat\Pi,\hat H] = 0 \) and non-degenerate eigenstates have definite parity. Then \( \langle x\rangle = \langle p\rangle = 0 \) in each such eigenstate, and matrix elements of odd operators vanish between states of equal parity. That is the origin of the parity selection rule in Perturbation Theory and the Variational Method.

Memory Tricks

Don’t Confuse

ThisvsThat
\( \langle\phi\vert\psi\rangle \): a number\( \lvert\psi\rangle\langle\phi\rvert \): an operator
Overall phase: unobservableRelative phase: changes probabilities
Commuting observables: simultaneous eigenbasisNon-commuting: uncertainty bound may still be zero in some states
\( \sigma_x\sigma_p \geq \hbar/2 \) (exact, state-independent bound)\( \Delta E\Delta t \gtrsim \hbar/2 \) (time is a parameter, not an operator)
\( \hat x = i\hbar\,\partial_p \) in momentum space\( \hat p = -i\hbar\,\partial_x \) in position space (opposite sign)
Schrödinger picture: states evolveHeisenberg picture: operators evolve; same expectation values

Common Mistakes

Practice

All questions below are original practice questions written for this page. They are not from any CSIR NET paper.

Level 1: Foundation

  1. Evaluate \( [\hat x,\hat p^2] \).
  2. Can a Hermitian operator have the eigenvalue \( 2i \)?
  3. What is \( \hat x \) in the momentum representation?
Answers
  1. \( 2i\hbar\hat p \). \( [\hat x,\hat p^2] = [\hat x,\hat p]\hat p + \hat p[\hat x,\hat p] = 2i\hbar\hat p \).
  2. No. Hermitian eigenvalues are real. If \( \hat A\lvert a\rangle = a\lvert a\rangle \), then \( a\langle a\vert a\rangle = \langle a\rvert\hat A\lvert a\rangle = \langle a\rvert\hat A\lvert a\rangle^* = a^*\langle a\vert a\rangle \), because the expectation value of a Hermitian operator is real. So \( a = a^* \).
  3. \( i\hbar\,\partial/\partial p \).

Level 2: Part B Style

Practice Q1. For \( \hat H = \frac{\hat p^2}{2m} + V(\hat x) \), the commutator \( [\hat x,\hat H] \) equals:

(a) \( i\hbar\hat p/m \) (b) \( -i\hbar\hat p/m \) (c) \( i\hbar\hat p/2m \) (d) \( i\hbar V'(\hat x) \)

Answer: (a)

\( [\hat x,V(\hat x)] = 0 \) and \( [\hat x,\hat p^2] = 2i\hbar\hat p \), so \( [\hat x,\hat H] = \frac{2i\hbar\hat p}{2m} = \frac{i\hbar\hat p}{m} \). This gives \( d\langle x\rangle/dt = \frac{i}{\hbar}\langle[\hat H,\hat x]\rangle = \langle p\rangle/m \).

Practice Q2. With \( A \) and \( \lvert\psi\rangle \) as in the worked example above (\( A \) has a 2 in the top-left and the Pauli-\( x \) block below; \( \lvert\psi\rangle = (1,1,1)^T/\sqrt{3} \)), the measurement is made and the result \( +1 \) is obtained. The state immediately afterwards is:

(a) \( (1,1,1)^T/\sqrt{3} \) (b) \( (0,1,1)^T/\sqrt{2} \) (c) \( (0,1,-1)^T/\sqrt{2} \) (d) \( (1,0,0)^T \)

Answer: (b)

The eigenvalue \( +1 \) is non-degenerate with eigenvector \( (0,1,1)^T/\sqrt{2} \). The state collapses onto it. A repeat measurement of \( A \) now gives \( +1 \) with certainty.

Practice Q3. For the ground state of an infinite square well, \( \sigma_x\sigma_p \) is approximately:

(a) \( 0.50\hbar \) (b) \( 0.57\hbar \) (c) \( 1.14\hbar \) (d) \( 0.32\hbar \)

Answer: (b) 0.57ħ

For the well \( 0 < x < L \), \( n = 1 \): \( \langle x\rangle = L/2 \), \( \langle x^2\rangle = L^2\left(\frac{1}{3} - \frac{1}{2\pi^2}\right) \), so \( \sigma_x^2 = L^2\left(\frac{1}{12} - \frac{1}{2\pi^2}\right) \). Also \( \langle p\rangle = 0 \), \( \langle p^2\rangle = \pi^2\hbar^2/L^2 \).

\( \sigma_x\sigma_p = \pi\hbar\sqrt{\frac{1}{12} - \frac{1}{2\pi^2}} = \frac{\hbar}{2}\sqrt{\frac{\pi^2}{3} - 2} \approx 0.568\hbar \). Above the bound \( 0.5\hbar \), as it must be: the sine is not a Gaussian.

Practice Q4. For \( \psi(x) = \frac{1}{\sqrt{a}}e^{-\lvert x\rvert/a} \), the momentum probability density \( \lvert\phi(p)\rvert^2 \) is proportional to:

(a) \( e^{-\lvert p\rvert a/\hbar} \) (b) \( \frac{1}{1 + p^2a^2/\hbar^2} \) (c) \( \frac{1}{(1 + p^2a^2/\hbar^2)^2} \) (d) \( e^{-p^2a^2/\hbar^2} \)

Answer: (c)

\( \phi(p) \propto \frac{1}{1 + p^2a^2/\hbar^2} \) (a Lorentzian, from the Fourier transform of a two-sided exponential), so its square is the squared Lorentzian. Option (b) is \( \phi \) itself, not \( \lvert\phi\rvert^2 \).

Level 3: Part C Style

Practice Q5. A spin-1/2 particle is in the eigenstate of \( S_z \) with eigenvalue \( +\hbar/2 \). Consider the statements:

  1. \( \sigma_{S_x} = \sigma_{S_y} = \hbar/2 \).
  2. The Robertson bound for \( S_x \) and \( S_y \) in this state is \( \hbar^2/4 \), and it is saturated.
  3. In the eigenstate of \( S_x \) with eigenvalue \( +\hbar/2 \), the Robertson bound for \( S_x \) and \( S_y \) is zero.
  4. \( \sigma_{S_z} = \hbar/2 \).

Which are correct?

(a) 1 and 4 only (b) 1, 2 and 3 only (c) 2 and 3 only (d) 1, 2, 3 and 4

Answer: (b) 1, 2 and 3 only

In \( \lvert{+z}\rangle \): \( \langle S_x\rangle = \langle S_y\rangle = 0 \) and \( S_x^2 = S_y^2 = \hbar^2/4 \) (as operators), so both spreads are \( \hbar/2 \). Statement 1 true.

Robertson: \( \sigma_{S_x}\sigma_{S_y} \geq \frac{1}{2}\lvert\langle i\hbar S_z\rangle\rvert = \frac{\hbar}{2}\cdot\frac{\hbar}{2} = \frac{\hbar^2}{4} \), and the product is \( \hbar^2/4 \). Statement 2 true.

In \( \lvert{+x}\rangle \), \( \langle S_z\rangle = 0 \), so the bound is zero (and indeed \( \sigma_{S_x} = 0 \)). Statement 3 true.

\( \lvert{+z}\rangle \) is an eigenstate of \( S_z \), so \( \sigma_{S_z} = 0 \). Statement 4 false.

Practice Q6. For \( \hat H = \frac{\hat p^2}{2m} + mg\hat x \) (a particle in a uniform field), the Heisenberg-picture operator \( \hat x_H(t) \) is:

(a) \( \hat x + \frac{\hat p t}{m} - \frac{1}{2}gt^2 \) (b) \( \hat x + \frac{\hat p t}{m} \) (c) \( \hat x - \frac{\hat p t}{m} + \frac{1}{2}gt^2 \) (d) \( \hat x\cos\omega t + \frac{\hat p}{m\omega}\sin\omega t \)

Answer: (a)

\( \frac{d\hat p_H}{dt} = \frac{i}{\hbar}[\hat H,\hat p_H] = \frac{i}{\hbar}(mg)(i\hbar) = -mg \), using \( [\hat x,\hat p] = i\hbar \), so \( [mg\hat x,\hat p] = i\hbar mg \). Hence \( \hat p_H(t) = \hat p - mgt \).

\( \frac{d\hat x_H}{dt} = \frac{\hat p_H}{m} = \frac{\hat p}{m} - gt \), so \( \hat x_H(t) = \hat x + \frac{\hat pt}{m} - \frac{1}{2}gt^2 \). The operator equations reproduce the classical trajectory exactly, as Ehrenfest’s theorem promises for a linear potential.

Practice Q7. Let \( \hat A \) and \( \hat B \) be Hermitian operators on the same space. Consider the statements:

  1. \( i[\hat A,\hat B] \) is Hermitian.
  2. \( \hat A\hat B \) is Hermitian if and only if \( [\hat A,\hat B] = 0 \).
  3. \( e^{i\hat A} \) is unitary.
  4. If \( [\hat A,\hat B] = 0 \), every eigenvector of \( \hat A \) is an eigenvector of \( \hat B \).

Which are correct?

(a) 1, 2 and 3 only (b) 1 and 3 only (c) 2, 3 and 4 only (d) 1, 2, 3 and 4

Answer: (a) 1, 2 and 3 only
  1. \( (i[\hat A,\hat B])^\dagger = -i[\hat B,\hat A] = i[\hat A,\hat B] \). True.
  2. \( (\hat A\hat B)^\dagger = \hat B\hat A \), which equals \( \hat A\hat B \) exactly when they commute. True.
  3. \( (e^{i\hat A})^\dagger = e^{-i\hat A} \), and the product is \( \hat 1 \). True.
  4. False when \( \hat A \) has a degenerate eigenvalue: then only suitable combinations within that eigenspace are eigenvectors of \( \hat B \). Example: \( \hat A = \hat 1 \) commutes with everything, and every vector is its eigenvector, but not every vector is an eigenvector of an arbitrary \( \hat B \).

Quick Revision Table

IdeaFormula or fact
Completeness\( \sum_n\lvert n\rangle\langle n\rvert = \hat 1 \)
Probability of \( a_n \)\( \lvert\langle a_n\vert\psi\rangle\rvert^2 \)
Canonical commutator\( [\hat x,\hat p] = i\hbar \)
Function commutators\( [\hat x,f(\hat p)] = i\hbar f' \); \( [\hat p,g(\hat x)] = -i\hbar g' \)
Robertson\( \sigma_A\sigma_B \geq \frac{1}{2}\lvert\langle[\hat A,\hat B]\rangle\rvert \)
Minimum uncertaintyGaussian; \( \sigma_x\sigma_p = \hbar/2 \)
Box ground state\( \sigma_x\sigma_p \approx 0.568\hbar \)
Plane wave\( \langle x\vert p\rangle = e^{ipx/\hbar}/\sqrt{2\pi\hbar} \)
Momentum space\( \hat x = i\hbar\,\partial_p \)
Heisenberg equation\( d\hat A_H/dt = \frac{i}{\hbar}[\hat H,\hat A_H] + \partial_t\hat A \)
Virial theorem\( 2\langle T\rangle = \langle xV'\rangle \)

Key Takeaways

What’s Next