Skip to content

Harmonic-Oscillator Propagator

For the one-dimensional harmonic oscillator

H=p22m+12mω2x2,H=\frac{p^2}{2m}+\frac12m\omega^2x^2,

the coordinate-space propagator is

Kho(xf,tf;xi,ti)=[mω2πiℏsin⁡ωT]1/2×exp⁡{imω2ℏsin⁡ωT[(xf2+xi2)cos⁡ωT−2xfxi]},\begin{aligned} K_{\rm ho}(x_f,t_f;x_i,t_i) &= \left[ \frac{m\omega}{2\pi i\hbar\sin\omega T} \right]^{1/2} \\ &\quad \times \exp\left\{ \frac{im\omega}{2\hbar\sin\omega T} \left[ (x_f^2+x_i^2)\cos\omega T -2x_fx_i \right] \right\}, \end{aligned}

where T=tf−tiT=t_f-t_i and sin⁡ωT≠0\sin\omega T\ne0. At the caustic times T=nπ/ωT=n\pi/\omega, this expression must be interpreted with the correct phase prescription or replaced by the corresponding delta-function limit.

The oscillator propagator is special because the semiclassical expression is exact: a quadratic action makes all fluctuation integrals Gaussian.

For a general potential, an exact coordinate-space kernel is rarely available. The harmonic oscillator is an exception because both the Hamiltonian and classical action are quadratic. This means:

  • the Schrödinger equation is exactly solvable by Hermite functions;
  • the path integral is Gaussian;
  • the classical action controls the exact phase;
  • coherent states remain coherent under time evolution;
  • each normal mode of a free quantum field behaves like an oscillator.

The canonical oscillator spectrum and wavefunctions are reviewed in Quantum Harmonic Oscillator.

For a time-independent Hamiltonian, the propagator has the spectral form

K(xf,T;xi,0)=∑n=0∞ψn(xf)ψn∗(xi)e−iEnT/ℏ.K(x_f,T;x_i,0) = \sum_{n=0}^{\infty} \psi_n(x_f)\psi_n^*(x_i) e^{-iE_nT/\hbar}.

For the harmonic oscillator,

En=ℏω(n+12),E_n=\hbar\omega\left(n+\frac12\right),

and

ψn(x)=12nn!(1πℓ2)1/4Hn(xℓ)e−x2/(2ℓ2),ℓ=ℏmω.\psi_n(x) = \frac{1}{\sqrt{2^n n!}} \left(\frac{1}{\pi\ell^2}\right)^{1/4} H_n\left(\frac{x}{\ell}\right) e^{-x^2/(2\ell^2)}, \qquad \ell=\sqrt{\frac{\hbar}{m\omega}}.

Thus

Kho(xf,T;xi,0)=e−iωT/2∑n=0∞ψn(xf)ψn∗(xi)e−inωT.\begin{aligned} K_{\rm ho}(x_f,T;x_i,0) &= e^{-i\omega T/2} \sum_{n=0}^{\infty} \psi_n(x_f)\psi_n^*(x_i) e^{-in\omega T}. \end{aligned}

The summation is evaluated using Mehler’s formula for Hermite polynomials. With the convergence prescription e−iωT→e−iωT−ϵe^{-i\omega T}\to e^{-i\omega T-\epsilon}, the sum gives the closed form above.

The factor e−iωT/2e^{-i\omega T/2} is the zero-point-energy phase. Dropping it gives the wrong kernel.

The classical oscillator path joining (xi,ti)(x_i,t_i) to (xf,tf)(x_f,t_f) has action

Scl(xf,T;xi,0)=mω2sin⁡ωT[(xf2+xi2)cos⁡ωT−2xfxi].S_{\rm cl}(x_f,T;x_i,0) = \frac{m\omega}{2\sin\omega T} \left[ (x_f^2+x_i^2)\cos\omega T -2x_fx_i \right].

The mixed second derivative is

−∂2Scl∂xf ∂xi=mωsin⁡ωT.-\frac{\partial^2S_{\rm cl}} {\partial x_f\,\partial x_i} = \frac{m\omega}{\sin\omega T}.

Therefore the kernel can be written as

Kho=[12πiℏ(−∂2Scl∂xf ∂xi)]1/2eiScl/ℏ,K_{\rm ho} = \left[ \frac{1}{2\pi i\hbar} \left( -\frac{\partial^2S_{\rm cl}} {\partial x_f\,\partial x_i} \right) \right]^{1/2} e^{iS_{\rm cl}/\hbar},

again with the appropriate phase branch. This is the one-dimensional Van Vleck form, and for a quadratic Hamiltonian it is exact rather than merely semiclassical.

For small TT,

sin⁡ωT≈ωT,cos⁡ωT≈1.\sin\omega T\approx\omega T, \qquad \cos\omega T\approx1.

The oscillator kernel reduces at leading order to the free-particle kernel:

Kho(xf,T;xi,0)∼(m2πiℏT)1/2exp⁡[im(xf−xi)22ℏT].K_{\rm ho}(x_f,T;x_i,0) \sim \left( \frac{m}{2\pi i\hbar T} \right)^{1/2} \exp\left[ \frac{im(x_f-x_i)^2}{2\hbar T} \right].

The potential contributes at the next orders in TT. This short-time structure is why the free-particle kernel appears inside time-sliced derivations of path integrals.

The formula above is singular when

sin⁡ωT=0,T=nπω.\sin\omega T=0, \qquad T=\frac{n\pi}{\omega}.

These are focal times of the classical oscillator. At such times,

xf=(−1)nxix_f=(-1)^n x_i

for every classical trajectory that starts at xix_i, independent of the initial momentum. The kernel becomes a delta-function limit:

Kho(xf,ti+nπ/ω;xi,ti)=e−inπ/2δ ⁣(xf−(−1)nxi).K_{\rm ho}(x_f,t_i+n\pi/\omega;x_i,t_i) = e^{-in\pi/2} \delta\!\left(x_f-(-1)^n x_i\right).

The phase e−inπ/2e^{-in\pi/2} is the caustic, or Maslov, phase in this simple setting. More general caustic and stability-determinant issues belong to the semiclassical pages, especially Semiclassical Propagator and Van Vleck Determinant.

Number states evolve by phases:

U(T)∣n⟩=e−iωT(n+1/2)∣n⟩.U(T)|n\rangle = e^{-i\omega T(n+1/2)}|n\rangle.

A coherent state has the number-state expansion

∣α⟩=e−∣α∣2/2∑n=0∞αnn!∣n⟩.|\alpha\rangle = e^{-|\alpha|^2/2} \sum_{n=0}^{\infty} \frac{\alpha^n}{\sqrt{n!}}|n\rangle.

Therefore

U(T)∣α⟩=e−iωT/2∣αe−iωT⟩.U(T)|\alpha\rangle = e^{-i\omega T/2} |\alpha e^{-i\omega T}\rangle.

The packet remains a minimum-uncertainty Gaussian. Its center follows the classical oscillator motion:

xc(T)=2 ℓ Re⁡(αe−iωT),pc(T)=2ℏℓIm⁡(αe−iωT).x_c(T)=\sqrt2\,\ell\,\operatorname{Re}(\alpha e^{-i\omega T}), \qquad p_c(T)=\frac{\sqrt2\hbar}{\ell} \operatorname{Im}(\alpha e^{-i\omega T}).

This is a concrete example of exact quantum evolution reproducing classical motion for selected observables and states, without making the state a classical point particle.

A free quantum field decomposes into independent oscillator modes after Fourier expansion and choice of boundary conditions. Each mode has a Hamiltonian of the schematic form

Hk=ℏωk(ak†ak+12).H_{\mathbf k} = \hbar\omega_{\mathbf k} \left( a_{\mathbf k}^\dagger a_{\mathbf k} +\frac12 \right).

The oscillator propagator is therefore the finite-dimensional ancestor of Gaussian functional integrals in free field theory. The field-theory version adds infinitely many modes, regularization, source terms, and relativistic correlation-function conventions.

  • Using the closed-form expression at sin⁡ωT=0\sin\omega T=0 without the delta-function and phase prescription.
  • Dropping the zero-point phase e−iωT/2e^{-i\omega T/2} in the spectral derivation.
  • Treating the propagator as a probability density rather than an amplitude.
  • Forgetting that the square-root branch changes across caustics.
  • Assuming the exact Van Vleck form holds for arbitrary potentials; it is exact here because the Hamiltonian is quadratic.
  • Confusing coherent-state propagation with energy-eigenstate stationarity.
  • R. P. Feynman and A. R. Hibbs, Quantum Mechanics and Path Integrals, McGraw-Hill, 1965.
  • L. S. Schulman, Techniques and Applications of Path Integration, Wiley, 1981.
  • R. Shankar, Principles of Quantum Mechanics, 2nd ed., Springer, 1994.
  • J. J. Sakurai and J. Napolitano, Modern Quantum Mechanics, 3rd ed., Cambridge University Press, 2020.
  • H. Kleinert, Path Integrals in Quantum Mechanics, Statistics, Polymer Physics, and Financial Markets, 5th ed., World Scientific, 2009.
  1. Show that the oscillator propagator reduces to the free-particle propagator at leading order as T→0T\to0.
Solution

Use

sin⁡ωT=ωT+O(T3),cos⁡ωT=1+O(T2).\sin\omega T=\omega T+O(T^3), \qquad \cos\omega T=1+O(T^2).

Then the prefactor becomes

[mω2πiℏsin⁡ωT]1/2=[m2πiℏT]1/2[1+O(T2)],\left[ \frac{m\omega}{2\pi i\hbar\sin\omega T} \right]^{1/2} = \left[ \frac{m}{2\pi i\hbar T} \right]^{1/2} \left[1+O(T^2)\right],

and the phase becomes

imω2ℏsin⁡ωT[(xf2+xi2)cos⁡ωT−2xfxi]=im(xf−xi)22ℏT+O(T).\frac{im\omega}{2\hbar\sin\omega T} \left[ (x_f^2+x_i^2)\cos\omega T -2x_fx_i \right] = \frac{im(x_f-x_i)^2}{2\hbar T} +O(T).
  1. Verify that the classical action generates the endpoint momenta.
Solution

For

Scl=mω2sin⁡ωT[(xf2+xi2)cos⁡ωT−2xfxi],S_{\rm cl} = \frac{m\omega}{2\sin\omega T} \left[ (x_f^2+x_i^2)\cos\omega T -2x_fx_i \right],

the final momentum is

pf=∂Scl∂xf=mωsin⁡ωT(xfcos⁡ωT−xi).p_f=\frac{\partial S_{\rm cl}}{\partial x_f} = \frac{m\omega}{\sin\omega T} (x_f\cos\omega T-x_i).

The initial momentum is

pi=−∂Scl∂xi=mωsin⁡ωT(xf−xicos⁡ωT).p_i=-\frac{\partial S_{\rm cl}}{\partial x_i} = \frac{m\omega}{\sin\omega T} (x_f-x_i\cos\omega T).

These are exactly the endpoint relations for the classical oscillator trajectory.

  1. Use the number-state expansion to derive U(T)∣α⟩=e−iωT/2∣αe−iωT⟩U(T)|\alpha\rangle=e^{-i\omega T/2}|\alpha e^{-i\omega T}\rangle.
Solution

Start from

∣α⟩=e−∣α∣2/2∑n=0∞αnn!∣n⟩.|\alpha\rangle = e^{-|\alpha|^2/2} \sum_{n=0}^{\infty} \frac{\alpha^n}{\sqrt{n!}}|n\rangle.

Applying U(T)U(T) gives

U(T)∣α⟩=e−∣α∣2/2∑n=0∞αne−iωT(n+1/2)n!∣n⟩.U(T)|\alpha\rangle = e^{-|\alpha|^2/2} \sum_{n=0}^{\infty} \frac{\alpha^n e^{-i\omega T(n+1/2)}}{\sqrt{n!}}|n\rangle.

Factor out e−iωT/2e^{-i\omega T/2} and combine the remaining phase with αn\alpha^n:

U(T)∣α⟩=e−iωT/2e−∣α∣2/2∑n=0∞(αe−iωT)nn!∣n⟩.U(T)|\alpha\rangle = e^{-i\omega T/2} e^{-|\alpha|^2/2} \sum_{n=0}^{\infty} \frac{(\alpha e^{-i\omega T})^n}{\sqrt{n!}}|n\rangle.

This is e−iωT/2∣αe−iωT⟩e^{-i\omega T/2}|\alpha e^{-i\omega T}\rangle.

  1. What is the kernel after one full oscillator period T=2π/ωT=2\pi/\omega?
Solution

For one full period, n=2n=2 in the caustic formula:

Kho(xf,ti+2π/ω;xi,ti)=e−iπδ(xf−xi)=−δ(xf−xi).K_{\rm ho}(x_f,t_i+2\pi/\omega;x_i,t_i) = e^{-i\pi} \delta(x_f-x_i) = -\delta(x_f-x_i).

The unitary is U(2π/ω)=−IU(2\pi/\omega)=-I because every number state picks up the phase e−i2π(n+1/2)=−1e^{-i2\pi(n+1/2)}=-1. The global sign has no effect on a single ray but matters in phase-sensitive comparisons.