Consider a non-relativistic particle of mass $m$, moving along the $x$-axis in a potential $V(x) = m\omega^2x^2/2$. use path-integral methods to find the probability to find the particle between $x_1$ and $x_1 + dx_1$ if the particle is at $x_0$ at time $t = 0$.

One finds the propagator to be $$P(x_1, t_1; x_0, 0) = \sqrt{{{m\omega}\over{2\pi i\hbar \sin \omega t_1}}}e^{{{im\omega}\over{2\hbar \sin\omega t_1}}((x_0^2 + x_1^2)\cos\omega t_1 - 2x_0x_1)}$$and the requisite probability to be $$\left|\int P(x_1, t_1; x_0, 0)\psi(x_1)\,dx_1\right|^2$$for some wavepacket $\psi$ localized between $x_1$, $x_1 + dx_1$. In the limit $x_1 \to x_0$, we have$$\lim_{(x_1 - x_0) \to 0} P(x_1, t_1; x_0, 0) = \sqrt{{m\omega}\over{2\pi i\hbar \sin\omega T}}e^{{{im\omega}\over{\hbar\sin\omega T}}(x_0^2\cos\omega T - x_0^2)}.$$

My question is, what is the physical significance of taking this limit $x_1 \to x_0$?

  • $\begingroup$ if you want the probability in an infinitesimal interval then you don't need to integrate; the limit then represents the probability of finding the particle between $x_0$ and $\text dx_0$ at time $t$. $\endgroup$
    – Phoenix87
    Jan 22, 2015 at 9:49

3 Answers 3


The probability density $$\lim_{x_1\to x_0} P(x_1, t_1; x_0, 0)$$ is basically the probability density for a particle at position $x_0$ to be back at that position after a time $t_1$ has passed.


As $T = t_1 - t_0 \to 0$, we have$$\begin{align}\lim_{t_1 -~ t_0~ \to~ 0} \langle x_1, t_1\,|\,x_0, t_0\rangle &= \lim_{T~ \to~ 0}\left({{m\omega}\over{2\pi i\sin\omega T}}\right)^{1/2}\text{exp}\left[{{im\omega}\over{2\sin\omega T}}\left(\left(x_1^2 + x_0^2\right)\cos \omega T - 2x_0x_1\right)\right]\\ &=\lim_{\epsilon ~= ~iT/m~ \to~ 0} {1\over{\sqrt{2\pi\epsilon}}}\text{exp}\left[-{{(x_1 - x_0)^2}\over{2\epsilon}}\right] = \delta(x_1 - x_0).\end{align}$$This is just as expected from the orthonormality of $\hat{x}(t)$ eigenvalues at $t = t_0$.


Ideologically speaking, the absolute value of the propagator squared

$$\tag{1} |K(x_f,t_f;x_i,t_i)|^2 \mathrm{d}x_f ~=~ \frac{m\omega}{2\pi\hbar\sin\omega \Delta t}\mathrm{d}x_f, \qquad \Delta t~:=~t_f-t_i~>~0,$$

is the probability that a harmonic oscillator starting at $t_i$ in position $x_i$ will finish within the position interval $[x_f,x_f+\mathrm{d}x_f]$ at time $t_f$.

In particular OP can study the case $x_i=x_f$, i.e. the probability of returning to the same position in a given time $\Delta t$.

Counter-intuitively, according to eq. (1), the probability does not depend on the start and end positions $x_i$ and $x_f$ at all! This foresees the fact that the notion of absolute (as opposed to relative) probabilities of the Feynman kernel $K(x_f,t_f;x_i,t_i)$ cannot be maintained on a non-compact position space, cf. e.g. Ref. 1 and this Phys.SE post.

In general, the probabilistic interpretation of eq. (1) only holds for short times $\Delta t\ll \tau$, where $\tau$ is some characteristic time scale of the system.


  1. R.P. Feynman and A.R. Hibbs, Quantum Mechanics and Path Integrals, 1965.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.