# Semiclassical propagator convergence at $t=0$

For harmonic oscillators the prefactor for the semiclassical propagator is $Fe^{iS}$ where $$F=\sqrt {m\omega/{2πi\hbar\sin(ωt)}}$$ and $$S={m\omega[(x_0^2+x_1^2)\cos(\omega t)-2x_0x_1]}/{2\sin(\omega t)}.$$

Suppose there is a wavefunction $\psi(0)$. I want to calculate $$\int dx_0<x_1|U(0,0)|x_0><x_0|\psi(0)> ,$$ which is exact and gives $\psi(x_1,0)$.

Now if $<x_1|U(0,0)|x_0>$ were to be substituted with $Fe^{iS/\hbar}$, at $t=0$, both the exponent and the prefactor $F$ blow up, how could it converge to $\psi(x_1,0)$?

First note that as $t\rightarrow 0$ your propagator evaluates to zero for $x_0 \neq x_1$ because of the destructive phases coming from $iS$. Therefore, the only relevant part of $\psi(x)$ is $\psi(x_1)$, which, assuming the function is smooth at the vicinity, you can take it out of the integral and ignore.
Then, you can carry out the integration with respect to $x_1$, which I believe should bring down a factor of $sin(\omega t)$, which prevents the blowout.