In trying to answer $\sigma$ transition and angular momentum conservation the following question arose.
Suppose we have an electric dipole $\boldsymbol{d}$ which is oriented and rotating within the horizontal ($xy$) plane. i.e. initially $\boldsymbol{d}$ points along $\boldsymbol{x}$, and then a quarter period later it points along $\boldsymbol{y}$.** What will be the resultant radiation pattern? A complete answer would give something like:
$\boldsymbol{E}(r, \theta, \phi) = |E(r, \theta, \phi)|e^{i\Phi(r, \theta, \phi)} \boldsymbol{\epsilon}(r, \theta, \phi)$
That is, it will give the amplitude, phase, and unit polarization vector as a function of the 3 polar coordinates, $(r, \theta, \phi)$.
My guess is that $\Phi$ and $\boldsymbol{\epsilon}$ are independent of $r$, though I'm not actually 100% certain about this. I could imagine the light is emitted such that it seems like it emanates from a point slightly off the z-axis, in which case I think $\Phi$ and $\boldsymbol{\epsilon}$ have a non-trivial dependence on $r$.. I think this would be related to the presence of non-zero orbital angular momentum.
** The case which is usually treated is a dipole oriented along $\boldsymbol{z}$ and oscillating along this axis. This corresponds to a $m=0$ type oscillation while I am interested in $m=\pm1$ type oscillations.