@Thorgott Okay this answers one question but opens two others.
Here (eq3) is the exact form. Two problems: 1. The result, which is stated is
exactly equal to the -ln+arctan term but it should be an approximation (with epsilon). 2. I forgot. I hope I remember.