The definition of Moliere radius used here is: The radius of a tube centered on the photon momentum vector such that 90% of the DETECTED shower is inside the tube. (i.e. this does not consider any shower leakage through the back of the detectors).
The simulation defined an array of 22 x 22 PbWO4 detectors with dimensions 2.125 x 2.125 x 20 cm3. Photons were sent in normal to the front face of the detectors and centered on a central crystal.
Here is an example of a shower from a single event.
Here is a plot of the energy deposited as a function of distance(in cm) from the initial photon position. R here is in cylindrical coordinates with phi and z integrated over. This contains a large number of events and the y-scale is really arbitrary. The shape is what's important.
This is the integral of the above plot vs. R. The 90% point is used to determine the Moliere radius. The number here seems inconsistant with the plot below (??). Histogram bin widths alone given an error of +/-0.005 cm.
Here is a plot of Moliere radius vs. incident photon energy. This is different from the above method in that the Moliere radius is calculated for every event.
There are two things to notice here:
The average Moliere radius from this plot is around 2.0cm. This differs from the above by about 4mm. I'm not sure where the discrepancy is coming from, but I suspect the one above may have problems with how the information is extracted and so I'd trust this result first. It also agrees better with the value of 2.0cm from the literature.
