There is something peculiar going on with the angle reconstruction. The limit of theta is affected by the direction of phi. I believe this is because of the way the transform function is implemented.
To see if this is a problem which I introduced with the new tranform function, I reconstructed a subset of showers from 2005 with the old transform function which limits theta to 54.5 degrees:
Different shaped cutoff, but still not straight.
And with another transform function included in the code which limits theta to 63 degrees:
I looked back at some old chiquita data and did not seem to have this problem. It uses the TF which limits theta to 54.5 degrees (.95 radians):
Here is the code that was used for the old simulation above.
This looks like a very even cut-off at .95 radians.
So... where did this start to become a problem?
Checking whether the Log-Likelihood method is a problem... reconstructed a subset of showers with Method 1 (chi-squared), LDF 6, and 54.7-deg transform function.
This looks the same as using this transform function with the LogL code.
Testing as well as I can reproduce old code: Method 1, LDF 5, 54 deg:
Looking back through old data...
November 6, 2006 (simulated chiquita data):
February 15, 2007 (real chiquita data):
