I used a datafile generated by Ken Bloom because it was handy for
a quick test, with results herein. The phi
distribution looks a smidgeon odd, but is the same in both.
Whether this curious phi distribution is merely statistical or
algorithmic remains to be seen. I suspect statistical. Note that
the distribution is the same between both the new (BMU) and old (IMU)
stub finding. The t0 and chisquared distributions
show some curiosities as well. Note that the t0 is the result of a
3-parameter fit, and is thus only sensible for 4-hit stubs.
The West and East BMU t0 distributions peak on different sides of 0,
suggesting an algorithmic issue in the fitting. The IMU t0 distributions
are both offset, which may reflect a genuine issue with the use of
t0 in simulation and reconstruction. The chi-squared distributions are
quite similar between the old and new reconstructions.