Make track variables consistent between trigger and offline
This is following up on #12 (closed): I'm moving every track output to use d0
and z0SinTheta
rather than btagIp_d0
and btagIp_z0SinTheta
. I did some tests to see if it changes anything:
- Dumped a few thousand jets with
h5ls
, which totaled around 22k tracks - looked for the biggest deviations between the two definitions1
The results:
- 102 tracks have any difference for
d0
, the largest is 1e-5 but that's only 5 tracks - 121 tracks have a difference for
z0SinTheta
, the largest is 1e-5 again, with 9 tracks - The difference is always in the last digit I print
I could dig in more, but I'm not surprised there are very small differences (the calculation of the two variables isn't exactly the same). I'm pretty willing to bet there won't be a physically measurable effect from the switch.
-
I used something like
h5ls -ld output.h5/tracks | egrep -o '[^ ]*d0=[^ ]*' | egrep -v nan | egrep -v 'IP.D' | tr '\n' ' ' | tr 'b' '\n' | tr = ' ' | awk '{if ($2!=$4) print $2-$4}' | sed -r 's/^-//g' | sort -n | head
, that's a fun one to work out what it's doing.↩
Edited by Dan Guest