hi
online the new/old libBlasts are .2.18g and 2.18h respectively.
I have reverted to the old, so recent runs are analyzed like that.
The old has can crush if pid is switched on.
Our blast_params is pro2003/blast_param, not the cvs. The sources for
the new are in ~cvs_v2/BlastLib2 (for the old they are "gone" because
I updated to the new tip.)
I made some comparison plots for elastics (coplanarity, zwl-zwr,
ee-eecacl, th_e - th_e_calc, th_p - th_p_calc) and left them in
pro2003/analysis/new_and_old_lib. The relevant definitions are in
..utils/strings.C
NOTE I did not compare the same runs, but I used the same cuts
I think this new libBlast works quite well, it seems better than the old
although we are still waiting for the "good" calib file.
However, for the little data I took the momentum reconstruction seems
worse in resolution. Note I took the new libBlast out after onlineGui
was not giving me right sector tracks but that seems to be another issue
(if at all).
What do you think ? What is the result from the benchmark runs ? I am not
going to recrunch present data until we make a decision about the tag.
Another issue, not related to the new fast fit. We seem to be not centered
at zero e.g. in theta_prot - theta_prot_calc(theta_e). This is not a
resolution effect of course, and it was also in the old lib. As anybody
tried to investigate this yet and how to attack it ?? Ideally we should
try it w/ the montecarlo
-- t
________________________________________________________________________________
Tancredi Botto, phone: +1-617-253-9204 mobile: +1-978-490-4124
research scientist MIT/Bates, 21 Manning Av Middleton MA, 01949
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
This archive was generated by hypermail 2.1.2 : Mon Feb 24 2014 - 14:07:29 EST