Hi Adrian,
thanks for the email.
PID is NOT reliable. it will not be reliable till the TOF offsets are
straightened out, especially, when the absolute offset between left and
right sector are resolved. right now reconstructed mass spectra are simply
different in left and right sectors. so a proton in left is not the same
as a proton in the right ;)
I ran nsed on 3691 and I have the following ep events by the program:
145, 229, 348, 610, 676, 1037, 1091, 1337, 1553, 1599. 10 of them, not
very shabby compare to 7 with old version. In fact the new version is
constantly getting a 40-50% more tracks than 2_96. I believe with the
proposed new calibration, we will get even more tracks.
these events are all recongnized by the program as havine on electron in
one sector and a proton in the other sector. some electrons are
misidentified as pi-.
unfortunately, 3691 is not among the ones in my batch recrunching so I
can't simple pull out the root file and look. but I believe lrn should
give the same as nsed.
there are a few things you might want to check:
please make sure to update to the head version of both BlastLib2 and
Blast_Params. I realized that the Blast_Params in blast account is not
under CVS. so for quite a long time, the blast.sc_cal in CVS was not the
one used by blast account crunching. this applies also to Epics.map,
electronics.map and a few other environmental files. I have checked in so
that the environmental files are the now same as those used by blast.
also make sure that your reconstruction is using Blast_Params in the head,
not with version 2_96. that is to say, you do not need to use two
Blast_Params, one for blastmc and one for reconstruction. Now everything
should have converged to the same environment directory. I suspect you are
using the correct version of Blast_Params since it ussually seg-faults if
not. so make sure to update. use "cvs up -AC" if you would. this will
force an update with the latest version, save any locally modified files
to .#...., and get an identical copy from CVS.
in the end, when you do configuration, try:
configure -enable-static ... -enalbe-other things you ussually do ...
Chris added this static link option, that will link lrn to individual
object files, not libBlast.so. When one has multiple copies of BlastLib2,
sometimes it lrn could link to the .so file in wrong directory. this
ussually would result in link time error or run time error before the
first line of main, but who knows, it does not hurt to play safe. static
link also makes gdb, gprof easier to use.
42 and 43 are our own generic positive/negative particle ID. this means
program could not identify the particle further than its charge.
finally v3 IS debugged. I can already run show_ep_asym.C on data crunched
with v3. ^_* I crunched through all the runs in RUNLIST.new which contains
about 2 weeks of data, the new lrn never quit.
Chi
On Sat, 3 Jan 2004, Adrian T Sindile wrote:
> Hi, Chi!
> I have a few questions about the new version of BlastLib:
>
> 1) is PID reliable yet? Using V2_96 on the first 2000 events of run 3691 I
> get 7 (seven) e-p coincidence events. Using V3 I get none, for the same
> events and run number;
>
> 2) is the nsed one gets from CVS (I followed Chris' email to get stuff)
> OK? I am asking it because I am able to reproduce its results with my
> cruncher, so if nsed is correct, why don't I get the results I was getting
> with V2_96? Maybe I am doing something wrong - could you verify the same
> run number and tell me if you get results consistent with V2_96?
>
> 3) In the run number I mentioned above, I was getting lots of PID 42 and
> 43 - I did not look too deep (I am sure they are named someplace in the
> codes) but on the CERN webpage there is no particle for 42 and 43... what
> are those particles?
>
> With the hope this might help debugging V3 or at least answering my own
> questions... Thank you!
>
> Adrian
>
> -------------------------------
> Adrian Sindile
> Research Assistant
> Nuclear Physics Group
> University of New Hampshire
> phone: (603)862-1691
> FAX: (603)862-2998
> email: asindile@alberti.unh.edu
> http://einstein.unh.edu/~adrian/
>
>
This archive was generated by hypermail 2.1.2 : Mon Feb 24 2014 - 14:07:30 EST