Hello, a problem of the last few weeks is that lrn.C has never (note:
willingly and with repeated warnings) been put in sync with cvs. This had
to do with various changes online.
We are now approaching a new version of reconstruction (include the new
FF, and the changes made on line such as spin bits..). This per se may not
be enough to recrunch existing data. However the present crunch is still
missing (in the lrn file) much abs info (ligit, dissociator & compression
tube pressure, hf) scaler info and coda info (e.g. the eventclass flag is
not in all lrn ntuples, while we mostly had at least 2 trigger types, also
no coda time info). Note that we may (may not) have a sign error on the
helicity (this will be tested today). Another change is that the charge is
now happily out of lrn.C and the charge info does/can not come from epics.
I'd like to make sure that the "right" lrn gets in cvs possibly today. I
propose to change slightly its definition to have a small scaler and epics
ntuple attached to the lrn file (right now we have only an epics ntuple).
The Epics part could be
Epics(blastps, hfield(2), helicity, dissociator-p, ligit, c-tube, beam
position(2), flow, tcell, time, spare1, spare2) <- for some runs hf and tcell are
not true readings
while Scaler could minimally have
Scaler(tgt_type, tgt_gas, inh_beamcurrent, non_inh_beamcurrent, time,
spare1, spare2).
Of course we already "have" this info but not all in one place and it
would be convenient to have at least a minimal set in lrn. This should not
increase the lrn file too much. Analysis should anyways use the filter
data! (since it is so much faster) and the scaler/epics mini-ntuples could
go to the filtered ntuple as well. With a small change to init.C we can
chain those as well.
I'll be happy to make all of the above changes and all required tests
with v2_20 and then put the updated macros in cvs for private use. Also
investigate why lrn.C hangs on some reconstructions.
Recrunching the past "asymmetry" runs after these changes , would likely
be done in < 12 hrs.
Note:
It was agreed that tps will look back at the dst, it's definition and
format is not clear yet and we are anyways looking for a long term solution.
On the other hand lrn.C has everything you need for physics *now* and that
should be our primary concern (particularly students).
In this sense I don't think we need a dst now and certainly all blast
analysis can survive on lrn.
The lrn code can be easily read.
-- ________________________________________________________________________________ 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