at a recent analysis meeting, it was decided that we should adopt a new
monte carlo formalism. in the past, we had been using DGen to generate
cross-section weighted events, then using blastmc to propagate these
events, then using the reconstruction library to reconstruct the
asymmetries.
however, using the reconstruction on monte carlo data is extremely slow.
additionally, as our setup currently stands, variations in theoretical
parameters (e.g. percent of galster in Gen) had to be handled one data set
at a time.
for these reasons (as well as a couple of others), it was decided that we
should change the monte carlo formalism. well, not so much as include
other options. the other that is being added is as follows:
1) use blastmc to generate whitely distributed events for the desired
reaction channel
2) take the events from 1) that pass acceptance constraints and send
into DGen, where cross sections, asymmetries, and whatever else can
be calculated for as many models as desired
3) rather than use the reconstruction explicitly on monte carlo data,
instead just use the reconstruction on real data to find resolutions
and smear the monte carlo results in accordance
anyway, i have been working on step 1), and a bunch of new and updated
files have been checked into CVS. i have written a white generator for
the electroisintegration channel (i.e. d(e,e'pn)). other channels will
follow as time permits and necessity demands. additionally, i have
updated the blastmc library so that it will now, in addition to the
event.coda file, spit out an event.dgen file. for the moment, this file
is simply a text file. if another format is needed, this can be changed,
but, for the present, it seems okay.
to implement the new formalism, do the following:
1) update the following directories in your CVS sandboxes:
blastmc, include/blast, Blast_Params
2) re-make the blast executable
3) go into the fort.99 file and change the KINE distribution style to 20.
if you read through the KINE documentation in this file, you'll see
that this style corresponds to whitely-distributed deuterium
electrodisintegration. other distribution styles will be available
shortly (e.g. d(e,e'd), p(e,ep), etc.)
4) still in the fort.99 file, check out a new command called KEPN
("Kinematic bounds for deEPN") and use it to set up the bounds for the
kinematic variables for electrodisintegration
5) run the blast executable
6) you will see a file called event.dgen. if you open it, you'll see
information on each event.
i'm out of time at the moment, but i'll send an e-mail soon explaining the
format of the data in the file.
aaron
This archive was generated by hypermail 2.1.2 : Mon Feb 24 2014 - 14:07:31 EST