Hi,
this has been a long discussion by now, but I believe that however we do
it, as long as dgen is happy and clear, we'll manage..
Read the readme file in the blastmc directory
dgen generates events for a given I_beam*Delta_T (= integrated charge,
note the beam current is a constant) and a given target thickness. The
values assumed are listed. Also listed is the integrated luminosity in
1/cm2. The montecarlo must obviously assume a certain cell geometry,
target mass (for conductance/thickness calculation). Furthermore:
Luminosity: target thickness x (Integrated Coulombs/1.6e-19)
You can separate these two components from the dgen readme (you have
to calculate 80 mA x 20 hrs). Then the ratio of the histograms
of montecarlo events over data events will be obviously equal to
R = H_mc/ H_data = (MC tgt thickness/ target thickness) * (MC charge/charge)
For the data, a standard Xsection macro like compare_eep.C or compare_ep.C
will give you the total charge (essentially, all the work is done in
init.C, you can easily see how). This number is obtained summing over all
charge files. Note: if a file in the list is skipped, so is its charge file.
Then, once you have R you can easily arrive at the integrated (or average)
target thickness over the set of data files you considered. This you can
compare with naive expectations based on flow, molecular-atomic fractions,
gas temperature, cell geometry.. See also "standard" macros like
compare_ep.C and compare_eep.C
In this sense you normalize data and montecarlo.
Of course you truly arrive at (target thickness) x (recon efficiency),
where you can/should assume recon efficiency=1 for the tofs. You of course
assume the mc has the right elastic cross section folded over the right
acceptance.
For instance: you will compare the montecarlo to polarized and unpol data.
equal pol and unpol flows result in different thicknesses (for abs
is ideally a flow of atoms, for unpol flow of molecular D2,H2). Also,
the cell conductance is for mass 2 and 4, respectively.
Also, for ABS, you could have an uncertainty in how much gas is really
injected into the cell (gas can flow to the sc. chamber otherwise)
Even if you don't know how to calculate the target thickness, you can
always normalize to known detection channels like elastic. Then you can
always apply the same effective normalization factor to data taken
simultaenously on different reaction channels to arrive at a cross
section.
One last note, h_data should be corrected by a factor 1/lvtime due
to daq dead times. In this case lvtime can be easily obtained from
f/lr->Draw(bgbeam/beam,beam>0)
--
________________________________________________________________________________
Tancredi Botto, phone: +1-617-253-9204 mobile: +1-978-490-4124
research scientist MIT/Bates, 21 Manning Av Middleton MA, 01949
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
On Sat, 12 Jul 2003, Aaron Joseph Maschinot wrote:
>
> is it "correct" to normalize the Monte Carlo count rates by the real-data
> charges? if not, is their some way to calculate the MC charge (i.e.
> whatever dgen uses)? the scalers aren't in the MC, so i can't just run
> charge.C on a MC file, i don't think.
>
> aaron
>
>
This archive was generated by hypermail 2.1.2 : Mon Feb 24 2014 - 14:07:29 EST