Hi all, I seem to be having problems with submit.pl (in
/nfs/farm/babar/AWG23/clarke1/newfitter1-24/VirVubFitter/fittest).
Using Antonio's instructions (see attached) I am able to find the
background and signal pdf parameters (by ./submit.pl -MC -bkg -lepcut -enr
-dumppar -intpur 0.1 -mxl -30 -mxh 30) but when It comes to running over
signal + background for the scan the fitter returns this error message:
RooRealVar::setRange(mes) new range named 'mesint' created with bounds
[5.27,5.2891]
RooRealVar::setRange(mes) new range named 'bmesint' created with bounds
[5.27,5.2891]
USING CUTS mes> 5.22 && mx > 4.200000 && mx < 5.000000 && q2 > -999.000000
&& q2 < 300.000000 && lepYaSe>0 && ksele == 0 && intpur > 0.100000
Number of entries in dataset: 108
scalefactor= 4.74858e-05
WARNING argusfrac NOT FOUND in paramIP.txt
Error: illegal pointer to class object argusfrac 0x0 791 FILE:fittest.C
LINE:563
*** Interpreter error recovered ***
EXIT
103.510u 7.350s 1:56.15 95.4% 0+0k 0+0io 51467pf+0w.
BTW, I use parameter file paramsIP.txt for parameterforscan.txt. Also I
put the argusfrac values in the paramsIP.txt file so I'm not sure what's
going on?
C.C.
Date: Wed, 26 Jul 2006 17:48:44 +0200
From: Antonio Petrella <[log in to unmask]>
To: Chukwudi Kweku Clarke <[log in to unmask]>
Subject: Some instructions
[ The following text is in the "ISO-8859-15" character set. ]
[ Your display is set for the "ISO-8859-1" character set. ]
[ Some special characters may be displayed incorrectly. ]
Hi Chukwudi
so the goal is to compute the correction factor for different integrated
lumiosity cut: the formula is
(S/P)_MC enriched
------------------ * (S/P)_data depleted.
(S/P)_MC depleted
To compute S/P you can use the macro in fittest directory.
Let's take MC enriched as example and let's suppose the integrated
purity cut is 0.1
The logical steps are:
- find bkg pdf parameter fitting only background MC on entire sample.
Let all parameters floating... the command would be:
./submit.pl -MC -bkg -lepcut -enr -dumppar -intpur 0.1 -mxl -30 -mxh 30
(with -dumppar you will have a dump of the results of the fit on a txt
file).
- control that the fit is good ie. the error matrix is ok and fitted
values are all ok.
- find signal pdf parameters fitting only signal MC on entire sample.
Let all paraemters floating. The command is:
./submit.pl -MC -sig -lepcut -enr -dumppar -intpur 0.1 -mxl -30 -mxh 30
- control that the fit is good ie. the error matrix is ok and fitted
values are all ok.
Now take the two txt files with dumped parameters and create a new one
by copying the signal stuff from the signal dumped parameter file and
background related stuff from the background dumped parameter file.
You will use this new file as input parameters for scan in bin of mx.
Let's assume that you call this file "parameterforscan.txt"
set to fixed (1) in this file the following parameters:
Thosig_r, Thosig_xc, sigma_r2, Thosig_n, Thosig_alpha, mean_cb,
sigma_cb, alpha_cb, n_cb, cbendpoint.
Now you are ready to run the scan (allcut now):
./submit.pl -MC -all -allcut -enr -intpur 0.1 -parfile
parameterforscan.txt -mxscan
- verify that the fits are good (look the plots and also the output of
the fits).
- look into the txt file with results: you will have also the S/P +-
error printed in this file (look also README, section 5- OUTPUT FILE).
So you have S/P for MC enriched. Do the same procedure for MC depleted
and data depleted. At the end you can compute the correction factor.
One more comment: instead of fitting MC all compontents, you can take
the P values doing a scan on background only
(./submit.pl -MC -bkg -allcut -enr -intpur 0.1 -parfile
parameterforscan.txt -mxscan),
and the S values counting events in the signal dataset.
You can find this number on the shell output of the fit when running
./submit.pl -MC -sig -allcut -enr -intpur 0.1 -parfile
parameterforsca.txt -mxscan). In this case you avoid fitting all MC.
I hope everything is clear.
Let me know, if you have problems I can read emails from time to time...
Bye
Antonio
|
