05/01/2007
Check the Fix sigma 6.5 do the lineshape fit
FCN=-24484.1 FROM HESSE STATUS=NOT POSDEF 92 CALLS 820 TOTAL
EDM=6.77097e-05 STRATEGY= 1 ERR MATRIX NOT POS-DEF
EXT PARAMETER APPROXIMATE INTERNAL INTERNAL
NO. NAME VALUE ERROR STEP SIZE VALUE
1 N 1.17027e+03 3.26428e+01 1.53003e-05 -1.52241e+00
2 Nbkgd1 2.32601e-07 1.01150e+00 3.22687e-05 -1.57080e+00
WARNING - - ABOVE PARAMETER IS AT LIMIT.
3 Nbkgd2 2.54130e-06 1.48565e+00 3.90989e-05 -1.57080e+00
WARNING - - ABOVE PARAMETER IS AT LIMIT.
4 Nbkgd3 8.27289e-05 1.93249e+00 4.44188e-05 -1.57081e+00
WARNING - - ABOVE PARAMETER IS AT LIMIT.
5 NbkgdFlat 8.67463e+00 3.38345e+00 3.57559e-06 -1.56663e+00
6 f1a 4.64593e-01 2.62622e-01 2.08080e-03 1.03209e+00
7 f1b 9.32339e-02 2.97921e-02 2.60178e-04 -5.63124e-01
8 md 1.86943e+00 4.58895e-05 4.26540e-05 2.90204e-01
9 s1a 1.50001e+00 2.39092e+00 5.00000e-01 -1.56714e+00
10 s1b 3.19106e+00 9.22709e-01 7.18106e-04 3.60615e-01
ERR DEF= 0.5
- Use three background components to be zeros.
sigmacommon1=new RooRealVar("sigmacommon1","sigmacommon1",0.0065);
RooRealVar Nbkgd1("Nbkgd1","Nbkgd1",0.0);
RooRealVar Nbkgd2("Nbkgd2","Nbkgd2",0.0);
RooRealVar Nbkgd3("Nbkgd3","Nbkgd3",0.0);
dhad fit floatmass fitparams diag sigmacommon1 6.5 -t s —tag d-m203,203 —qsub
Fit with BES lineshape
dhad bf BES2006lineshape Gamma 0.0286 M 3.7718
05/02/2007
Fit regular BF
- Get sys err file by script
- Found error in the original input file:
- 0010 should be 0.0101
- Do the BF fit and compare with original result
dhad bf
dhad table compare bf original
The comparison looks no different.
Compare BES lineshape result
dhad table compare bf BES2006lineshape/Gamma/0.0286/M/3.7718
post on HNews
05/04/2007
BES2006 linshape fit
dhad fit BES2006lineshape -t d —qsub
dhad fit BES2006lineshape -t d —tag d —qsub
Gamma 0.0286 M 3.7718 fit
dhad fit Gamma 0.0286 M 3.7718 -t d —tag d —qsub
05/07/2007
Gamma 0.0286 M 3.7718 fit
Change the .C file back to normal and resubmit the jobs for double tags
dhad fit Gamma 0.0286 M 3.7718 -t d —tag d —qsub
BES2006 linshape fit
dhad fit BES2006lineshape -t d —tag d —qsub
Update BF fits
check the output file first and backup
05/08/2007
Start the 7.05 version
dhad upgrade 7.05
05/09/2007
New BES lineshape func
Message from Steve:
Good news! I found a bug in our implementation of the BES line shape. (This is good news because it may explain the funny behavior we've seen.) The bug is that the numerator of the Breit-Wigner is taken to be the sum of the D0D0bar and D+D- widths. The numerator should be one or the other, depending on whether we're fitting neutral or charged D's. (Details at the end.) I've made plots comparing various forms of the line shape (for e+e- -> psi(3770) -> D0D0bar). I copied the code directly from the function where it's implemented in RooFit -- Anders' copy of RooDEnergyImp.cc. I plotted the following functions (the number in parentheses is the "mc" parameter that identifies the line shape function): (1) MARK III form (our default) - D0D0bar and D+D- branching fractions taken to be equal (yet another difference between the paper draft and what we actually do), no [1+(r*p0)2]/p03 factors, nonrelativistic BW (0) modified MARK III form - This uses correct D0D0bar and D+D- branching fractions and a relativistic BW. (3) BES line shape (incorrect form, but it's what we've used in our "BES lineshape" fits) - correct D0D0bar and D+D- BFs, relativistic BW, includes [1+(r*p0)2]/p03 factors. Here the second term in the denominator is (M*Gamma)2, as opposed to the modified MARK III form which uses (E*Gamma)2. The numerator of the BW is wrongly taken to be the sum of the D0D0bar and D+D- widths. (4) corrected BES line shape (my addition) - This is the same as above, except that the correct numerator is used. I used the same parameters for all four versions: M = 3.774 GeV and Gamma = 27.5 MeV. All are normalized to equal 1 at E = M = 3.774 GeV. Here are the lineshapes as functions of energy: http://www.lns.cornell.edu/restricted/hypernews/srs63/20070508/3770lineshapes.eps The purple line is covering the black and blue lines. That is, the original MARK III, modified MARK III, and fixed BES line shapes are almost identical, and the incorrect form of the BES line shape (red) is the only one that's significantly different. Here are plots of the ratios to our default, the original MARK III: http://www.lns.cornell.edu/restricted/hypernews/srs63/20070508/3770lineshapeRatios.eps http://www.lns.cornell.edu/restricted/hypernews/srs63/20070508/3770lineshapeRatios_zoomed.eps The original MARK III, modified MARK III, and fixed BES line shapes don't differ by more than 1%. The incorrect form of the BES line shape differs a lot more. We'd expect smaller ISR tails with this incorrect form. This ought to explain why we've seen strange differences with the "BES lineshape," even when the parameters are the same. For those interested, the bug and my fix are shown below. Steve ************************ THE BUG: double gamma=GammaD0D0+GammaDpDm; return gamma/( (e*e-m_mres*m_mres)*(e*e-m_mres*m_mres)+m_mres*m_mres*gamma*gamma); MY FIX: double numerator; if (m_md < 1.867) numerator = GammaD0D0; else numerator = GammaDpDm; return numerator/( (e*e-m_mres*m_mres)*(e*e-m_mres*m_mres)+m_mres*m_mres*gamma*gamma); ------------------------------------------------------------- Visit this HyperNews at CLEOc message (to reply or unsubscribe) at: https://hypernews.lepp.cornell.edu/HyperNews/get/DHadGroup/154.html
- In version 7.03, changed the RooDEnergyImp.cc code and compile:
// return gamma/( (e*e-m_mres*m_mres)*(e*e-m_mres*m_mres)+m_mres*m_mres*gamma*gamma);
double numerator;
if (m_md < 1.867) numerator = GammaD0D0;
else numerator = GammaDpDm;
return numerator/(
(e*e-m_mres*m_mres)*(e*e-m_mres*m_mres)+m_mres*m_mres*gamma*gamma);
gmake -f GNUmakefile.standalone clean gmake -f GNUmakefile.standalone
dhad fit BES2006lineshape1 -t d —tag d —qsub
dhad fit BES2006lineshape1 -t d —qsub
Upgrade to version 7.05
dhad upgrade 7.05
dhad7.05
05/10/2007
Make plots page for bes2006lineshape1
done.
Comparison table
dhad table compare yields BES2006lineshape1 -t d -v1
R=0 for data fit
dhad fit R 0 -t d —qsub
05/11/2007
R=7.5 for data fit
dhad fit R 7.5 -t d —qsub
Compare chisqr for R=0 and R=7.5
dhad table compare chisq R/0 -t d -v1
dhad table compare chisq R/7.5 -t d -v1
Prepare for the 7.05 version
dhad upgrade 7.05
05/12/2007
Prepare for the 7.05 version
dhad upgrade 7.05
Create pages structure
done
05/14/2007
Make Default version 7.05
DHad.py
Get Yields
- Reprocess signal MC
dhad yield regular -t s —qsub
Problem in qsub:
failed opening input/output file because: 05/14/2007 10:08:21 [1821:31141]: error: can't open output file "/home/xs32/work/CLEO/analysis/DHad/ Use "qmod -c <jobid>" to clear job error state once the problem is fixed.
Uncomment lines in qsub() OK.
dhad yield regular -t s —qsub
dhad yield regular -t s —tag d —qsub
- As data and generic don't change, just link them
cd /home/xs32/work/CLEO/analysis/DHad/7.05/dat/evt/regular ln -sf /home/xs32/work/CLEO/analysis/DHad/7.03/dat/evt/regular/data_* . ln -sf /home/xs32/work/CLEO/analysis/DHad/7.03/dat/evt/regular/generic_* .
Post the chisq comparison on HN
chisq comparison for R=0 and R=7.5 Dear all, As David suggested on the last group meeting, I fitted the single tags for data with R=0 and R=7.5 respectively. The related plots can be found at: (Under the section "Single Tag") http://www.lepp.cornell.edu/~xs32/private/DHad/7.03/plots/ I also made tables comparing their chisq with the regular fits which has R=15: (Under section "Comparison" : "chisq") http://www.lepp.cornell.edu/~xs32/private/DHad/7.03/tables/ The trend looks like what we expected, R=7.5 has closer chisq with R=15. Xin
Fit for signal MC
dhad fit floatmass fitparams diag -t s —tag d —qsub
05/15/2007
Plot the params
Create table
dhad table line_shape_parameters
- For Mode 202, we used the version 7.03 value as initial value
s1a=new RooRealVar("s1a","s1a",2.25,1.5,4.0);
f1a=new RooRealVar("f1a","f1a",0.166,0.0,1.0);
s1b=new RooRealVar("s1b","s1b",4.00,1.5,4.0);
f1b=new RooRealVar("f1b","f1b",0.049,0.0,0.4);
dhad fit floatmass fitparams diag -t s —tag d -m202,202 —qsub
Retriev RCS1.1 after done the fit
05/16/2007
Restore the double tag fit code to normal (1.1)
OK.
Message from BES
Thank you for your message. The "r" from the fit is 41+/-458 fm from
the fit mentioned in the paper (hep-ex/0612056). Howevere, the "r"
value from the different fits change a lot. From the fit to the data,
we found that the other parameters are not sensitive to the "r"
value. If we fixed the "r" at a value from several fm to several
thousand fm in the fit, the quantities M_{psi(3770)},
Gamma_{psi(3770)}, Gamma^{ee}_{psi(3770)} and R_{uds} are almost
unchanged.
By the way, Did CLEO-c also make the cross section scan over the
psi(3770) resonance? How may energy points at which CLEO-c collected
the data? And what are the energies?
best regards,
Rong Gang
lineshpae para table
dhad table line_shape_parameters
Single tag fit
dhad fit regular -t s —qsub
dhad fit regular -t d —qsub
dhad fit regular -t s —tag d —qsub
dhad fit regular -t d —tag d —qsub
05/17/2007
Plots for regular fit
done.
Tables for regular fit
dhad table single_signal_eff
dhad table compare version 7.03 single_signal_eff 'Efficiency(%)' Indp
NoFSR yield
dhad yield noFSR -t s —qsub
dhad yield noFSR -t s —tag d —qsub
R= 0 and 7.5 for double tag fit
dhad fit R 0 -t d —tag d —qsub Log
dhad fit R 7.5 -t d —tag d —qsub Log
05/23/2007
Link the Old Dalitz file
signal/olddalitz
dhad yield olddalitz -t s —qsub Log
dhad yield olddalitz -t s —tag d —qsub Log
05/24/2007
Fit Old Dalitz MC
dhad fit olddalitz -t s —qsub
dhad fit olddalitz -t s —tag d —qsub
Plots done.
Tables
dhad table single_signal_eff olddalitz
dhad table compare column olddalitz_single_signal_eff,'Efficiency(%)' 7.03/single_signal_eff
Fit the file with the correct evt file
dhad fit olddalitz -t s —qsub
dhad fit olddalitz -t s —tag d —qsub
05/25/2007
Old Dalitz fit
Refresh plots done.
Table:
dhad table single_signal_eff olddalitz
dhad table compare column olddalitz_single_signal_eff,'Efficiency(%)' 7.03/single_signal_eff
dhad table compare column olddalitz_single_signal_eff,'Efficiency(%)' single_signal_eff
done.