Hi all,
I have run a few comparisions between the old model, the new
model, but as not all jobs run on the same of computer and the CPU scaling
factors are not very accurate, this gives only a rough estimate.
The old model needs around2 hours on a don. The new model needs around a
factor of 2 more time in running. If the endpoint corrections is turned
on, the factor to the old model is around 2.5.
Running the new model with a binned fit, is more or less instantly. Well,
around 10 to 20 minutes.
The only problem is that the results from the binned and unbinned fit does
not agree very well.
Here is an example:
Final number for the unbinned fit:
\def\allsemil{ 77847 \pm 743}
\def\allbgsl{ 4845 \pm 46}
\def\allslsub{73002 \pm 743}
\def\allepssl{ 0.91 \pm 0.03}
\def\allnu{ 257 \pm 47}
\def\allbgc{ 231 \pm 8}
\def\allbgo{ 20 \pm 2}
\def\allepsu{0.401}
\def\allepsmx{0.940}
\def\allbrbr{0.0298 \pm 0.0055}
\def\allbrbrerrmc{0.0020}
Final numbers for the binned fit:
\def\allsemil{ 72937 \pm 752}
\def\allbgsl{ 4580 \pm 47}
\def\allslsub{68357 \pm 752}
\def\allepssl{ 0.90 \pm 0.03}
\def\allnu{ 250 \pm 53}
\def\allbgc{ 217 \pm 8}
\def\allbgo{ 19 \pm 2}
\def\allepsu{0.386}
\def\allepsmx{0.976}
\def\allbrbr{0.0312 \pm 0.0066}
\def\allbrbrerrmc{0.0021}
If I look through some fits, I see sometime huge differences:
Unbinned, data, lepYes==1:
EXT PARAMETER INTERNAL INTERNAL
NO. NAME VALUE ERROR STEP SIZE VALUE
1 B 5.18577e+04 5.08987e+02 3.84504e-05 -1.05592e+00
2 P 4.46709e+04 1.11429e+03 1.41763e-04 -1.09368e+00
3 S 8.16126e+04 7.20654e+02 1.12338e-04 -9.20606e-01
4 ar -1.91128e+01 6.17403e-01 3.95979e-04 9.23130e-01
Binned, the same:
1 B 5.17364e+04 5.22213e+02 1.92864e-04 -1.05654e+00
2 P 4.95944e+04 1.15144e+03 1.35498e-04 -1.06753e+00
3 S 7.61453e+04 7.36531e+02 1.15204e-04 -9.43533e-01
4 ar -1.92077e+01 6.33879e-01 3.95123e-04 9.19643e-01
This is 5000 events differenc in P(peaking background), where the error is
~1000.
Cheers,
Wolfgang
-----------------------------------------------------------------------------
Wolfgang Menges
Queen Mary, University of London SLAC, MS 35
Mile End Road 2575 Sand Hill Road
London, E1 4NS, UK Menlo Park, CA 94025, USA
++44 20 7882 3753 ++1 650 926 8503
[log in to unmask]
-----------------------------------------------------------------------------
