Hi Urs,
I agree that we are seeing MC stat fluctuations. Maybe the final iteration
should be done without kaon killing. For the time being, assuming that
consecutive jobs have similar processID's, I would take the difference
between +/-1 sigma divided by two as error.
would you agree?
ciao
ric
On Mon, 9 Dec 2002, Urs Langenegger wrote:
>
> Hoi,
>
> here is the full list of systematics jobs I ran.
>
> ------------------------------------------------------------------------------------
> j00 default, ursl-112702bis 0.0186079 +- 0.00265021(stat) +- 0.00146837(MC stat)
> j01 trk killing 1.3%/1.6% 0.0190465 +- 0.00266228(stat) +- 0.0014944 (MC stat)
> j02 trk killing 1.3%/2.5% 0.0185922 +- 0.00264268(stat) +- 0.00153336(MC stat)
> j03 no kaon PidKilling 0.0183171 +- 0.0025495(stat) +- 0.0014283 (MC stat)
> j04 no neutrals smearing 0.0184739 +- 0.0025375(stat) +- 0.00140824(MC stat)
> j05 ka eff +2% 0.0182091 +- 0.0025957(stat) +- 0.00145816(MC stat)
> j06 ka eff -2% 0.0185147 +- 0.00259249(stat) +- 0.00142605(MC stat)
> j07 ka mis +10% 0.0194732 +- 0.00262255(stat) +- 0.00142315(MC stat)
> j08 ka mis -10% 0.0193142 +- 0.00258354(stat) +- 0.00140779(MC stat)
> j09 default again 0.0183824 +- 0.00258376(stat) +- 0.00145584(MC stat)
> j10 el eff +2% 0.0188634 +- 0.00256837(stat) +- 0.00142657(MC stat)
> j11 el eff -2% 0.0184217 +- 0.00255032(stat) +- 0.00142972(MC stat)
> j12 el mis +10% 0.0198809 +- 0.00257702(stat) +- 0.00141022(MC stat)
> j13 el mis -10% 0.0195494 +- 0.00257245(stat) +- 0.00139646(MC stat)
> j14 default again 0.0195263 +- 0.0025662(stat) +- 0.00139902(MC stat)
> j15 mu eff +2% 0.0198895 +- 0.00263745(stat) +- 0.00142837(MC stat)
> j16 mu eff -2% 0.0208533 +- 0.00272544(stat) +- 0.00151923(MC stat)
> j17 mu mis +10% 0.0185189 +- 0.00263228(stat) +- 0.00147756(MC stat)
> j18 mu mis -10% 0.0193403 +- 0.00265003(stat) +- 0.00147681(MC stat)
> ------------------------------------------------------------------------------------
>
> I am a bit concerned as to how useful these numbers are. As long as we
> run with kaon Pidkilling, we'll never reproduce the "default" j00, and
> the variations can be ~40% of the statistical error. See j09 and j14
> compared to j00. This might also go into the direction of what Daniele
> has seen in his evaluation of the neutral systematics.
>
>
> Cheers,
> --U.
>
|