This page documents pt spectra for WZ in that standard model at NLO, and compares it with LO spectra in standard model
as well as for anomolous couplings that are roughly a factor of 2 smaller than the D0 published limits using their ~300pb-1
- A single k-factor does not adequately desribe the difference between NLO and LO as the ratio of differential cross sections varies as a function of the pt of the Z by about 20%.
- Anomolous couplings affect the pt of the Z distribution most strikingly at large pt, while the difference between NLO and LO is largest at more modest pt. In particular, the core of the pt distribution is going to be quite uncertain due to higher order effects, while the change in the high pt tail is negligible compared to the effect of anomolous couplings. I think this means we can't just blindly fit the spectrum, ignoring NLO effects.
- The integral spectra show that for any pt cut-off our statistical error on the yield with pt larger than that cut-off is large compared to the theoretical uncertainty due to higher order corrections ot the differential cross section. Not clear to me how we can use this to our favor.
I'd say, if we can convince ourselves that the efficiency vs pt curve doesn't depend on aTGC then we might as
well use the NLO curves times the universal acceptance in our fit for aTGC, instead of the differential cross section at LO.
Bottom line, this is all very straightforward if Rami shows that the efficiency vs pt is universal, i.e. independent of
aTGC value. If he shows that there is a significant difference in the acceptance function for different aTGC then the fitting
requires more thought.
Ratio of Standard Model NLO divided by LO pt of the Z spectrum
The legend in these two plots should be selfexplanatory. All spectra are LO, except for the yellow one which explicitly says
in the legend that it is standard model NLO. As overall normalization we chose the number of signal events in our present data sample,
i.e. 13 events. The standard model LO prediction thus sets the scale for the overall normalization while the other curves are normalized relative to according to the total cross section they have as compared to the standard model at LO.
These are the same spectra, except that we integrate them towards pt=\infty . The dashed line is the standard model expectation
scaled up by 1.64 times its statistical error for our present data sample. I.e. at each point we form N+1.64*sqrt(N) to go from SM-LO to the dashed curve.
Comments on Technicalities
MCFM spits out two ntuples when it run in "tota" mode, i.e. when generating NLO. It's not fully clear what these two ntuples are.
In the above spectra we ignored the second of these ntuples because its contribution to the total is negligible, and we aren't sure
what it means anyway. For completeness, here the pt of the Z spectrum for this second ntuple.
- 30 Dec 2006
Topic revision: r1 - 2006/12/30 - 07:13:24 - FkW