Q: How much improvement does using a Selection Artificial Neural Network in
two dimensions (S.A.N.N.2) to fit the data as opposed to the dijet mass?
A: We have seen that using a S.A.N.N.2 is the equivalent of 2.5 x more data versus a dijet mass peak.
Q: What contributed to the improvement of your expected limit that you had with the same amount of data during the
summer 2006 analysis.
A: There are there main contributions:
We are now using a Missing transverse energy Projection Dijet Fitter (M.P.D.F.) to correct jet energies for the
missing energy in the event. We find that this improves the ZH dijet resolution from 16% to 10%. Also, the found dijet masses more
accurately reflect the mass of the Higgs Bosons used in MC simulations.
After correcting jet energies by MPDF subscribed amounts, we adjust all possible input variables used during S.A.N.N.2 optimization.
The reoptimization now emphasizes more angular variables of the event versus the energetic variables used in the previous result.
We split one and two b-tagged events. The two tagged events have a much purer signal to background ratio since
the amount of mistagged and c quark events are reduced compared to single tagged events.
Q: Are there any projections for higher luminosities?
A: Based on the current analysis and using a Higgs Mass of 120 GeV/c2 as a baseline,
for integrated luminosities of 2, 4, and 8 fb-1,
we expect an upper limit of 12, 7.6 and 5.2 X SM, respectively. However,
we hope to reduce the systematic errors and increase signal acceptance in time.