In
http://arxiv.org/abs/1109.4897v2 an alternative analysis of the previous data is mentioned. Quote from page 23:
"
An alternative method to extract the value of delta-t consists in building the likelihood function by associating each neutrino interaction to its waveform instead of using the global PDF. This method can in principle lead to smaller statistical errors given the direct comparison of each event with its related waveform.
"
Now there are 3 results:
The original: σt = (57.8 ± 7.8 (stat.) +8.3/-5.9 (sys.)) ns.
The alternative analysis: σt = (54.5 ± 5.0 (stat.) +9.6/ -7.2 (sys.)) ns.
The short pulse experiment: σt = (62.1 ± 3.7(stat) + 8.3/-5.9 (sys.)) ns
In the latter, I included the systematic error mentioned at page 29:
"
At first order, systematic uncertainties related to the bunched beam operation are equal or smaller than those affecting the result obtained with the nominal CNGS beam.
"
My conclusion is, that the three results are compatible, but I would like to see a more elaborate explanation of the systematic errors, especially for the alternative analysis and the short pulse experiment.
In fact, the alternative method suggests that it relies on the PEW amplitude at the event time minus the TOF.
Note that in this way all events are treated as if they occur at the same time and so it rules out any effect of uneven event spreading, something that cannot be said from the original analysis.
This also greatly reduces the effect of PEW parts not corresponding to the event time minus the TOF, in fact these PEW parts cannot contain any information about the start time of the proton/neutrino that caused the event, so these parts must be considered as noise.
Summing the PEWs around the event time minus the TOF, gives a Gaussian curve, its top indicating the TOF, with a resolution that is intrinsically equal to the 1 ns resolution of the digitizer.
Due to the large time uncertainties, the PDF is expected to be wider than the 5 ns period of the 200 MHz SPS radio frequency. Hence the Gaussian curve will show smaller adjacent Gausian curves, each at a distance of 5 ns,
as a result of the coloured noise due to the mentioned 200 MHz radio frequency.
However, with many events, the curve at TOF should still have the highest value. This leaves little room for greater statistical errors than 1 ns with respect to the PEW timing.
Now, this is all speculation, because the report does not indicate more details of the alternative analysis.
Can anybody tell more about the alternative analysis and the systematic errors of this analysis and the short pulse experiment?
Bert
peraweb.lngs.infn.it/Opera/publicnotes/note132.pdf+DETRMINATION+OF+CNGS+GEODOSY&hl=en&pid=bl&srcid=ADGEESi4fhQbojkl7CX4Hzz3WKCi0SwELICoz_PEUmWTREQsWZ79kpETPexVmWUevXnorAQOZoJFdR2AFMdzciVuF2hGwEILwl9T9eHvxIuF1BlZe4c3dSUDuK2lQA6Hd5qccGq45sx8&sig=AHIEtbTyci-nfj5SZEgAAKm4kShf_3_G8w
