2019-02-25 referees-meeting

Monday 25th February at 14:00 GMT


There was lots of discussion about the sampling. Because the momentum reconstruction now requires TOF2, and a chi2 cut has been added in TKD, the downstream sample is becoming more complicated and needs clarifying; ACTION need to have clear separation of upstream sample; downstream sample; and TOF/momentum selection. This means e.g. modifying Table 3 to make clear what is in the upstream and downstream sample. JN has already started to add a system whereby "cut" events are flagged rather than being completely removed so the removed events can be studied.

Fig. 1 - please include a distance scale
Table 3
  • JN was asked is chi2 cut significant? JN: it is a small cut
  • It was noted that first row of table 3 has the same number for all momenta; why is this? JN: because it comes before the momentum cut, this is the same number.
  • The TOF timing selection now has fewer events in compared to previous iterations; why is this? JN: because the sampling has changed

JN was asked why has the r0 cut changed? JN: has carried out extensive systematics studies on misalignment and revised cuts accordingly

The TOF reconstruction described in Sec. 3.2 was discussed along with Fig. 8. ACTION Sec. 3.2 Delta tof01 in the text should be Delta tof12 (it is a typo). JN was asked about eq 15; he indicated that it is historical. ACTION remove eq 15; Fig. 8b should be updated to reflect the new momentum reconstruction

Fig. 8d has a biased residual in momentum; JN said that this is because the plot was made for all data; ACTION redo the plot for data after the upstream cuts but before the downstream cuts

It was noted that the overflow bins are selected by momentum; so reconstructed momentum is needed for overflow bins, even if they don't hit TOF2. ACTION: use TOF01 for overflow bins

Fig. 2 and 3 - JN was asked why is the distance between projected tracks so different between empty and full data? Should be identical in the thin absorber approximation, which we are close to. ACTION: JN to investigate

Fig. 4a - JN was challenged on the differences between MC and data; the plots are for data with no cuts; ACTION JN will make plots with all cuts except for the one under study applied, as we have for other analyses.

Fig. 6: ACTION should be referenced in the text

Fig. 7 was discussed. It was noted that ACTION units should be [rad]. JN was asked to ACTION please add theta y plots

Tab. 4/5: it was noted that the mean momentum was not consistent (in the decimals). ACTION JN to investigate

Fig. 9-12: it was noted that the TOF selection affects the transverse distribution rather dramatically. Various sources were discussed including diffuser shoulder and dispersion; however there was a suspicion that a geometrical effect could be causing issues. ACTION JN to investigate.

It was noted that the TOF2 efficiency can be 95 % depending on the beam setting. ACTION JN was asked to demonstrate the TOF2 efficiency

Fig. 13: plot showed very nice agreement between data and MC. Label should be MICE internal and should say that this is the 200 MeV/c bin

Fig. 14: noted that the normalisation is quite different between different settings. ACTION JN to investigate

Fig. 16: queried how efficiency is handled in the overflow bin. E.g. tracks added for efficiency should be removed from overflow. ACTION JN said that he would do this.

Fig. 17: why is agreement better in the core than in the tail? JN said that an updated MC was needed. It was noted that the normalisation is different again compared to Fig. 14. ACTION investigate normalisation

Fig. 18:
ACTION please bin as width per mrad. ACTION please check that sum of probabilities is 1. ACTION For all plots, please repeat in MC and add MC to the plot.

JN was asked how the convolution can work with empty bins in tails; JN said that he relies on resampling but did not think empty bins were significant.

JN was asked how efficiency is treated in the convolution. He noted it was not applied in the convolution. ACTION include efficiency in convolution (we hacalculated and used by the deconvolutionve been around this loop before).

Fig. 19:
JN was asked about the outermost bin; JN said that it was just a statistical fluctuation

JN was asked whether the correlated error matrix was avaliable. JN said that he was not sure; it was noted that this may be needed for the chi2 calculation ACTION JN to check.

JN was asked to check whether the procedure conserves probability, including overflow bins. JN was asked to think about normalisation. ACTION: JN to check

Section 5 Systematic uncertainties

JN was challenged on the sytematic uncertainty due to TOF and that this should be an uncertainty in "momentum" rather than "angle". JN noted that an uncertainty exists due to the mislabelling of the empty distribution vs the full distribution; following discussion JN accepted that this was principally an error in momentum rather than angle, and should be treated as such. ACTION: TOF uncertainty should be treated principally as uncertainty on momentum.

It was noted that TOF resolution plots should be added for the empty data as in Fig. 8.

MCSNote.pdf (1.63 MB) Rogers, Chris, 25 February 2019 10:51

Goldsimple.pdf (28.3 KB) Nugent, John , 25 February 2019 13:55