Attempt to assign a systematic to the mass dependence of the time variation in the cross section ...

Previous blogs have shown that the mass dependence in the time variation of the cross section is driven by the TPC response and specifically by RDO board 18-1 dying on day 155. We have also seen that the simulation does not mock up the loss of this board well so that the mass dependence persists in the fully unfolded results.

With the cause of the mass dependence well established, the task now is to assign a systematic to this effect. I think the best way to do this is to compare the raw dijet mass yields before and after board 18-1 died and use the size of the difference as the systematic. The problem with this is how to disentangle the effect of board 18-1 dying from the overall drop in yield vs time due to luminosity monitoring problems.

Figure 1: The upper panels show the mass spectra for the the period before (Blue) and after (Red) board 18-1 died. The bottom panels show the ratio of the absolute value of the difference between the two yields divided by the pre board 18-1 yield (this ratio is shown reflected symmetrically about 0). The left column shows the yields and ratios normalized to the same integral and the right column shows the two yields and ratios each normalized by their appropriate integrated luminosity as determined by the BBC.

The bottom panels in figure 1 represent the size of the systematic due to the different spectra before and after day 155. When we normalize by the integral, we see the shape of the systematic, but we remove any change which may be due to reduced efficiency. When we normalize by the BBC integrated luminosity, we can again see the mass dependence, but it is conflated with the known luminosity change between early and late periods of the run.

We need a way to remove the luminosity only component from the systematic shown in the lower right hand panel in figure 1 so we are left only with the mass dependent change. We can do this by looking at the EMC only branch.

Figure 2: The ratio of the absolute value of the difference between the two yields divided by the pre board 18-1 yield for the EMC only dijets. Again, the points above and the histogram below zero are symmetric. The ratio should represent the size of the change in normalized yield due to luminosity differences. The ratio is fit with a P0 between 16 and 49 to determine the size of the luminosity difference.

Figure 3: The systematic seen in the bottom right hand corner of figure 1 with the ratios shifted by 0.17 as was determined from figure 2 by looking at the EMC only branch. This is my estimation of the size of the mass dependent systematic.

Figure 4: Comparison of the mass dependent systematic (Black) to the track inefficiency (Red) and track pt shift (Blue) systematics.

Figure 4 shows that the systematic from the mass dependence only exceeds the systematics from track efficiency and track pt in the first and last result bins (remember that 16-19 and 100-120 are not included in final result). As the effects I saw from board 18-1 dying were a drop in the average number of tracks (and a slight increase in average track pt over the entire TPC) I interperet figure 4 to mean that the mass depencence effect is covered by our current systematics for all but the first and last bin. I propose increasing the systematic in the 19-23 GeV bin to match the black curve above while leaving the other systematics the same. We could also discuss what to do with the last bin, but it looks like statistical fluctuations are larger out here and I'm not sure I trust the magnitude of the systematic in the 82-100 GeV bin.