Angelika has given me the timing for the fill 10507 vernier scan, and I have begun fitting.

After looking at the l2 plots for the barrel for that fill, I don't think there are any particularly hot towers, though one or two are questionable.  In that light, this first pass does not attempt to excise hot towers from the event counter.

Step 1)  plot a histogram of the timestamp of each event.  This is done by looping over the mudst and incrementing the histogram for each event that passed the BHT3 trigger.

Step 2)  find the offset between the t=0 of the step pattern and the t=0 of the run.  This is done by identifying the largest changes in the pattern (when we go from x or y= +/- 1 back to x=y=0) and looking for the offset that produces the largest change in rate across each of those times.  There is a fairly clear peak.

Step 3)  Rebin the timestamps into their scan steps.  Using the aligned step pattern, we integrate the middle twenty seconds of each step.

Step 4)  Fit the scan step histogram with the functional form discussed in previous blog posts, combined with the x and y positions that are known for each step.

The result of this is widths for the blue and yellow beams in each direction, and a variable corresponding to sigma_BHT3*k_b*f_rev*N_blue*N_yellow.  Unfortunately, there seem to be an uneven number of bunches in this fill, so k_b is not a single number.  I will have to think about how to correct this.  Additionally, I don't have a good 'official' number for the frequency of revolution.

The fit doesn't look that great just yet, and so I may need to do some background subtraction after all.