Hoi,

I used the command e.g.

  skimData -g 1620000-1799999 --tableprefix objy --s SP -m "B+B- generic"

to obtain the numbers in the table below on 01/08/03. I used the
runnumbers as given in

  http://www.slac.stanford.edu/BFROOT/www/Computing/Offline/Production/run_ranges.html

for the division into years:

  2000 with bug 600000-759999
  2000 w/o bug 1200000-1344999

  2001 with bug 770000-1009999
  2001 w/o bug 1370000-1609999

  2002 w/o bug 1010000-1199999 1620000-1799999

As of today, I find the following numbers of events:
=========================================================
MC <10.3.1a B+ (10^6 events) B0 (10^6 events)
=========================================================
2000 9.6 9.3
2001 24.9 24.9
---------------------------------------------------------
Total 34.2 34.2
=========================================================
MC >=10.3.1a B+ (10^6 events) B0 (10^6 events)
=========================================================
2000 22.2 22.2
2001 34.6 37.3
2002 56.6 54.7
---------------------------------------------------------
Total 113.4 114.2
=========================================================

The lumis for the data is
  2000, 1900V 10.6/fb
  2000, 1960V 9.6/fb
  2001, 1930V 35.6/fb
  2002, 1930V 25.6/fb

This amounts to 81.4/fb (on resonance), which I take to imply 90x10^6
BB pairs. The lumis are in a ratio of 1/1/3.5/2.5 or (in integers)
2/2/7/5. I do not understand Concezio's

 > Luminosity ratios: 2:1:6:3

>From this one can see that currently we have too much 2002 MC. The
2000 data should always be split 1:1 into 1900 and 1960 conditions.

If we want to
 o replace the runs with <10.3.1a
 o have 3x data statistics in MC

we would need an additional 42x10^6 BBbar events:
=========================================================
new MC B+ (10^6 events) B0 (10^6 events)
=========================================================
2000 8.5 8.5
2001 12.5 12.5
---------------------------------------------------------
In this minimal scenario the total sample is not properly lumi-mixed
since we still have too much 2002 MC.

This can be remedied with the following scenario (removing part of the
2002 MC):
=========================================================
new MC B+ (10^6 events) B0 (10^6 events)
=========================================================
2000 11.0 11.0
2001 24.0 21.0
2002 -12 -10
---------------------------------------------------------
This would give a total sample of
=========================================================
new MC B+ (10^6 events) B0 (10^6 events)
=========================================================
2000 33.2 32.2
2001 58.6 58.3
2002 44.6 44.7
---------------------------------------------------------
Total 136.4 135.2

which is quite close to the proper lumi-weighting. This is also quite
close to the 60x10^6 events needed posted by Concezio.

If we do not want to remove 2002 MC, we would need something along the
lines of:
=========================================================
new MC B+ (10^6 events) B0 (10^6 events)
=========================================================
2000 21.8 21.8
2001 43.4 40.7
---------------------------------------------------------

This picture will change somewhat with the additional 30x10^6 events
to be swept tomorrow, but since I don't know how much of those will be
BB, I cannot include them in my calculation at the moment.

Since I am still a bit jet-lagged, I may have erred somewhere above.

Cheers,
--U.