Hi Concezio, I have not gone through the details, but since it's already afternoon for you, here is the location of my logfiles first: /u/b2/kerstin/scra/Ibutestdss_Run12 and /u/b2/kerstin/scra/Ibutestdss_Run4 More later, Kerstin On Wed, 19 Jul 2006, Concezio Bozzi wrote: > Hi Kerstin, > can I have a look at your logfiles as well? One should compare your > results for Run1+2 and Run4 with the fit on the entire Run1-4 period at > > http://www.slac.stanford.edu/~bozzi/scra/Ibufitdss1_nounf_newfixcount/ > http://www.slac.stanford.edu/~bozzi/scra/Ibufitdss1_nounf_newfixcount/fitdss1_nounf_newfixcountfitresults.eps > > The other/vcb ratio on this sample is 0.55. > I would expect the data yield to be higher than the sum of the two > yields of your fits (393+436=829), whereas > > Data 1' bin = 725.552 +- 24.3418 > > The same is true for Vub, whereas Vcb and other seem to be OK: > > Vub 1' bin = 3219.71 +- 59.7701 > Vcb 1' bin = 1926.92 +- 46.0334 > Oth 1' bin = 305.603 +- 19.3678 > > > The interesting point is that the SP6/SP5 yield is different from > the ratio of Run4/Run1+2 luminosities, take e.g. your Vcb 1'bin numbers: > > SP6/SP5 = 1327.69/482.67 = 2.75 > Run4/Run1+2 = 100ifb/80ifb = 1.25 > > > Concerning the fitted components, the fit on the full sample gives: > > Vub comp = 0.105665 +- 0.00958735 > Vcb comp = 0.185907 +- 0.00628845 > Oth comp = 0.102527 +- 0.0113147 > > ratio other/vcb = 0.551496, > which is smaller than both ratios of your fit. > > > The Chi Square of the Fit (per dof) is 4.31296 > NDOF = 7 > > At this point I am puzzled, and tempted to drop SP5 completely since it > seems that the effect due to D** is bigger than any detector-conditions > effect. Or we could have a other/vcb ratio which depends on run period > (not easy to implement in VVF), or fit the run periods and combine the > results (we lose some statistical power due to the mES fits on data). > Any opinions? > > Concezio. > > > Il giorno mar, 18-07-2006 alle 16:04 -0700, Kerstin Tackmann ha scritto: > > Hi Heiko, > > > > > I have not read the details of your posting yet. > > > There is seomthing where we might have an inconsistency: > > > I assume you are using the ratio other/vcb from the fit > > > on data concezio has performed, right? > > No, actually. I am performing non-unfolded 3-parameter fits > > separately for Run1+2 with SP5 and Run4 with SP6. Wasn't > > this what we talked about? > > > > > If so, one needs to determine this ratio separately for > > > the SP5 and Sp6 period first, right? > > That is what I did here effectively. I thought this is what > > we talked about this morning. Would you like me to use these > > numbers and perform 1-bin-unfolded fits on the two samples > > as well? > > > > Cheers, > > Kerstin > > > > > > > > here are the results from the tests concerning the differences in the D** > > > > modeling we talked about this morning. This is done with Concezio's recent > > > > VVF tag and the correction factors are picked up from > > > > corrratiosigpeakmx_newdatadepl.txt. The D** are moved into the "other" > > > > component (-fitdss 1). > > > > > > > > Please find the plots here: > > > > http://costard.lbl.gov/~kerstin/vubunf/testdss_Run12fitresults.eps > > > > http://costard.lbl.gov/~kerstin/vubunf/testdss_Run4fitresults.eps > > > > > > > > By eye, the fits do look different and here are some numbers: > > > > > > > > **Fit on Run 1+2, SP5: > > > > chi2/ndof = 1.64116 (for 7 dof) > > > > ratio other/vcb = 0.296045 / 0.287339 = 1.03 +- 0.17 > > > > (with errors: > > > > Vcb comp = 0.287339 +- 0.0183957 > > > > Oth comp = 0.296045 +- 0.0442523) > > > > > > > > **Fit on Run 4, SP6: > > > > chi2/ndof = 1.90748 (for 7 dof) > > > > ratio other/vcb = 0.117799 / 0.141224 = 0.83 +- 0.13 > > > > (with errors: > > > > Vcb comp = 0.141224 +- 0.00785652 > > > > Oth comp = 0.117799 +- 0.0174853) > > > > > > > > where vcb is D and D*, other is D** and the non-sl bkgd. > > > > > > > > The differences might or might not be from the D** - roughly, in SP6 we > > > > would think that the D** should smear down to smaller masses (since > > > > some widths are larger and the minimum masses, with which D** can be > > > > produced can be lower). We subtract less "other" and still the data > > > > spectrum has fewer entries in the bins around 2GeV (and the D** do > > > > smear this far down). > > > > > > > > > > > > Here is another observation from these fits: > > > > > > > > Run 1+2, SP5, numbers of events in the first mX bin: > > > > Data 1' bin = 393.016 +- 26.1089 > > > > Vub 1' bin = 1809.31 +- 45.3825 > > > > Vcb 1' bin = 482.67 +- 22.7482 > > > > Oth 1' bin = 73.6857 +- 9.62637 > > > > > > > > Run 4, SP6, numbers of events in the first mX bin: > > > > Data 1' bin = 436.967 +- 27.2767 > > > > Vub 1' bin = 1528.4 +- 41.7351 > > > > Vcb 1' bin = 1327.69 +- 37.7632 > > > > Oth 1' bin = 186.534 +- 14.9827 > > > > > > > > after the mES fit... compare the number of data events (quite similar) > > > > and the number of Vcb events (quite different). As far as I know there is > > > > no SP5/SP6 weight on the generic MC. In the case that we think > > > > efficiencies might be sensitive to the run period, this looks like a > > > > potential problem, doesn't it? And taking Run 3 into account, this looks > > > > even worse. I am confused - I thought we were supposed to have SP5 and SP6 > > > > roughly in the ratio of Run1-3 to Run4. And if this is true before any > > > > cuts and bkgd subtraction, it probably should not be vastly different > > > > afterwards. > > > > > > > > Cheers, > > > > Kerstin > > > > > > > > > > > > > > > > > > > > >