This is a very interesting and important analysis. The chib/upsilon ratio is very import in undertanding the upsilon and chib productions.
I have one general comment: why don't you use a very loose CL_gamma cut. I wonder if tightening this cut can reduce the background (if you can reduce the background, the chib(3P) mass maybe look better). I've just checked it and saw only a 0.5 Mev improvement, so in comparison with systematic it's insignificant.
Here are my detailed comments:
1)L16: A measurement of the mass of the chib(3p). The meson was recently observed not the mass
2)L40 and other relavant lines: generally by "primary vertex" we mean the PP vertex, here I think you mean the dimuon vertex
3)L43: KL distance larger than 5000
[x ] 4)L48: The trigger requirement is different for 2011 (1.68) and 2012 (2.56?)
Yes, it's 1.48 (2011) and 1.68 (2012)
5)L74: corresponds to one Upsilon(nS) signal repectively.
6)L120-121: In this case the bias and resolution effects of the Upsilon mass from
7)L135-140: I think you can simply refer to the DCB and say only the left tail is used.
I leave it as it is
8)L151-152: how is the ratio of resolution, k chosen, from MC?
From MC. I've added a sentence on it
9)L161-162:So the sum of six ...in these decays ---> Each decay is parameterised with a CB funtion.
10)L168-169: how are these orders of background determined, they are so different (2,4,5).
It's due to background shape at the left side of plot. You can see it on the Figure B.1.
11)L191: I think one explanation why the mass depends on the lambda is good to show here.
"-" I think it's follow from fit description that we could not separate xb1 and xb2 states.
12)L195: make sure the pt distribution in data simulation is similar to allow this parameter fixing
In each bin it's fixed to the value from simulation - so they are different in each pt bin (i added a few words on it)
13)Figure 3.12: how is the mass difference between the chib1(3P) and chib2(3P) determined.
It's fixed to 12Mev as said in section 3.2.
14)L303: is it possible to free lambda in the fit to make a cross check (maybe only for high statistic range)
I've tried it, but have no success.
1. It seems for me that the
mass difference spectra are inconsistent for 2011 and 2012 data
for the signal of chi_b(3P). For example if we look carefully
at fig 3.1 we see rather clean signal of chi_b(3P) for sqrt(s)=7 TeV
and practically no such signal for sqrt(s)=8TeV. This fact is
reflected in table 3.5. If we take the ratio of the yields of 1P, 2P, 3P
and background for 2012 and 2011 we get:
2.43+/-0.11 for 1P, 2.24+/-0.31 for 2P and 1.47+/-0.56 for 3P.
The ratio for background is 2.71+/-0.05.
The difference of this ratio for 3P and 1P is 1.7 sigma.
This effect is also reflected in fig. 3.3 bottom, where
the points for sqrt(s)=8TeV are systematically below the points
for sqrt(s)=7TeV.
For comparison the ratios for Upsilon 1S, 2S and 3S are:
2.33, 2.32 and 2.29+/-0.02.
These are consistent within stat. errors.
For the case of Upsilon(2S) gamma the situation (fig.3.6, table 3.6) is the same:
ratio for 2P is 2.7+/-0.4, for 3P is 1.6+/-0.7. The difference is 1.5 sigma.
the ratio for background is 2.5 which is consistent with 2P ratio.
For Upsilon(3S) gamma the situation is quite different:
the ratio for 3P is 2.5+/-0.7, for background is 2.8+/-0.4.
Please also prepare figs.6.1 and 8.1 for chi_b(2,3P)->Upsilon(1S)gamma
in readable format by putting upper limit of y-axis at around 10%.
2. On page 1 you give formula 1.1 and assume that the efficiency for
Upsilon is canceled. This assumption is fulfilled only if you
apply the same selection cuts for Upsilons from chi_b and for
inclusive Upsilons. In your case there is no cut on mu+mu-
mass for the inclusive Upsilons case but this cut is applied
for Upsilons from chi_b. For example you apply cut
10300<M(2mu)<10526MeV in table 3.2 for Upsilon 3S which corresponds
to -1sigma +3sigma from mean value. In this case the ratio of
efficiencies for the inclusive Upsilon selection to Upsilon
from chi_b is about 1.2. This value can not be neglected in the
analysis. The situation is the same for Upsilon(2S)gamma channel.
[ ] 3. In p.2 l.70 you write that you performed unbinned maximum
likelihood fit. For this case the p.d.f. function used in the
fit should be normalized to unity. In your defintion of the
fitting function in eqn(2.1) the normalization parameter N is
floating. This is allowed for minimum chi2 fit but not for
unbinned maximum likelihood fit. Please check which kind of
fit you really performed.
The same problem is in the section about chi_b yield extraction,
eqn(3.1) and eqn(3.2).
You give the chi2/n.d.f. in many tables with the results of the fits,
please argue that the chi2/n.d.f is a suitable parameter for
the quality of the unbinned maximum likelihood fit
Absract: Are you sure you measured the mass of spin one
state of the chi_b(3P) tripleT in the situation when you can not
separate the spin states?
P.1 l.25 The logics of this statement is not clear, please rephrase it.
Please see also general comment 2.
p.2 l.43 What do you mean by "primary vertex of dimuon candidate"?.
Is it different from mu+mu- intersection point? Is it different
from the primary vertex of the event? Please clarify this because
the cuts on line 44 look strange.
P.2 l.63 Before the use Neural Networks you should demonstrate that
all the variables used by NN are well reproduced by MC or give a reference
where this is shown.
P.3 table 2.1 Cut on chi2 of decaytreefitter is missed.
P.3 l.73 Define the tau parameter.
P.3 l.82-83 Before fixing the parameters at the MC values
you should demonstrate the good description of the data
by MC. Please show in the same plot the mu+mu- mass spectrum
for the data and MC.
P.4 fig.2.1 Please show the pull and give the chi2 and n.d.f. or
other parameter of the fit quality.
P.5 l.89 How did you get the number 3+/-2 MeV? From table 2.2
the mass difference from PDG is -3.3+/-0.3 for sqrt(s)=7TeV and
-4.7+/-0.3 for sqrt(s)=8TeV.
We see significant difference of the mu parameter for the 2011 and
2012 data (more than 10 sigma). Does MC reproduce this difference?
P.6. l.93 The masses for 2011 and 2012 are statistically incompatible,
so you can not combine them by taking mean value of them. This will
give you wrong efficiency of mass band cut for both samples.
P.7 l.102 Finally Upsilons are selected by cut on M(mu+mu-). Before
the description the photon selection please describe this cut.
P.8 l.112 respect -> with respect
P.9 eqn(3.2) From the definition of the CB function the
parameter alpha should be positive. You fix it at negative value.
Please explain it.
P.9 l.142 Before fixing the parameters at MC values you
shoul demonstrate the good description of the data by MC.
P. 11 Fig.3.1 Some discussion on the difference in the
chi_b(3P) mass range for 2011 and 2012 is needed (see
general comment 1.).
P.13 l.189-191 It seems the explanation of the effect of the
shifting of the visible mass is not satisfactory because
the same effect of mass shift with P_t is seen for Upsilon(nS)->mu+mu-.
P.16 fig.3.8 The legend does not correspond to the points, please check.
P.17 fig.3.9 The legend does not correspond to the points, please check.
P.18 fig. 3.10 The errors of the points are asymmetric. What is meannig of this
asymmetry? Please give some explanation.
In the shown pull the errors of the points are different from unit and from each other?
How it can be explained?
P.18 table 3.7 This table should be identical to the one at page 56 tabel D.1 but
we see the different values for sqrt(s)=7TeV.
P.18 l.221 The values 10517+/-4 and 10505.5+/-2.4 are statistically inconsistent, so
you can not combine them into 10508+/-2(stat.).
P.19 l.226 Again the combination to 10508 is not allowed.
P.19 fig.3.12 Please check the errors of the points in the pulls.
P.24 l.266 and P.25 fig.5.4 Are you sure that you have left (not right) tail in
the mass difference distribution?
P.27 fig.6.1 Please redraw the figures with the proper scale of y-axis.
P.44 fig.8.1 Please redraw the figures with the proper scale of y-axis.
P.60-64 tables E.1-6 In these table the efficiencies for chi_b2 are systematically lower than
for chi_b1. From our previous study we expect the opposite behaviour.
This is due to the fact that the mass of chi_b2 is about 20 MeV higher than of chi_b1 and
hence the photons from chi_b2 are more energetic and have higher efficiencies for registration.