From vickey@uiuc.edu Tue Jul  8 19:54:22 2003
Date: Mon, 07 Jul 2003 09:40:00 -0500 (CDT)
From: Trevor Vickey 
To: burkett@fnal.gov, smaria@fnal.gov, williams@williams.hep.upenn.edu,
     castro@fnal.gov, slaughter@fnal.gov
Cc: Tony Liss , catutza@fnal.gov, derrede ,
     doksus , eddy , hskim ,
     jakraus , kpitts ,
     lannon , Lucio Cerrito ,
     leh , serrede ,
     strolog , trj ,
     vickey , catutza@uiuc.edu,
     Chris Marino , anyes@fnal.gov
Subject: Univ. of Illinois comments on your paper draft

Dear all,

The University of Illinois group has reviewed the Kaluza-Klein Graviton
Emission draft paper.  We would like to congratulate both the authors and
the godparents for getting this analysis out.  You will find our comments
listed below.


  General comments:

  The first paragraph is difficult to understand.  What does it mean for
  the "metric field" to be "Fourier expandable in terms of sines and
  cosines of the extra dimension", etc.?  Somehow the start of the paper
  should be understandable without having followed the subject.

  One possibility would be to start the paper by stipulating that there
  are theories of TeV scale gravity.  Then go to the Feynman diagrams
  (which are general to other TeV scale gravity theories, right?).
  Then say that this paper is about ADD gravitons and explain what that
  means.  ADD specifically comes in for the cross section calculation and
  the specific event signature, right?  That would allow the paper to
  start off in a way that was not so opaque.  The reader would know what's
  what before getting lost in the specifics of TeV scale gravity theories
  and extra dimensions. 

  Response
  This is the best way we could briefly and comprehensively introduce what
  a Kaluza-Klein mode is.  Some readers may not be familiar with Kaluza-Klein
  towers, so it is important that we explain this in the paper.  We went 
  through several iterations with the GPs on this before finalizing the first
  draft.  We had agreed that we would not gloss over everything, but write it 
  out as explicitly as possible.  We have tried to improve this section in
  the new draft.
  

  The introduction is too long, only one page focuses on the analysis
  and it suffers because of this.  We would like more discussion of the
  analysis, if necessary at the expense of the introduction.  For example,
  the background removal procedure (see our comments below) which needs to
  be described in this paper, is instead left to reference 11.
  Response
  In response to all of the comments, we have added quite a bit of 
  explanation to the discussion of the analysis.  In other papers, the 
  background removal described in reference 11 is omitted completely.  At 
  least here we mention that these backgrounds are removed and point the 
  reader in the direction of more information.   
        

  Are any of the KK states massive?  Perhaps some of these states decay
  and then the paper needs to mention something about their lifetimes?
  Response
  This sounds like you want a longer introduction.  In discussing the 
  signature we have added a mention that these states are stable and 
  non-interacting.


  Do we need to worry about decaying gravitons?  If not, perhaps mention
  that the gravitions can pass through the detector undisturbed.
  Response
  Not in the model we are using.  Again, we now state that the gravitons
  are stable and non-interacting.


  What does the signal look like?  Consider overlaying the signal missing
  E_T over the background.
  Response
  The signal is a smooth excess over the backgrounds for most of the
  kinematic distributions.  In CDF5151 we have plots (also on our webpage
  at http://harv11.fnal.gov/~burkett/KKG) that show the addition of the 
  signal for M=1TeV and n=2.  The plots with 
  the signal added become too busy to be easily read.  In talks we have 
  emphasized the difficulty of trying to observe the signal in the MET 
  distirubtion.  In the latest draft we have added a sentence to explain 
  this.


  In the paper you use both "standard model" and "Standard Model".
  Response  
  Thanks.  We thought we had uniformly "standard model" as prl requires.


  General comment on the figures:  The text in all of them is TOO SMALL .
  They will be shrunk to single-column width in prl, by which time the
  text will be unreadable. It would be nice to use the same font for
  all labels and keys in Figures 2, 3 and 4.
  Response  
  OK.  We have increased all font and marker sizes, matched the fonts in
  the last figure to those in the previous two, and changed the shading
  of the hatched regions.


  Specific comments:

  Abstract:

  Don't use "C.L" in the abstract.  Write explicitly "confidence level
  (C.L.)" in the abstract and throughout the paper use "C.L.".
  Response  
  OK.  Fixed.


  Page 1:

  The first paragraph is difficult to understand.
  Last sentence: expands as a series of states, say "expressed as a
  series of states"  
  Response  
  We have rewritten the sentence to try to improve it.



  Page 2:

  Second sentence:  "each quantum" instead of "each quanta".
  Response  
  OK.  Fixed.

  In the equation $m^2=\vec{p}_n^2$ it would be better not to use the
  arrow, if we are talking about the 4-momentum.
  Response  
  It is the momentum in the extra dimensions.  It is stated explicitly now.

  It would be nice to mention that the natural units are used (hbar=c=1),
  so that the equation where the effective Planck scale is defined makes
  sense (since length multiplies mass).
  Response  
  We think it is fine, it is an HEP paper

  First paragraph, first sentence:  Remove the comma after "M_D".
  Response  
  OK.  Fixed.

  At the end of the first paragraph after the M_pl equation:  "the
  measured value of M_pl" .  This is a bit of a misnomer.  No one measures
  M_pl .  It depends on fundamental constants that are measured.  It would
  be better to just say "use M_pl = 10^19 GeV/c2" or whatever they
  actually use.
  Response  
  Thanks.  You are exactly right.  Fixed.

  First paragraph, last sentence:  Should R for the n=1 case be 10^11 and
  not "R=10^13 m"?  Reference 5 uses R=10^11 for the n=1 case.
  Response  
  Yes.  Fixed. 

  Third paragraph:  You only list three processes for production of
  Gravitons, there is also the Z* mode:  qqbar->gamma/Z* G.  Consider
  listing all of these processes on a single line, or perhaps put them
  together with the Feynman diagrams in Figure 1.
  Response  
  We have rephrased it as "in the emission of a KK graviton tower plus 
  a hadronic jet"

  G is defined.  How about q and g?
  Response  
  OK.  Fixed.


  Page 3:

  Figure 1 caption:  Remove "Upper", "Middle" and "Lower".  The Feynman
  diagrams are already labeled in the figure itself.
  Response  
  OK.  Fixed.

  Second paragraph:  You don't need to mention that the jets are ordered
  in transverse energy.
  Response  
  We don't believe it hurts and it doesn't take up space.

  Second paragraph, last sentence:  Maybe you want to say the "cone axis"
  instead of "the event vertex, and the cone center"?
  Response  
  OK.  Fixed.


  Page 4:

  First paragraph, first line:  The expression "accelerator-" looks
  strange.
  Response  
  We now say "beam and detector-related..."


  First paragraph, second line:  The background removal procedure, which
  is left to reference 11, needs to be described here.
  Response  
  The first and second cleanup selection of the MET trigger sample has
  been used in several published analyses (stop, sbottom, gravitino, 
  gluino, H(bbar)Z(\nu\bar{\nu}) and some make no mention of it at all.   
  We believe that the space would be better used in describing the removal
  of the SM backgrounds.


  In particular, we have no understanding of how the QCD background was estimated.  We went
  to the CDF note for this analysis but it also offered little information
  on the QCD background.  It's key to any exotic analysis, without it
  there's no sensitivity.  This paper is short.  There's room.
  Response  
  We have added something to describe how this is estimated.  Note that
  the 2to2 QCD background in a monojet final state is small, so the 
  sensitivity for this analysis is for the most part independent of the 
  QCD. You could cut higher in MET and then there is an upper limit on 
  the QCD contribution that is dependent on the resolution uncertainty 
  alone.  We went a step further, keeping a lower MET cut and calculating 
  the QCD background using a direct normalization of Herwig 2to2 (1400)
  to the jet20+jet50 dijets.  The procedure is the same as in the 3-jet 
  case.  It is direct luminosity normalization of the MC using data.  The 
  3-jet case is described in cdf5152 and cdf5222.  In this case we use a 
  sample of dijets in the data.  Further details on this can be found in 
  our note, CDF5151.


  Reference 11 does not exist.  Did you mean volume 87?
  Response  
  The reference has been fixed.

  LO and NLO are not defined.
  Response  
  OK.  Fixed

  First paragraph, last line:  Don't define the K factor as K =
  sigma_NLO/sigma_LO, (usually K factor is factor for higher order
  corrections), maybe you mean sigma_signal/sigma_herwig?  Also, you
  should state that the K factor is applied to the background.
  Response  
  The K factor is the ratio of next-to-leading order/leading order cross
  section, and it is applied to the signal.  The backgrounds have been
  normalized to the data.

  Second paragraph, first line:  "QCD produced events" - This is bad
  nomenclature.  At a hadron collider ALL events are QCD produced.  What
  you mean is something closer to "QCD final states", but perhaps that is
  not exactly right either.
  Response  
  OK.  We now use multijet events.


  Page 5:

  As it stands right now, Table I does not contain any information
  that is relevant to the description of this analysis.  The table
  would be relevant if it included the expected number of signal
  and background events passing each cut, for it would then show
  the relative cut rejection and efficiencies.  We suggest including
  this information in the table or removing the table completely.
  Response  
  In most CDF papers the data selection and reduction is shown.  We
  could choose values of M_D and n to get a representative efficiency,
  (thought it varies with n) and show the signal efficiency after each 
  cut.  We will consider this. 

  Also on page 5, the systematic errors are listed for signal efficiency.
  No discussion about the systematic errors on the background are
  included at all.  Presumably the uncertainties listed in Table II are
  the systematic uncertainties on the expected backgrounds, but we
  have no way of knowing what contributes to these effects or how
  they were estimated.
  Response  
  The significant backgrounds are normalized to the data, so the
  uncertainties come from the luminosity uncertainty.  We have added 
  detail to this in the new draft.
 
  Figure 2 caption:  Just stop after "The distribution is plotted with a
  variable bin size."
  Response  
  We have a conflict of advice here.  Our GPs want it as is


  Figure 4 should be a cross-section instead of number of signal events.
  Also, the x-axis label "M?D!(GeV)" is too close to the axis.
  Response  
  Since the limit is for a specific model, we do not want people to use
  cross section numbers and apply them to other models.  Given that it is
  a model, and CDF specific, the acceptance times efficiency is hidden
  if you give a straight cross section number. If we had set a model-
  independent limit, say as a function of the MET cut, we would use the
  cross section.  We will discuss if there is a better way to show it,
  or maybe display it in a table.  


  Second paragraph:   PYTHIA MC is used to model the signal.  Are there
  kinematic effects due to the fact that the graviton is spin 2?  If so,
  are these effects modeled in PYTHIA?  On the cross section calculations
  - what is the uncertainty in the theoretical predictions?  There is
  nothing indicated in Fig. 4 .  The limit should be taken at the lower
  end of the theory uncertainty (which is typically grossly underestimated
  anyway).

  Response  
  The spin 2 effects of the graviton are more important for virtual graviton
  searches.  As for the theoretical uncertainty, the variations on the Q^2 
  scale directly vary the output spectrum of the emitted graviton, so that 
  is included in our uncertainty on the acceptance.  (See our answer to your 
  later comment regarding the combination of the uncertainties.)  Also, recall 
  that we focus on the limits for K=1.0, so we are being safe, if not 
  extra-conservative.

  Anything about systematics on the backgrounds?  Flush out the background
  systematics.
  Response  
  The most significant backgrounds have been normalized to the data, so
  their largest uncertainty comes from the luminosity uncertainty.  There
  is an additional uncertainty on the QCD background, which is much smaller.
  (See also our replies to Andrea Castro's comments.)  We have added a 
  discussion of these uncertainties in the new draft.

  C.L. is still not defined.
  Response  
  Fixed according to your earlier suggestion.


  Second paragraph, first sentence:  There are three things here but you
  don't explicitly explain any of them.  Could you explain how using each
  of these, the 95% C.L. upper limit on the number of signal events is 62?
  Response  
  To explain exactly what is done would take much more room than we have.
  The details of the method are in CDF4476, by John Conway and Kaori.  The
  new draft says:
    "Using a Monte Carlo technique to convolute the uncertainty on the 
     background estimate with the relative systematic uncertainty on
     the signal efficiency, the 95 C.L. upper limit on the number of 
     signal events is 62."
  


  References section:

  Reference 11:  Does not exist, did you mean Volume 87?
  Response  
  The reference has been fixed.  Thanks. 


  Again, we would like to congratulate you on getting this analysis out.

			University of Illinois High-Energy Group


  Response  
  Thanks very much for reading the draft and helping us out with your comments.