% Responses to Toronto
% Version 2.0 Feb 15, 2001

>Dear Authors and Godparents,
>
>Some members of the Toronto group have read the first draft of the
>"Searches for New Physics in Events with a Photon and b-quark Jet at CDF"
>PRD.  We have prepared a list of general and specific comments on your
>paper. We're sorry for the delay in getting these comments to you.
>
>We hope you find these useful. Thank you again for all your hard work,
>
>Dan MacQueen,
>for the CDF Toronto group.
>
>==========================================================================
>
>General comments:
>-----------------
>
>It is not clear what the motivation for this paper is. Instead of
>investigating the relevance of a specific model, you start from a very
>general signature (b + high Et photon) and add several requirements to set
>limit on various supersymmetric models. It follows that your additional
>cuts (for instance at the end of section 2) are unmotivated, which can be
>very confusing for the reader. Instead of starting from such a general
>stance, it might be useful to relate the additional cuts to more specific
>models in the introductory chapter, so that the reader has an idea of what
>you are looking for (given that this paper is a PRD, adding a couple of
>paragraphs to relate the phenomenology of the models to the experimental
>requirements should not be a problem).
>

We have added some motivational prose to the introduction. In addition, we 
have added some references to the growing literature on signature-based 
searches, which may help the reader understand the idea of testing the 
standard model as an hypothesis without relying on specific models of new 
physics.

>It seems that, like the e-e-gamma-MET paper, we are trying to turn a few
>events into the justification for a complete study. It is not clear this
>is enough. We see many parallels between the approach taken here and that
>espoused by our colleagues from Frascati in their "discovery of the
>s-bottom". One important difference here is that we have not discovered a
>signal of any significance, so we are concerned with setting limits. As in
>that case many parts of this paper might be more appropriately authored by
>a smaller subset of individuals rather than the whole CDF collaboration.

We beg to differ. We are looking in specific signatures, with
(as best we can) a priori cuts.  We have not tuned any cuts a posteriori 
to enhance signals, and there is a good physics motivation for looking in 
this channel. In addition, we have not claimed the observation of an
anomaly, or supplied a lengthy theoretical explanation for any effect.   
This is old-fashioned physics- we test an hypothesis (the SM) in a region
where new effects could reasonably be expected to show up, based on possible
hints in the data of where to look. 

>We feel that much of the discussion in sec 4 surrounds events that we end
>up dismissing, thus diluting the main discussion of events which are "more
>significant".

We have added some prose explaining that these events pass our a priori
cuts, and consequently need to be discussed. It's important to state
that there does not seem to be evidence of new physics here, and so one
needs to discuss the probable SM origin of the events. Again, this is
standard procedure: the top PRD discussed the mumugamma event, for example,
as being non-top, although it passed the cuts. 


>The appendix attempts to quantify a difficult subject. It is not clear to
>us that there is much scope for generalised acceptance*efficiency
>calculations. In cases where the method proposed by the authors works it
>should be possible for clever theorists to know what parts of phase space
>a model excludes "safely". In the interesting cases (where on 30%
>exclusion can be achieved in the simplified approach) it is not clear that
>concluding this without the benefit of a full simulation is risky. Should
>the CDF collaboration really be endorsing this?


    Yes, we believe that this can save a lot of work in those cases where it is
not risky. There is a growing literature on this, in particular the work
on Sleuth by Bruce Knuteson  and others at D0, which is along these lines
(if anything, much more radical). CDF has been at the cutting edge of this
work, and it would be a pity not to publish on a topic that is becoming
of great interest.

>Specific Comments:
>------------------
>
>Abstract:
>
>1) It would be a good idea to put some numerical results in the abstract.

    Abstracts should be short- this one is already long. there is no single
number that seems sufficiently more important than some of the others 
to be singled out.

>Section 1: Introduction
>
>1) "Major responsibility" sounds awkward, and seems to place
>the wrong emphasis on things -- perhaps refer to the search for new
>physics as a "major opportunity"?

    Fixed

>2) The first sentence in the second paragraph should be divided in two
>["...heavy particles which can decay into constituents (quarks, bosons,
>etc.)  Due to the large mass..."]

    Fixed

>3) "transveese" -> "transverse".
    Fixed
>
>4) "a disproportionally large branching ratios" should be
>"disproportionally large branching ratios."
    Fixed
>
>5) The statement that "these hypothetical particles have disproportionally
>large branching ratios" is too strong -- this is only true in some models,
>although those are the models you look at.
    Fixed
>
>6) Throughout the paper, you alternate between E_T and E_t for transverse
>energy. It would be better to standardize this to E_t.  As well, you
>should put the symbol for missing transverse energy in italics, rather
>than putting a slash through a Roman E_t.

    Fixed
>
>7) You should explain, expand, or define CDF, MSSM, and GMSB, rather than
>just putting the acronyms in without explanation.

    Fixed
>
>Section 2: Data Selection
>
>1) Section 2.2: "Bremstrahlung" -> "bremsstrahlung".  

    Fixed 
>
>2) In the final sentence of section 2.2, you state that the requirement
>that the vertex be less than 60 cm from the centre of the detector is made
>so the "projective nature of the calorimeters is preserved."  This is
>actually done to ensure the detector contains the entire jet.

   It's both- but you're right, the initial introduction was to prevent
blue-sky events (which, by the way, don't preserve the projective
nature of the calorimeter). Fixed.

>
>3) In section 2.3, paragraph one, you state that you require the EM
>cluster be isolated to remove backgrounds from jets and electrons.  In
>fact, photons can't be found near jets (that is they may be there, but we
>can't look there.)  You should rephrase this to explain what this does for
>the efficiency.

     Perhaps you misread it.  The main background to a photon
is a jet with a large electromagnetic component (the worst is a high-pt
pizero from a jet).  By asking that the EM cluster be isolated we discriminate
against this background. we've looked again at the sentence, and think it's
clear- will you look again?

>
>4) Section 2.3, third paragraph -- "to be less" -> "be less".
     Fixed
>
>5) Section 2.4: B-tagging is a 2-pass algorithm, where the first pass
>looks for two tracks, while the second pass looks for more than two
>tracks.  You have combined the two passes into one sentence here.

    We made this more explicit.
>
>6) Section 2.5: "...one signature is not obviously more likely than any
>other" seems like editorializing.

   Yes, it (partly) is. However, we qualify the sentence with 'in most cases'.
However again, we believe it to be true, if not widely appreciated in
the experimental community (yet). 

>
>7) Section 2.5, second paragraph, first sentence: Missing E_t has already
>been defined, so this sentence isn't needed.

    We don't want to start a sentence with a symbol, and this is of such 
importance that it is a good first sentence for a paragraph; it leads
into what we mean by 'measured Et'.

>
>8) Why did you choose Delta Phi (photon - missing E_t) < 168 degrees,
>rather than some other value?  As well, this optional requirement is
>included in the background, but the other optional requirements are not.

We have added prose to explain this.  This cut was used because it 
showed a clear gain in signal-to-noise.  The other distributions which 
show an advantage such as Met are also used.  The value of the cut is
historical, 15 bins in the delpa-phi plot, cut out the highest bin.

>9) It should be made clear that your H_t is different from the H_t used in
>other analyses.

We state clearly what it is- it's much harder to say what it's not. 
Ht is different in many analyses, including between D0
and CDF In each case one has to state clearly how it is defined, and we do.

>
>Section 3: Background Estimates
>
>1) Is table 2 really necessary? it doesn't provide any additional
>information and can be confusing. Column 2 is labelled "calculation", when
>it really is "method" and the content of the second column is not clear to
>whoever hasn't read the whole chapter (for instance, why using CES+CPR as
>a way to find the mistag+real photons background?).

    We thought it would be helpful, as it's hard to remember how each is
calculated. We have changed 'calculation' to 'method of calculation'.

>
>2) section 3.1: The sentence "are significantly diluted in their
>discrimination power" is not very clear. Why not just say that the
>discrimination power is low at high energy?
   Done (thanks)
>
>3) section 3.3: the last paragraph could be removed.
     The paper is fairly dense, and we thought this was helpful to
the reader.

>
>4) Where does this 20% additional uncertainty for unmeasured effects come
>from? (explain in a sentence or give a reference).
>

  MLM replies today:
  "I don't think we need to explain more about these "additional
  effects", provided you use the sentence I suggested in my letter:
  "... which cannot be determined by simply changing the scale
  dependence." It simply states the fact that varying the scale is a way
  of assessing the potential error induced by the lack of higher order
  corrections, but is no substitute for a full NLO calculation. "

  We had rephrased his sentence, so will go back to quoting him.


>5) section 3.5: this section is redundant with section 3.1 (apart from the
>actual number of photon background). Wouldn't it be better to merge the 2.
>

	Our intention was to summarize the methods 
in the beginning so that the individual sources would be shorter
and not interupted by longish digressions.  We don't like the solution of
moving one piece to the other since it breaks up the organization
either way.

>I don't understand how you evaluate the fake photon background at high Et
>(in section 4, you claim it is an approximation made by applying the fake
>photon measurement and the positive tagging prediction to the large
>statistics untagged sample": is it some sort of extrapolation from the low
>Et sample?).
>

If we take the untagged sample with Met>20 and apply the fake photon 
background method, we have the number of fake photons for all M(b,j).  
We then apply the positive tagging prediction, to arrive at the number
of expected tags.  This procedure can be repeated with cuts on M(b.j)
to find the number of event expected in the boxes. 


>6) In section 3.3, you claim that you expect different "mistag" background
>values in the QCD sample and in the photon+missing Et sample. Can you
>measure or quantify in some way the difference? Then you wouldn't have to
>quote a 50% uncertainty on the mistag background (section 3.6).
>
	Since the backgrounds are not used to set the model limits,
it is hard to justify the effort in renewing the study (particularly
right now, with the run starting), but we agree that it would
be great to investigate this more. It's unfortunately not easy.

>7) section 3.7: "bremstrahlung" -> "bremsstrahlung".
      Fixed
>
>Section 4: Data Observations
>
>1) Table 3: The significant digits are not good here: it should be 0 +/- 5
>rather than 0.3 +/- 5, 0.0 +/- 0.1 +/- 0.1 instead of 0 +/- 0.1 +/- 0.1,
>etc.
>
	Fixed

>2) In Figure 1a, it is difficult to tell the dashed background prediction
>from the dotted SUSY model prediction.
>
	Perhaps we could change to a solid line but then we have to change
all the figures - do you insist? 

>3) In Figures 2b & 2d, you label the y-axis Events/2.1 -- 2.1 what?  Also,
>is the Delta phi between the missing E_t and the nearest jet referred to
>in the text at all?

    Fixed.

>
>4) In tables 4, 5, and 9, we don't think that the run and event numbers
>need to be recorded.
      
Run and event numbers were given in the top evidence PRD and in the
gamma-gamma+X PRD, to name two CDF examples. If one ever wants to revisit this
analysis  sometime in the future, this is the key to finding the events.

>
>5) We're unclear on the motivation behind section 4.2 -- why do you
>discuss four events which don't pass the cuts if the backgrounds are not
>established for them?

	Section 4.2 is about 5 high-mass events- we assume your `four'
is a lapsus digitae. These events DO pass our cuts, and thus the
question of their significance arises. In addition, why do you say the 
background is not established?  
The whole section is about the background.  If we do not
address the events, then you are left with the plot showing the five events
separated from the rest, but left unaddressed. The question will
immediately arise if this clustering is significant. We address this question
using the method that Zeus developed for the events on the tail of
their distribution (which is very similar to the methods subsequently 
developed for Sleuth for D0).


>
>6) In table 8, significant digits are not expressed properly -- 0 +/- 2
>+/- 0.4 should be 0.0 +/- 2.0 +/- 0.4, etc.
>
	Fixed

>7) Last sentence of second paragraph should read "... we note nothing
>else..."
    Fixed
>
>8) In figure 6, you should mention in the caption that the solid dots
>represent the data.  Text inside the figure should also be removed, as
>that information is in the caption anyway.
>
	We often summarize the contents of the plot on the plot itself, so
that when the plot is shown in talks it is self-contained.
Will you grant us this style point?

>Section 5: Limits on Models of Supersymmetry
>
>1) p22. You include a lot of terms that are not defined (or poorly defined)
>in the introduction of this section. Examples include: "model limits",
>"signature-based limits", "gauge-mediated concept", "SUGRA".

       We have explained these a little more and pointed the reader to 
the references.

>
>2) p22, 2nd paragraph of sec. 5.1: Have you every really defined the MSSM
>spectrum of particles in this paper? We looked back at the introduction
>and didn't find it there. Couldn't a few sentences be added to draw the
>parallels between the SUSY charginos and the SM charged bosons?
>3) p22, 3rd paragraph of sec 5.1: By the same token you don't define M_1,
>M_2, beta, mu. Maybe only particle physicists will get this deep into the
>paper, but even for some of them only beta will be an "everyday term".

	
We have published one other SUSY PRD from run 1B and that did not define any
of these terms; it assumed a working knowledge of SUSY.  We started
to add brief definitions and found it is fairly awkward since you just
need more and more jargon and detail. Reference 19 is an excellent
introduction (if we say so ourselves- we are authors) to the field, and
is referenced in this sentence.

>4) p23, 1st full paragraph: Which is the cause and which is the effect.
>You seem to imply that region 2 merits the most attention because Pythia
>models it reliably. Is this because they chose to code it up first because
>there is something more reliable about the physics? That would be a
>stronger argument to use. Basing our choices on what phenomenologists have
>put in their models seems like poor motivation.

     Fixed, rewritten to be more clear.

>5) p23, 2nd full paragraph, 2nd sentence: "To decide where in the region
>to place..." seems to missing a verb. Maybe if you read "=" as "is equal
>to" then you get a complete sentence. But this is not very good English.

     Fixed

>6) p24, line 7&8: "The charginoS and neutralinoS parameters..." You don't
>need the "S"s here.

      Fixed

>7) p24, line 11: It took us a long time to figure out what the physical
>significance of the top-stop threshold was. Another sentence explaining
>how they are neutralino produced (should it be top-anti-stop?) might make
>this a little easier for people who don't deal with this physics every
>day.

	Fixed, rewritten to be more clear.

>8) p24, end of 4th full paragraph: Doesn't the fact that you couldn't come
>up with a set of parameters that doesn't predict a Higgs boson that should
>have already been found at LEP make the "nominal point" of the model much
>less interesting?

      Indeed, it makes our model probably, technically excluded by LEP.
This is exactly one of the reasons to move to signature-based searches.
The same thing happened to other searches where the LSP was inconsistent
with the moving LEP limits.   In these cases the models can be retuned (we
could probably find a point with a heavier Higgs with enough effort)
and the bottom line does not change, we have only spent a few weeks redoing
all the numbers and plots.  The real bottom line is we observe no significant
excesses and no anomalous events.  We would like to spend less time 
on the details of the rest.

>9) p24, 4th paragraph: Haven't we published cc-MET search that Regina and
>Andrei were doing a few years ago? Even in a conference proceeding?

	Yes in PRL, we added this reference.

>10) p25, 2nd full paragraph: Haven't we settled on a 4% luminosity
>uncertainty for all of run I now that the top cross-section has been
>published?

We have recalculated everything based on the new (!) 4% number. It makes
no difference, but we have done it right. (One of us (HJF) notes that E811
and E710 might differ with our faith in the 4%  number).

>Section 6: Model-Independent Limits:
>
>1) In the first paragraph, mention that the script L is integrated
>luminosity.

	Fixed
>
>Section 7: Conclusions
>
>1) The concluding sentence should say what you mean by "with a reliability
>of approximately 30%" -- does this mean a 30% uncertainty?

    It means the difference between the approximate and detailed answers was
less than about 30%.  We changed it to uncertainty.

>
>References:
>
>These need a careful grooming for punctuation. Separate papers in the same
>reference should be separated by semi-colons and end with a period. It
>looks like you know this but have mixed and matched references from other
>places as some end with periods in the middle and others have a semi-colon
>at the very end. See for example: 1, 2, 9, 11 and 17.

    Fixed

>1) "et al." should follow the first author and have a comma after it. cf.19 
>and 20.

     Fixed

>2) When you quote a second paper by the same authors you should include
>the authors name again. cf. 1 and 11.

     A quick check of the style page on 
http://publish.aps.org/STYLE/ms.html#citations
for two sources confirms that what we have done is correct.

>3) Ref 3 should not have the paper title included.

    Fixed

>4) Nucl. Instrum. and Methods Phys. Res., Sect. A should be used
>consistently (cf. 4 (ok), 5 (nice), 7 (could use work), 24 (missing the
>"A") and 25 ...)

      Fixed

>5) Couldn't ref 15 just be a footnote, or better yet included in the text?

    Please grant us this style point. (see the CDF guidelines for godparenting)

>6) You should use serial commas consistently in lists of 3 or more
>authors. We think ref 22 is the only place you use a "," in front of the
>"and" as in `xxx, yyy, and zzzz'.

       Fixed


>7) Ref 30 is in appropriate if Weiming is an author on this paper, does it
>really differ from 31. Guess so, but still.

	Removed

>8) Ref 32 is not very satisfying. There are no proceedings available from
>this workshop (yet?). Maybe a reference to the webpage for the meeting
>would be better than just the statement that a workshop was held -- which
>is almost clear from the text in the paper body.

    Fixed