#39 10-MAY-1997 19:45:40.80 MAIL
From: FNALD0::DEMARTEAU "Marcel Demarteau"
To: SMTP%"brock@msupa.pa.msu.edu",BROCK
CC:
Subj: comments
From: FNALD0::KATHYS "KATHY STREETS - D0 X2779" 7-MAY-1997 17:00:34.75
To: BROCK
CC: DEMARTEAU,KATHYS
Subj: comments/answers to PRD questions
=============================================================================
5/7/97
Chip, below we've listed the comments you wanted us to check and/or comment
on and the status.
Kathy & Marcel
Also, about the scaled vs unscaled masses, here's what Marcel & I agreed on:
- we don't want to give unscaled masses
- which unscaled do you give? - adc counts, first correction at reco level,
TB correction, with sampling fractions, final scaled values? , which
scale?
Also people would then worry about how there errors are done etc.
I think giving unscaled values would be more confusing...
- in the data section, show the W and Z distributions unscaled
- in the parameters section we determine \alpha and \delta and can therefore
now calculate the true energy
- in the fitting section, we should
- quote Z mass scaled (91.3 or whatever)
- quote W mass scaled
- can give ratio if you want, but in reality people will use
the LEP Z mass in any calculation that involves the ratio
- also there's some things to think about when giving a ratio:
- our ratio is dependent on the LEP Z mass!
- it's not just a multiplicative factor to get to W mass because
of the offset
- if the LEP Z mass changes, our Z contour from the energy
scale will change and things do not necessarily scale and the whole
thing would have to be redone
- what we're really doing is just multiplying the values of the energy
from W events in the MC by mz_lep/mz_d0. The reason we say it's the ratio
is that the errors are calculated using the ratio and therefore a lot of
the errors are reduced.
=============================================================================
BELOW:
ANSWERS TO MW PRD COMMENTS - 10MAR96 VERSION OF PAPER
- these are the questions/comments that Chip wanted Marcel/Kathy to check
=============================================================================
Mincer:
- page 9, colum 2, last paragraph in 4:
Do I understand correctly from the fact that the missing Et was corrected
that our event vertex measurement is worse than the vertex calculated
from a single electron?
Chip==>
M&K: have made no change, but what do you want to do?
Kathy's answer ===>
Yes, the event vertex has worse resolution than a single electron.
But the point here is that 2 different vertices are used: (nominal
vertex for miss_ET and intersection of axis for electrons) and
therefore a correction has to be made for this. I say leave it as is.
Marcel's answer ===>
His understanding is correct. That's our detector. No change in
text needed.
=========================================================================
Serban:
Motivation paragraph for zero-suppression correction (Marcel will write?)
Marcel ==> done. sent you a draft.
=========================================================================
Mont:
- no questions for Marcel/Kathy
=========================================================================
Uli:
section II.D.1
11) p.7 the definition of f_iso is wrong. The quoted definition is the
one used offline. L2 uses a different one.
Chip: what say you, Marcel and Kathy?
Marcel's answer ===>
He is correct. In L2 an isolation in a 5x5 window is used. I suggest
not referring to the f_iso definition, but just say: "electrons were
required to be isolated at the trigger level". Just leave it at that.
Any more detail won't be necessary and only confusing.
21) p.10 "during the main ring cycle" is too vague. Define a clear main ring
veto condition in section II.D.2 and use it here.
Chip: not done...will think.
Marcel's answer ===>
Agree, could be made more clear.
22) p.10 The list of data samples is incomplete and/or inconsistent. There are
several extended data sets used, eg the CC/EC Z sample, which are not
described. Either restrict yourself to the main W and Z samples here
(ie the ones used for the mZ and mW measurements) or describe all data
sets completely and with comparable detail (eg give sample sizes for all
or none of the samples). Phrases like "slightly enlarged" are vague and
should be left out or made more specific. What is the relationship of
this list with table IV? Why does table IV refer to so many more Z
samples than are described in the list?
Chip: should elaborate more, I agree.
Marcel/Kathy ==>
It would be too complicated to include all data samples (W,Z, min bias,
zsup, single electron, pion, j/psi etc. etc.) so we think we should
just describe the main W and Z samples here and describe other
data samples as they're used
24) p.10/11 the EC EM scale is not discussed (it is used in the recoil
calibration and CC/EC Z masses are quoted later).
Chip: elaborate?
Marcel's answer ===>
we discussed this at the meeting. When discussing the determination of
the hadronic scale and when we say we include EC events, state that the
uncertainty on the hadronic to EM ratio from the uncertainty on the
energy scale of the EC's is negligible.
28) p.12 "This is consistent with the result obtained from the Z data
only..."
What result is referred to? Quote the numbers for the Z data only to
show their consistency.
Chip:
not always obligated to detail such an instance of agreeement...?
What about this time?
Marcel/Kathy:
agree with Chip (leave it)
31) p.12 "The offset results in a 5 MeV correction". This doesn't jive
with the the error quoted on mW due to delta (25 MeV later in the
paper). If a uncertainty of 210 MeV on delta causes an uncertainty
in mW of 25 MeV then a correction of delta=-160 MeV must cause
more than a 5 MeV change in mW.
Chip:
done, but nothing done yet about the 5MeV statement. The last sentence was
removed as being identical to wordsused where the actual uncertainties are
enumerated.
Kathy's answer ===>
We put the offset into the MC and the mass changed by 5 MeV only!
The shift that Uli originally gave in his paper was an estimate and
assumed that f_W and f_Z were equal (I think).
The effect might not be linear... and we were conservative
with errors.
Marcel's answer ===>
Uli is correct in pointing out that the shift is very small, too small
given the total uncertainty quoted. But, the total uncertainty quoted
is for ensemble tests. If our data sample happens to give in this case
just a 5 MeV shift, well, that's what the data tells us.
34) p.12 "An additional cut is imposed..." What is the cut?
Chip: we talked about this, and I don't remember what we decided.
Kathy/Marcel: This additional cut is explained on p. 10 - we should
just make a reference to it somehow
35) p.12 What does "completely consistent" mean? Can something be
incompletely consistent? The best way to convince a reader of
the consistency of two results is to list them both and let
him/her decide. Quote the CC/CC results.
Chip: done, wording....should we quote?
Marcel/Kathy: yes, we should quote and get rid of completely
40) p14 equation 23 seems to describe an energy resolution but the text talks
about a transverse momentum resolution. Fix the equation.
Chip: is this wrong?
Marcel/Kathy: all should be E_T. For the recoil that's the only
thing we have. We don't know p_L, the longitidunal momentum, of the
W.
43) p.14 something was consistent again but you don't show the numbers.
This section needs much more detail: plots of trigger turn-ons, how
measured.
Chip: We had trouble deciding about this..why was that? Why shouldn't we
have the turnon curves?
Marcel's answer ===>
we decided to throw it out because it adds to much detail. People know
how to measure a trigger turn-on curve and as such is standard stuff.
Wat is different in a W-mass analysis is the u// eff. That is something
most analyses don't need to worry about and that's why it gets more
attention here. I think it is justified to have a perceived disparity
between trigger eff and u// eff. After all, trigger eff's are in all
cross section papers. That is was also a controversial issue within
the group,.... that should not affect the paper.
We do need to quote some numbers (see page 30 of d0note 2929)
47) p.16 Quote a mass window for the 7.4% background under the Z signal.
Chip: need numbers.
Marcel's answer ===>
Window is 70-110 GeV.
52) p.19 " this contribution has also been determined from minimum bias
events" It may be interesting to know what the energy flow in minbias
events is, but I don't see why this has much relevance for W decays and
why the error in the energy flow should be the difference between that
in W and mb events. Can you explain?
Kathy ==>
We put this in because it gives people a reference point to
know how different the energy flow is from something they know.
We have to have something to base the errors on and the minimum
bias energy flow should be the minimum there is and that from the
W events the maximum, so we take the difference as the error, to be
conservative.
55) p.21 the discussion of the u|| efficiency is somewhat of a repeat of
the discussion in section IV.D.2. Put them together.
Chip: Marcel?
Marcel's answer ===>
I thought Chip didn't want to combine the discussions. Will be happy
to combine them if he wants.
56) p.22/23 The results of the Z mass fits appear to be quoted for
alpha_em = 1 and the results of the W fits are quoted for alpha_em =
something else. No explanation is given. This is very confusing. At least
it needs to be explained and the value of "something else" needs to be
given. Else the reader will wonder why our final result is not
80.35/87.01 * MZ(LEP). I really think that we need to quote these
numbers for the same calibrations.
Chip: We discussed this at the meeting, but I have to implement the decision
(which I admit I dont'remember!). Certainly, a part of it was to eliminate
the straddler Z results
Marcel's answer ===>
we decided to quote the Z-mass scaled with the alpha from the energy
scale section. It will then not come out to be exactly the LEP Z-mass.
Scaling is justified, since we studied that long time ago and results
in only a shift of 1 MeV. The figure of the Z-mass needs to be updated.
Will do that, also by simply scaling.
=========================================================================
Kotcher:
p9
col2
====
par2 - "...the MET is corrected for the difference in the two methods".
what does this mean? perhaps "the MET is recomputed
using the z position of the new vertex."
Chip: okay someone check what I wrote.
Kathy: I looked at sec_prd_27apr.tex. I don't like the way it's written
now. I think it should say:
The met is recomputed using the z position of the vertex obtained
from the electron.
Marcel: agree with Kathy.
p10 col1
====
top - "...within 10% of the module boundary for cc only". should qualify
this a little (one sentence or phrase). there's nothing in the
cal portion that indicates why this is appropriate:
modular, phi geometry, cracks in CC but not EC, etc.
Chip: huh?
Kathy: Just say that "within 10% of module boundary for cc only, due
to cracks between modules which can affect electron identification
and energy measurements"
Marcel: we don't have cracks between modules. We have uninstrumented regions
due to the mechanical construction of the modules.
p12 discussion at end of col1 is very confusing. i think you do want to say
something that we determined this stuff at the test beam
(already did, so that's ok), and that this is waht gave us 4%
discrepancy. But want also
to say clearly that delta, C are completely redetermined. quoting
166 mev is good, but must make it clearer, less cumbersome, and more
logical.
Chip: made some changes...
Kathy: I put a couple of prepositions into your sentences
Marcel: we also discussed this at the eb meeting and we agreed that the
text should say explicitely that only the beta's, the sampling
weights, are carried over from the TB but that all other numbers
are redetermined from a calibration in situ.
p14
bottom of page - the jet resolution you use (80%) is determined from
jets in situ, i believe. you imply that we obtained this for pions,
which isn't true. (it was 50% FOR PIONS).
Chip: check what I wrote.
Kathy: I checked what you wrote - you still need to change it.
Jon is right, it from the jet energy resolution at D0, not
pions from the TB!!!!
See how it's written on page 23 of d0note 2929 - that's what
you need to write!
scaled with hadronic energy scale
smeared using d0 jet resolution
Marcel: I agree with Kathy.
p19
overall comment on underlying event energy in electron/recoil: this is
a very difficult subject to write about clearly. i think the logic
you've used here is good, and your first stab at it this subject here is
also very good. i do, however, think that some things have to be
arestructure so that you don't lose the reader. i don't want to rewrite
it, but i'll give you a few comments:
Chip:
This was far from the 'first stab'...indeed we almost lost it completely over
this section! It was very hard to write and I think now is the best that this
complicated subject has been explained. It could be better and we talked
about this at the meeting. Marcel was going to write a motivational (no, not
like Schwartzkopf) section. All of this should be looke at by Marcel to
correlate with Jon's concerns.
Marcel: wrote a section, perhaps not a la Schwartzkopf, but a la
Schwarzenegger, whatever that sounds like.
Jon:
first, i think you should start this part by listing verbally the effects
of interest in an introductory-type fashion -- tails of shower, low energy
mb crap, zero-suppression, etc. -- along with why this subject
matters (correlated with the electron direction, etc.) this will warm
him/her up to the subject, give meaning to the eqns that follow, and
allow you to refer back to these concepts as you go.
for example, later, when you discuss rotating the electron cluster to get
the underlying event, you can then immediately say what effects are
included in this type of a determination: the energy is zero-suppressed,
so that it has problems that have to be addressed, etc. unravel
the problem by laying it all out in words first, then taking the problem
piece by piece (as you've done), mentioning it's short (and long) comings.
par1 under 1., p19 - "would get rather involved" -- too informal.
why is it numerically equivalent to 5X5?
near bottom -- would say "...25 0.1X0.1 delta_eta X delta_phi readout
towers" here, and just say "readout towers" or "0.1X0.1 readout
towers" later in this section. (for example par3, col2, p19)
Marcel: done, sent draft.