Dear Jens Jorgen, Hiro and Steve,
well done on the multiplicity paper. On Monday
Wit Buza was grilling me about our 200GeV results. I think that Phobos cannot
be far behind.
Physics
=======
One thing that strikes me about the paper is that if it was about antiprotons
we would only be mentioning dN/dy and never mentioning dN/dPt.
If we plot the distriubtion of events vs multiplicity it should look like
the plot below.
| * * Data
Log | * : AMPT
1/N dN/dM | * * * :
| *:
| :*
| : *
--------------------
Multiplicity
We cut on this curve to define our centrality bins. However it would be
interesting to compare this curve directly to AMPT and also to characterise
its slope at the high multiplicity end. This is nicely characterised by
RMS/sqrt(N). Note in my example the DATA and AMPT have the same mean but
different slopes at the high end of the distribuiton.)
I would suggest that the 3 right panels of Fig 3 be replaced by such plots
for the same centralities as the left panels. Thus we would be able to
compare to models in two dimensions instead of just 1 as at present.
Also it seems to me the Khazeev and Levin are ONLY predicting the ratio
of the distributions are sqrt(s)=200 and 130. Thus we should just compare
our ratios to their's.
If we are going to extract alphas and Betas why not put them (and corresponding
130GeV numbers) into table 1. This could be done with some fancy latex footwork.
It would both save space and add information content. Also why not add the
width of dN+-/dEta for each centrality to Table 1.
My second concern is for the errors shown. It is obvious from Figs 2 and
3 that the
systematic errors are correlated, since one can draw a simple curve that goes
through every error bar with a chi**2/NDF of about 0.1. Surely some of the
errors only effect the Y axis, namely they increase the multiplicity by
5% or so. These should be listed in the text and indicated on the figures by
a single line. Secondly we should demonstrate to the reader
that our Eta resolution is smaller than our Eta bins.
Those errors that effect the shape could be shown as a band
(colored if necessary). It is also OK to have two sets of systematic errors,
one for the MA and another set for the BBC.
I think that you can support your centrality procedure by referring to the
following preprint.
1. nucl-th/0110020 [abs, src, pdf, other] :
Title: Geometric relation between centrality and the impact parameter in
relativistic heavy-ion
collisions
Authors: W. Broniowski, W. Florkowski
We show, under general assumptions which are well satisfied in relativistic
heavy-ion collisions,
that the geometric relation of centrality c to the impact parameter b,
namely c ~ pi b^2/sigma_inel,
holds to a very high accuracy for all but most peripheral collisions. More
precisely, if c(N) is the
centrality of events with the multiplicity higer than N, then b is the value
of the impact parameter for
which the average multiplicity <n(b)> is equal to N. The corrections to this
geometric formula are of
the order (Delta n(b)/<n(b)>)^2, where Delta n(b) is the width of the
multiplicity distribution at a
given value of b, hence are very small. In other words, the centrality
effectively measures the
impact parameter.
The AMPT predictions you show do not involve changing the minijet cutoff to
2GeV but rather the fixing of several physics bugs in the code. These reduce the
multiplicity by 4 and 7% at 130 and 200GeV respectively.
Since the numbers are useful to people I suggest we spend the $50 to store
our data an EPAPS web site, see http://prc.aps.org/authors.html and click
on auxilarly material.
Plotsmanship
============
The labels of the plots should be larger, consistent and centered.
Each plot should have a key to the symbols and lines. This should somewhere
in the plot but should not be repeated for every panel. Hopefully there could
be some indication of the common systematic errors on the X and Y axis of
each plot. For PRL each panel needs to be labled (a) (b) (c) etc and we should
refer in the text to "Fig. 4(a)" not "the left panel of Fig. 4" We should
decide on a consistent set of notation for text and figures. I myslef like
dN^+-/dEta and dN^+-/dEta . 2/N_part.
For Fig. 2 PRL requires that the axis labels have the same number of digits.
Therefore the labels should be 4.0 4.5 etc. However if you set the maximium to
4.6 or or 4.4 then root may automatically set the labels to be 0, 1, 2 etc.
The x label should be eta-y_beam and it should be centered.
For Fig. 3 the vertical space between the panels should be removed. For the
vertical axis they should either all be set to be the same (to highlight the
change in yield) or each panel should be 1.15*(the maximium of dN/dEta).
This choice of scales would highlight any difference in shape. The centrality
ranges should written on each plot.
I think that the current Fig 4 dN/dEta * 2/N_part
would look nicer as a lego plot. If you like it as is why not but the centrality
ranges on the top x axis?
Text
====
Like Zbig I worry about length. There is a new Revtex on the
APS web site that is supposed to give the "look and feel" of PRL. We should do
this as a check on size and the clarity of figures.
Try to use the small anglo saxon word rather than
the long french or latin one, eg on page 4 paragraph 2 2nd sentance
"They sit next to the beam pipe +-2.5m from the nominal vertex. Are we using
both the Si and the TMA. If we are not using something it should not be
described in detail. Minimize the ZDC description.
"The ZDC measure neutrons emmited along the beam direction [cit NIM and the
infamous ZDC physics paper M. Chiu et al, nucl-ex/0109018]. They server
as the min bias .... 97% of cross section) and locate the vertex with a
resolution of 3.6cm"
References should be grouped together ie \cite{ref1, ref2}. In captions you
may need to use \protect{\cite{ref3}}. The references need to be correctly
ordered.
Abstract ``limiting"
Page 2 2nd paragraph change
predicted~\cite{partonsat83,Escola00,
Kharzeev_and_Levin}, which would limit the production of
charged particles.
to
predicted~\cite{partonsat83,Escola00,
Kharzeev_and_Levin}. This would limit the production of
charged particles.
Next paragraph use sqrt(S_nn)=200GeV rather than 100+100 AGeV
Page 7
Para 1 2nd sentance
change "For the 4 show rapidities" to
"For the 4 rapidities shown"
In concluding paragraph change "migh" to "might".
Well done!,
Michael
Michael Murray, Cyclotron TAMU, 979 845 1411 x 273, Fax 1899
Dear Jens Jorgen, Hiro and Steve,
well done on the multiplicity paper. On Monday
Wit Buza was grilling me about our 200GeV results. I think that Phobos cannot
be far behind.
One thing that strikes me about the paper is that if it was about antiprotons
we would only be mentioning dN/dy and never mentioning dN/dPt.
If we plot the distriubtion of events vs multiplicity it should look like
the plot below.
| * * Data
Log | * : AMPT
1/N dN/dM | * * * :
| *:
| :*
| : *
--------------------
Multiplicity
We cut on this curve to define our centrality bins. However it would be
interesting to compare this curve directly to AMPT and also to characterise
its slope at the high multiplicity end. This is nicely characterised by
RMS/sqrt(N). Note in my example the DATA and AMPT have the same mean but
different slopes at the high end of the distribuiton.)
I would suggest that the 3 right panels of Fig 3 be replaced by such plots
for the same centralities as the left panels. Thus we would be able to
compare to models in two dimensions instead of just 1 as at present.
My second concern is for the errors shown. It is obvious from Figs 2 and
3 that the
systematic errors are correlated, since one can draw a simple curve that goes
through every error bar with a chi**2/NDF of about 0.1. Surely some of the
errors only effect the Y axis, namely they increase the multiplicity by
5% or so. These should be listed in the text and indicated on the figures by
a single line
Those errors that effect the shape could be shown as a band
(colored if necessary). It is also OK to have two sets of systematic errors,
one for the MA and another set for the BBC.
I think that you can support your centrality procedure by referring to the
following preprint.
1. nucl-th/0110020 [abs, src, pdf, other] :
Title: Geometric relation between centrality and the impact parameter in relativistic heavy-ion
collisions
Authors: W. Broniowski, W. Florkowski
We show, under general assumptions which are well satisfied in relativistic heavy-ion collisions,
that the geometric relation of centrality c to the impact parameter b, namely c ~ pi b^2/sigma_inel,
holds to a very high accuracy for all but most peripheral collisions. More precisely, if c(N) is the
centrality of events with the multiplicity higer than N, then b is the value of the impact parameter for
which the average multiplicity <n(b)> is equal to N. The corrections to this geometric formula are of
the order (Delta n(b)/<n(b)>)^2, where Delta n(b) is the width of the multiplicity distribution at a
given value of b, hence are very small. In other words, the centrality effectively measures the
impact parameter.
At some point we will have to care about length. There is a new Revtex on the
APS web site that is supposed to give the "look and feel" or PRL. We should do
this as a check on size.
This archive was generated by hypermail 2b30 : Thu Oct 25 2001 - 14:57:51 EDT