<html>
<head>
<meta content="text/html; charset=ISO-8859-1"
http-equiv="Content-Type">
</head>
<body text="#000000" bgcolor="#FFFFFF">
<div class="moz-cite-prefix">What are the arguments for proposed
binning in x?<br>
the last bean is clear, as much data in this configuration as
possible, but splitting of SHMS data taking<br>
in 3 intervals is not discussed from the optimalization point of
view, it would be good to have it in the<br>
presentation.<br>
<br>
RESPONSE : (Hmm. Not sure how to answer this. Simple answer is
that the points represent the largest<br>
spread in x that allows a reasonable overlap
with HERMES in a reasonable amount of beam-time.<br>
Ellie has optimized to avoid large
systematics from F1, and suppression of rates. )<br>
<br>
--------------------------------------------<br>
<br>
Ellie: The x=0.49 point was chosen to compare with the largest
non-zero Azz & b1 found in the HERMES data. It will be taken
at a slightly higher x than HERMES due to combining statistics
from both the HMS an<br>
d the SHMS. The HMS was not able to be pushed any lower in x
without sacrificing statistics, and so x=0.49 was chosen to
balance these two factors.<br>
<br>
The x=0.16 point was chosen to optimize the uncertainty in b1 by
balancing available statistics against F1 which increases
dramatically in the low-x region. <br>
<br>
The x=0.28 and 0.36 points were chosen to map out any potential
zero-crossing behavior with similar statistics in Azz for both the
high-x and low-x points. These bins are smaller than the two end
bins due to overlapping contributions from nearly all of the
spectrometer settings.<br>
<br>
Take care,<br>
Ellie<br>
<br>
<pre class="moz-signature" cols="72">Elena Long, Ph.D.
Post Doctoral Research Associate
University of New Hampshire
<a class="moz-txt-link-abbreviated" href="mailto:elena.long@unh.edu">elena.long@unh.edu</a>
<a class="moz-txt-link-abbreviated" href="mailto:ellie@jlab.org">ellie@jlab.org</a>
<a class="moz-txt-link-freetext" href="http://nuclear.unh.edu/~elong">http://nuclear.unh.edu/~elong</a>
(603) 862-1962</pre>
On 06/09/2013 11:07 PM, Karl Slifer wrote:<br>
</div>
<blockquote
cite="mid:CAKYfHotxUu9=yWTMrGM-7sw2EgX1TGorBNOv9_vgXcY6AnZNBA@mail.gmail.com"
type="cite">
<div dir="ltr">
<div><br>
</div>
<div>Hi all,</div>
<div><br>
</div>
<div>Below please find a draft response to the readers. I'd
appreciate any/all feedback.</div>
<div><br>
</div>
<div>Thanks to Oscar and Dustin for sending comments already. I
hope I have incorporated them satisfactorily, but let me know
if not.</div>
<div><br>
</div>
<div>Dustin : any ETA for the updates to the technote? Ideally
we can send them this short email and then provide your note
for the full details. </div>
<div>Most important would be to clarify the overall drift
numbers, as I had the same confusion that Ellie raised.</div>
<div><br>
</div>
<div>I'll have time to work on this tomorrow morning and early
afternoon, but have to leave for DC mid afternoon. I'd love
to get a response to Ewa before then if possible.</div>
<div>
<br>
</div>
<div>thanks much,</div>
<div><br>
</div>
<div>
-Karl</div>
<div><br>
</div>
<div><br>
</div>
-----------------------------<br>
<div class="gmail_quote"><br>
The measurement is very sensitive to the systematic effects
and good control of them is the<br>
key point. Therefor I would like to know if there are any
estimates of expected size of effects from:</div>
<div class="gmail_quote"><br>
</div>
<div class="gmail_quote">RESPONSE: We agree with the TAC
assessment that systematic errors from drifts must be
mitigated, but</div>
<div class="gmail_quote"> that they are
manageable with a combination of hardware upgrades and a
dedicated collaboration </div>
<div class="gmail_quote"> effort. We note
that the recent g2p experiment involved a similar situation
where a significant</div>
<div class="gmail_quote"> commitment was
made to install and run this polarized target, but separate
groups were tasked</div>
<div class="gmail_quote"> with substantial
tasks of beamline, DAQ and detector upgrades. We are grateful
to the TAC for</div>
<div class="gmail_quote"> pointing out
several effects which were not explicitly discussed in our
submitted proposal. We </div>
<div class="gmail_quote"> have written a
short note discussing each of these and conclude that the
overall systematic</div>
<div class="gmail_quote"> uncertainty is
still of the same order as estimated in the proposal.</div>
<div class="gmail_quote"><br>
1. beam - one aspect is the stability in terms of position and
divergency<br>
this can change acceptance and produce false asymmetries<br>
<br>
</div>
<div class="gmail_quote">RESPONSE: (The TAC report pointed out
that the false asymmetries from beam position drifts are
easily</div>
<div class="gmail_quote">
removed by "regression". Can someone fill
in the details? I assume that this just means</div>
<div class="gmail_quote"> that the parity
feedback on position is very good, but we need some
numbers/examples to</div>
<div class="gmail_quote"> back this up. This
is dealt with in section 1.1.4 of the note, but there are few
details.)</div>
<div class="gmail_quote"><br>
</div>
<div class="gmail_quote">
2 beam polarization - how the unpolarized beam will be
obtained what kind of effects are<br>
expected from beam polarization (ie. to which level exact
averaging of opposite beam polarity<br>
is needed and how the phase space of the beam is
polarization dependent)</div>
<div class="gmail_quote"><br>
</div>
<div class="gmail_quote">RESPONSE: JLab E06-010 (Transversity)
spin-averaged a highly polarized (~80%) beam in order to</div>
<div class="gmail_quote"> obtain an
"unpolarized" beam. The parity feedback allowed for
knowledge of the residual</div>
<div class="gmail_quote"> beam polarization
at the 2.2*10^{-5} level, according to the lumi monitors.
Please see attached</div>
<div class="gmail_quote"> plot, which shows
the beam asymmetry from that experiment.</div>
<div><br>
</div>
<div class="gmail_quote"><br>
3. temperature effects on the efficiency (and stability of the
detectors allignment) - proposed scheme<br>
of polarization reversals will give data with target
polarization during the day and unpolarized<br>
at night or vice versa. This can introduce false
asymmetry related to any kind of temperature<br>
dependence in efficiency or allignment. Was it estimated
? Are there any studies of this kind of effects<br>
in previous experiments?</div>
<div class="gmail_quote"><br>
</div>
<div class="gmail_quote">RESPONSE: (This could be addressed by
the transversity slides, but I'm not sure if the pion yield
plot </div>
<div class="gmail_quote">
addresses this. The main sensitivity to
temperature will be the BCMs and Dave has plans</div>
<div class="gmail_quote"> to isolate them,
but I don't have any details of that yet.)</div>
<div class="gmail_quote"><br>
</div>
<div class="gmail_quote"><br>
</div>
<div class="gmail_quote"><br>
4. for the drift of efficiency and its time dependence (page
25) linear evolution in time is assumed.<br>
for which effects it is justified? It is clear that for
example changes in packing factor of dilution factor<br>
can have "step like" characteristics. Are there any ideas
to what level such effects can be controlled<br>
during the run?<br>
<br>
RESPONSE : (Linear and sinusoidal drifts are the only type
that I've seen. Higher orders could theoretically </div>
<div class="gmail_quote"> be present, but if
they were I suppose they would become an issue for all
experiments, not just ours.)<br>
"Step-like" changes in the packing factor
or dilution factor have only been observed once in 700 hours</div>
<div class="gmail_quote"> of running the
polarized target. It was immediately obvious from the change
in polarization. If it</div>
<div class="gmail_quote"> occurs during this
experiment, it would impact only a single pol/unpol cycle,
which is either a 12 hour or</div>
<div class="gmail_quote"> 24 hour portion of
data. This data would need to be either discarded or handled
with care.</div>
<div class="gmail_quote"><br>
</div>
<div class="gmail_quote"><br>
In the proposal "consistency checks on measured cross section
for each run" is mentioned.<br>
What precisely is meant? At what level it can be done for the
proposed measurement? Please give<br>
more detail, especially on the precision of such test.<br>
<br>
RESPONSE : Typically we can monitor the unpolarized yields to
the better than 1% level. Luminosity monitors</div>
<div class="gmail_quote"> installed around
the beamline can be monitored to the ?? level (J.P.?)<br>
<br>
<br>
The other test mentioned in the proposal, where I would like
to have some more comments on is "the<br>
measurements of dilution and packing factor - with carbon
target "- what exactly is planed and which<br>
precision can be obtained? is it included in the beam time
estimate?</div>
<div class="gmail_quote"><br>
</div>
<div class="gmail_quote">RESPONSE: The polarized target material
is deuterated ammonia (ND_3). </div>
<div class="gmail_quote">
We determine the dilution factor by the
ratio of simulated radiated rates on D to total</div>
<div class="gmail_quote"> rates. The pf is
calculated by interpolating the ND3 measured rates between
simulated </div>
<div class="gmail_quote"> rates for different
packing factors, with the simulation calibrated by the
measured rates</div>
<div class="gmail_quote"> on a carbon target
of known thickness. The systematic uncertainty of this
process</div>
<div class="gmail_quote"> is at the 4% level.
It is important to note that the dilution factor is a scale
factor so the </div>
<div class="gmail_quote"> uncertainty is an
overall scale factor. We have 6 hours assigned to this task
in our overhead table 4. </div>
<div class="gmail_quote"> This is a
relatively short time since all that is needed is to measure
the unpolarized cross section </div>
<div class="gmail_quote"> from a carbon
disk, and the rate is usually quite high. </div>
<div class="gmail_quote"><br>
</div>
<div class="gmail_quote">
<br>
It would be interesting to see comparison of expected
statistical errors in each bin with expected false<br>
asymmetries from time variation of beam and
efficiency/acceptance.<br>
<br>
RESPONSE: We now have plots graphically showing the full
systematic uncertainty, both from the normalization</div>
<div class="gmail_quote"> dependent factors
and the possible drifts. The plots are shown in the technote.<br>
<br>
<br>
What are the arguments for proposed binning in x?<br>
the last bean is clear, as much data in this configuration as
possible, but splitting of SHMS data taking<br>
in 3 intervals is not discussed from the optimalization point
of view, it would be good to have it in the<br>
presentation.</div>
<div class="gmail_quote"><br>
</div>
<div class="gmail_quote">RESPONSE : (Hmm. Not sure how to
answer this. Simple answer is that the points represent the
largest</div>
<div class="gmail_quote">
spread in x that allows a reasonable
overlap with HERMES in a reasonable amount of beam-time.</div>
<div class="gmail_quote"> Ellie has
optimized to avoid large systematics from F1, and suppression
of rates. )</div>
<div class="gmail_quote"><br>
</div>
<div class="gmail_quote"><br>
<br>
In general, also the authors call the measurement "ratio
method" it is the cross section difference method<br>
as the two data sets are taken at different time. Advantage of
"ratio method" can be fully used when two<br>
target cels are exposed at the same time and next order is
reversed. Such configuration allows several<br>
additional cross checks, but requires two cell target.<br>
<br>
RESPONSE : We have examined a two cell configuration, and
while it is attractive for the reason you point</div>
<div class="gmail_quote"> out, it was not
clear at the time of the proposal submission that it
significantly reduced the </div>
<div class="gmail_quote"> overall
systematic uncertainty. We will continue to examine this
option and are open to using </div>
<div class="gmail_quote"> it if we are
convinced the systematic improvement is significant. </div>
<div class="gmail_quote"> </div>
<div class="gmail_quote"><br>
<br>
test of Close-Kumano sum rule - it is very hard to do such
test with limited coverage in x, may be some<br>
estimates of contribution measured/extrapolated can be done
for specific models. For the models mentioned<br>
on page 17 - can one get some idea what would be the
contribution in measured range?<br>
</div>
<br>
<div>RESPONSE : Yes, this in an excellent suggestion. G. Miller
and M. Sargian have provided us their curves. </div>
<div> We need to integrate to see the
contributions for x<0.15 and x>0.5. (Ellie, I think you
have these </div>
<div> now. Can you look into this?)</div>
<div> </div>
<div><br>
</div>
<div><br>
</div>
</div>
<br>
<fieldset class="mimeAttachmentHeader"></fieldset>
<br>
<pre wrap="">_______________________________________________
b1_ana mailing list
<a class="moz-txt-link-abbreviated" href="mailto:b1_ana@jlab.org">b1_ana@jlab.org</a>
<a class="moz-txt-link-freetext" href="https://mailman.jlab.org/mailman/listinfo/b1_ana">https://mailman.jlab.org/mailman/listinfo/b1_ana</a>
</pre>
</blockquote>
<br>
</body>
</html>