<html>
<head>
<meta http-equiv="content-type" content="text/html; charset=UTF-8">
</head>
<body text="#000000" bgcolor="#FFFFFF">
<font size="-1" face="Times New Roman, Times, serif"><font
face="Helvetica, Arial, sans-serif">Dear all,<br>
I tried to derive photons corrections for MC and data, following
the procedure described in this note: <a
class="moz-txt-link-freetext"
href="https://misportal.jlab.org/ul/Physics/Hall-B/clas/viewFile.cfm/2011-004.pdf?documentId=632">https://misportal.jlab.org/ul/Physics/Hall-B/clas/viewFile.cfm/2011-004.pdf?documentId=632</a><br>
Basically, this is based on the exclusive reaction gamma p ->
p pi+ pi- gamma, <b>where all the particles are measured</b>.
The measured gamma 3-momentum is compared to the 3-momentum
obtained from a 1C kin. fit to the gamma p -> p pi+ pi-
(gamma) hypothesis. <br>
For Kin. fit, I am using the package provided by MK. Simulations
have been performed with MK DB and runIndex = 10.<br>
<br>
After deriving MC corrections (and waiting for data to be
skimmed), I also looked at photon resolution and covariance
matrix for MC. For MC, cov. matrix can be determined comparing
the "true" gamma 3-momentum with the measured one. The "true"
3-momentum can be the generated one, or the one obtained from
the kin. fit. For data, the same procedure can be adopted, and
the "true" gamma 3-momentum will be the missing one obtained
from the kin. fit. However, when using as "true" 3-momentum the
one obtained from the kin. fit, the resolution (i.e. the width
of the "kin.fit-measured" distribution) is given by the
(quadratic) sum of the error on the measured quantity, and the
error on the missing photon 3-momentum from kin. fit. If the
latter is "small", it can be neglected. <br>
<br>
I attach a plot (MC1.png) showing the MC distribution of:<br>
<br>
* Generated photon energy - kin. fit photon energy (black)<br>
* Measured photon energy - generated photon energy (red)<br>
* Kin. fit photon energy - measured photon energy (green)<br>
<br>
The width of the black curve is smaller than the one of the red
curve, showing the the kin. fit photon energy error is much
smaller than the measured photon energy error. This guarantees
that the resolution studies based on comparing the kin. fit
photon energy to the measured one are not biased.<br>
<br>
I also attach a plot (MC2.png) showing the MC distribution of:<br>
<br>
* Generated photon theta - kin. fit photon theta (black)<br>
* Measured photon theta - generated photon theta (red)<br>
* Kin. fit photon theta- measured photon theta (green)<br>
<br>
In this case, the width of the "generated - kin. fit"
distribution (black) is LARGER than the width of the "generated
- measured" distribution (red), as if the experimental error on
the measured photon theta angle is SMALLER than the error on the
theta angle of the missing photon from kin. fit. If this is
correct, the corresponding resolution can't be extracted using
the kin. fit information as the "true" one: not a problem for
MC, but for data. <br>
Or, maybe, the EC simulation (gsim+gpp) is not smearing enough
the EC response for angles - however, I checked that EC
resolution determined from generated 3-momentum is compatible
with what described, for example, in <a
class="moz-txt-link-freetext"
href="https://misportal.jlab.org/ul/Physics/Hall-B/clas/viewFile.cfm/2009-011.pdf?documentId=533">https://misportal.jlab.org/ul/Physics/Hall-B/clas/viewFile.cfm/2009-011.pdf?documentId=533</a><br>
<br>
Do you think it may be possible that the error on the measured
photon angle in EC is smaller than the missing-photon angle
(after applying the kin. fit?). Note, finally, that the same
conclusion applies even when not using the kin. fit, but just
taking the missing photon 3-momentum from final state p, pi+,
pi-.<br>
<br>
Thanks!<br>
Bests,<br>
Andrea</font></font>
</body>
</html>