<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=us-ascii">
</head>
<body style="word-wrap: break-word; -webkit-nbsp-mode: space; line-break: after-white-space;" class="">
<br class="">
<div>Hi Igal,</div>
<div><br class="">
</div>
<div>
<blockquote type="cite" class="">
<div class="">On Apr 13, 2020, at 9:57 PM, Igal Jaegle <<a href="mailto:ijaegle@jlab.org" class="">ijaegle@jlab.org</a>> wrote:</div>
<div class=""></div>
</blockquote>
<br class="">
<blockquote type="cite" class="">
<div class="">
<div style="font-style: normal; font-variant-caps: normal; font-weight: normal; letter-spacing: normal; text-align: start; text-indent: 0px; text-transform: none; white-space: normal; word-spacing: 0px; -webkit-text-stroke-width: 0px; text-decoration: none; font-family: Calibri, Arial, Helvetica, sans-serif; font-size: 12pt;" class="">
Also, I still do not understand why this should change the gain and prevent from improving the resolution.<span class="Apple-converted-space"> </span></div>
</div>
</blockquote>
<br class="">
</div>
<div>It doesn't in principle, however it complicates the background. If you don't model this background very well then your fit parameters (which you are using to tell you information about gain and the resolution) develop systematic biases and no longer reflect
the true gain and resolution. I think you can start to see this if you look very closely at the plot you sent (re-pasted below). The peak of the red curve appears shifted just a little bit high in this plot: the top of the red curve is at the rightmost
of a few bins that appear to have the same contents, and just above the peak the curve systematically overestimates the bin contents (before systematically underestimating the bin contents beyond that). A plot of the residuals would confirm this, but I don't
see these features in the other plot you sent.</div>
<div><br class="">
</div>
<div>The problem is that the precision of your extraction of the mean (and hence the gain) is being limited by systematic errors rather than statistical errors on the fit parameters. If the systematic errors are fully correlated across all runs then this is
no problem for extracting relative gains: you make the same systematic mistake every time and your results still reflect relative changes. However, it appears this is not the case, the background changes as the runs evolve (due to changing physical conditions
or how your algorithm deals with the presence of tracking hits) and hence these systematic biases change and lead to apparent significant variations of the gain.</div>
<div><br class="">
</div>
<div>The eta fits seem much cleaner and less susceptible to these background problems, with one exception: if part of the shape distortion in the pi0 is due to variation of the true event vertex with respect to the assumed event vertex (center of target) then
that distortion would be present in the eta also. </div>
<div><br class="">
</div>
<div>Matt</div>
<div><br class="">
</div>
<div><br class="">
</div>
<div><br class="">
</div>
<div><img apple-inline="yes" id="1C64B17E-9163-42A9-9530-66B557B290D2" src="cid:18862133-E7BA-481A-87F4-98EAA3FD2930@hsd1.in.comcast.net" class=""></div>
</body>
</html>