<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
</head>
<body>
<div class="BodyFragment"><font size="2"><span style="font-size:11pt;">
<div class="PlainText">Dear All,<br>
<br>
Please find attached minutes of our meeting yesterday which I’ve posted to the HadStruct GitHub page. Also there is Wayne’s notes, but I attach them also. Remember our next call is WEDNESDAY AT 1215PM.</div>
</span></font></div>
<div class="BodyFragment"><font size="2"><span style="font-size:11pt;">
<div class="PlainText"><br>
<br>
Regards,<br>
David<br>
<br>
**************************************<br>
<br>
Present:<br>
<br>
BJ,CM,CK,CE,ER,JQ,JK,RS,SZ,TK,WM,DR,KO<br>
<br>
Computing time<br>
<br>
SZ noted he had an award of 300K node hour at Jean Zay in Paris (1 node = 4 V100 GPUs), and was writing an application under PRACE<br>
<br>
DR noted we had a six-month extension on Bridges, and some remaining time on Stampede2<br>
<br>
Gluon Distribution<br>
<br>
WM, AR and RS discussed the polarized gluon distribution matrix elements that were needed, and distributed some notes here.<br>
<br>
One thing that would be needed for the gluon distribution would be the matrix element corresponding to the unpolarized momentum fraction, and how that would be renormalized, e.g. RI Mom, etc. The Wilson-flow analysis would probably entail performing the same
procedure for both the gluon and flavor-singlet quark distributions. We would arrange a meeting later in the week (1pm Friday) to discuss this.<br>
<br>
PDFs<br>
<br>
CK and CE gave an update on their progress with the PDFs using distillation, and ER an update on the data on the large-volume 48^3 near-physical lattices. With the 16 operators under consideration, but with a small number of sources, the ground-state appeared
well described, but excited states not so. TK showed his analysis at the 32^3 lattices where correlators were computed from every time slice, and identified 9 or 10 operators that described the system. For the largest lattice, the practical limit on the number
of eigenvectors was probably 128, and therefore finding a good basis was important.<br>
<br>
KO asked three questions:<br>
<br>
• How the Boosting paper was progressing. CE and DR said there was a draft with the two-point functions essentially complete, but we were discussing whether to include the matrix elements.<br>
• Running after GPDs on Frontera - there was still some time necessary runs on the GPDs.<br>
• State of fitting to the PDFs. Here it was noted that the sGEVP method was probably needed to properly benefit from the multiple operators with distillation.<br>
There was discussion about the fitting form for GPDs. AR advocated using the double-distribution which would give a parametrization we could use, in the manner of the "global-fitting-like parametrization" we were using for the PDFs.<br>
<br>
Next Meeting:<br>
<br>
Wednesday 15th April at 12:15pm.</div>
</span></font></div>
</body>
</html>