[Hadstruct] Fwd: [EXTERNAL] Questions on your USQCD proposal

[Konstantinos Orginos]

Dear All,

Apart from discussing the answers to the SPC we should also discuss:

1) Progress with disconnected diagrams (Chris C.)
2) DA calculations (Daniel)
3) GPD paper status

Cheers,

Kostas


Sent from my iPad

On May 9, 2023, at 10:27 AM, David Richards via Hadstruct <hadstruct at jlab.org> wrote:


Dear All,

Please find attached some questions from the SPC.  I’m not sure I can respond “by return”, but certainly should be able to respond by tomorrow night.   FYI, I enclose the proposal.


Begin forwarded message:

From: "Bhattacharya, Tanmoy" <tanmoy at lanl.gov<mailto:tanmoy at lanl.gov>>
Subject: [EXTERNAL] Questions on your USQCD proposal
Date: May 8, 2023 at 9:58:39 PM EDT
To: David Richards <dgr at jlab.org<mailto:dgr at jlab.org>>


Dear PI,

The SPC are discussing the proposals received for the USQCD computational resources, and will finalize allocations over the next couple of weeks.  As I mentioned at the AHM, the demand is almost a factor of two over available resources, except for the GPU resources (The 21g is oversubscribed by a factor of 2 as well).  If you haven't mentioned it in your proposal, but can use the non-21g GPU resources efficiently, now would be a good time to mention it.

In addition, while reading your proposal, we had some questions.  We would appreciate if you could provide some clarification by return email, as soon as possible, and definitely by this weekend.  This will help us better evaluate the expected benefit to USQCD of the allocation we recommend for your proposal.

Q1. Which deliverables would you prioritize in case the allocation is reduced?

KNL = 50% contractions for flavor-singlet hadron structure and 50% peramabulators and genprops on 48^3 lattice
21g = perambulators and generalized perambulators

We can certainly move 19g -> 21g - note latter undersubscribed.  I would prioritize two-point functions on KNLs.


Q2. Can one comment on the cost/gain ratio for the use of the distillation
method?


OK, these old chestnuts are the best.
Q3. What is the range of t and ξ values expected to get good signal within the
obtained statistics?

Maybe we should think about this.

Regards,
David

Thank you
Tanmoy
USQCD SPC Chair

--
Dr. Tanmoy Bhattacharya, Scientist 5 and Laboratory Fellow,
Org:T-2 MS:B285 DP:03012305U Rm:03-0123-178 W:+1(505)665 4733 C:+1(505)660 5404

<USQCDstruct2023-3.pdf>
_______________________________________________
Hadstruct mailing list
Hadstruct at jlab.org
https://mailman.jlab.org/mailman/listinfo/hadstruct
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.jlab.org/pipermail/hadstruct/attachments/20230510/bca3047e/attachment-0001.html>