Zoom Logo

IRIS-HEP Retreat Plenary Tuesday - Shared screen with speaker view
Robert Tuck
23:45
https://docs.google.com/document/d/1E4zBMmW86-DmunJcVKapHNp1CqFxk__9gaY7YZAEZKk/edit
Robert Tuck
23:53
Session minutes
Frank Wuerthwein
46:59
Same Q as Kyle: Can Pete/Gordon/Brian simply assign the number of milestones each area should aim for?
Frank Wuerthwein
50:29
Ok, so OSG-LHC should shoot for ~4 NSF visible milestones for the next 2 years. One every 6 months. And then maybe internal ones as we see fit.
Frank Wuerthwein
51:23
Do we need to have milestones for Year 5 in the PEP? Or only Y3 and Y4?
Gordon T. Watts
51:30
That sounds ok.
Gordon T. Watts
51:36
OSG is a little different, of course.
Kyle S Cranmer
51:37
Are others having problems hearing Pete or is that on my end
Gordon T. Watts
51:46
Yeah - he is competing with fortnight.
Kyle S Cranmer
51:59
Ok, I wasn’t sure if it is on my side or not
Ken Bloom
52:27
We’re all competing with Fortnite.
Frank Wuerthwein
53:05
Ok, good enough for now. Thanks!
Gordon T. Watts
54:33
HE - High Energy LHC! Awesome! ;-)
Vassil Vassilev
01:07:12
Question to Kyle: Could you elaborate on how differentiable programming would fit in the field.
Gordon T. Watts
01:07:15
Snowmass is another obvious place to engage with the theory community.
Kyle S Cranmer
01:11:56
I’d love to talk, sorry if I didn’t understand the question
Kyle S Cranmer
01:13:19
Pete I think I misunderstood your “overloading” point.
Kyle S Cranmer
01:14:29
Mike: https://iris-hep.org/about/team
Mark Neubauer
01:17:31
Yes, Snowmass efforts us (IRIS-HEP) several opportunities, one of which is theory engagement with our developments and directions, especially in AS.
Frank Wuerthwein
01:27:20
“One site transitioning to caching-only” is a T2? Or any site?
Kyle S Cranmer
01:28:10
Brian, can you comment briefly on SkyHook and Awkward Array.
Jim Pivarski
01:30:45
If they're Arrow in the storage layer, it's not a big step to make them Awkward in the analysis.
Alexander Held
01:33:13
SkyHook / ServiceX: If I use ServiceX to make a request and get some data, then it delivers the data to me in some fashion - does SkyHook replace this delivery mechanism?
Benjamin Galewsky
01:41:53
@Alexander — we are imagining ServiceX delivering columnar data to Sky Hook.
Carlos Maltzahn
01:44:16
@Alexander, @Ben — in my mind ServiceX Frontend is the user interface for both requests and delivery. Requests get stored in Skyhook and delivery accesses Skyhook through the frontend.
Alexander Held
01:45:38
Ok, so a user of ServiceX wouldn’t themselves interact with SkyHook? And presumably right now there’s something within ServiceX that performs what SkyHook would do in a simplified fashion?
Gordon T. Watts
01:46:08
There is an ad-hoc discussion about the details of this, this afternoon (US Time)
Mark Neubauer
01:48:50
“This” you refer is the Snowmass data hosting and job submission? Are there more details about the meeting you mention?
Brian Bockelman
01:58:24
I find the iPad is one of the most reliable videoconferencing devices...
Gordon T. Watts
02:09:54
@Rob - 5 minutes left before 5 minutes of question time.
Frank Wuerthwein
02:12:42
@Pete: if we are going the route of serviceX acceleration than we need to find a deal with the ops program to share buying a testbed.
Frank Wuerthwein
02:13:03
2 hosts, both 100Gbps capable,
Kyle S Cranmer
02:13:54
I would also imagine SSL playing a role in the “Analysis (200TB) Grand Challenge”
Rob G
02:23:53
hi Nhan! thanks will take a look
Gordon T. Watts
02:30:05
@David - you have 5 minutes before questions start.