NPPS Leadership Team

["Huang, Jin" <jhuang AT bnl.gov>]

Hi Torre

 

Thanks for sharing the slides. Few comments just for your consideration:

 

• On slide 3:
  it might be useful to point out one challenge of NP AI application is to understand its systematic bias and unexpected failure modes, as many of our measurements are systematic uncertainty dominated. I think this is reflected on slide 8
• On slide 3: “Engaging all talents”:
  it might be useful to bring up a constructive relationship and expertise complementarity with CSI. I did enjoy and benefited from working with CSI
• Slide 3: “See agenda of the EIC UG Summer 2021 https://indico.bnl.gov/event/11463/timetable/#20210805.detailed”:
  Might be useful to also link to our last AI4EIC workshop: https://indico.bnl.gov/event/10699/
• Slide 4: “MC simulations of detector responses using GANs”:
  It would be useful to emphasize the novel AI approach to bridge simulation and data in the LDRD 21-029 as lead by Brett, which has two levels of application in sPHENIX both in detector simulation and in jet-medium simulation (though scheduled to later part of that LDRD)
• Slide 4: “we could also have an opportunity for outreach - data repository may attract external teams (TrackML “like”)”:
  In recent discussion at HSF reco working group, we considered possibility to publish a follow up to TrackML data with calorimeter or tracking+calo data. With some effort investment, this dataset could come from sPHENIX or EIC simulations to foster ML particle flow development.
• Slide 12:

• It is thrilling to be quoted in this context 😊
• Let me rephrase the last part of the bullet 1.2:  “After completing the primary goal of the adoption of FELIX-style streaming DAQ by the EIC community, LDRD 19-028 has relocated 50% of the resources to the AI compression development”

Slide 13: just to note there is a NP (sPHENIX) component tooSlide 13: Also LDRD 21-023 mentioned in the last slide is also IO+CSI+NPSlide 18: Last bullet I would also encourage to recognize sPHENIX as a useful testing ground enroute to EIC, as reflected by e.g. the last DOE award to LANL+MIT.Slide 20: I am not sure it is intended to be shown, but since then ECCE has used a specialized DNN for electron and muon IDs based on full simulation and reconstructions, as used in this week’s DPAP homework response.

 

Hope these are useful, and enjoy the retreat ~

 

Cheers

 

Jin

 

 

______________________________

 

Jin HUANG

 

Physicist, Ph.D.

Brookhaven National Laboratory

Physics Department, Bldg 510 C

Upton, NY 11973-5000

 

Office: 631-344-5898

Cell:   757-604-9946

______________________________

 

From: Torre Wenaus <wenaus AT gmail.com>
Sent: Thursday, January 20, 2022 7:37 PM
To: NPPS leadership team <Phys-npps-mgmt-l AT lists.bnl.gov>; Huang, Jin <jhuang AT bnl.gov>; Michael Begel <michael.begel AT cern.ch>; Aschenauer, Elke <elke AT bnl.gov>
Subject: NP AI/ML slides for mini-retreat tomorrow

 

Hi all,

Jamie and Eric have organized a mini-retreat on NP computing tomorrow, unfortunately with a rather short invite list.

https://indico.bnl.gov/event/14140/

 

Jerome and I were asked to give a presentation on AI/ML. Our slides are here; my slides begin on #8. I'd really appreciate any comments you have. 

https://docs.google.com/presentation/d/1waM1VkA3elB3YJUxTTJPvS6uXApaT80jN1Q08r3QZK8/edit#slide=id.g10f0e7284f3_3_111

 

The meeting begins at 1pm tomorrow. Sorry for the late request.

 

  Torre

 

--

-- Torre Wenaus, BNL NPPS Group, ATLAS Experiment

-- BNL 510A 1-222 | 631-681-7892 |  wenaus AT gmail.com | npps.bnl.gov | wenaus.com

-- NPPS Mattermost room: https://chat.sdcc.bnl.gov/npps/channels/town-square