NPPS Leadership Team

[Torre Wenaus <wenaus AT gmail.com>]

Hi,

As FYI (not more than that yet unfortunately) Meifeng is writing (with my encouragement) an LDRD A proposal for submission to CSI, the short description of which is

We propose a collaboration between CSI's HPC group and the Physics Department's NPPS group to combine their expertise towards developing, deploying and applying distributed machine learning services. CSI's HPC expertise combined with NPPS's experience building ML services using PanDA and iDDS and working with physicists in identifying and supporting physics analysis use cases will enable us to effectively leverage large scale HPC resources for ML to a much greater extent than either CSI or the NPPS team can on their own.

We're currently discussing requested funding/effort breakdown. From my point of view, we can bring any level that fits the funding and our available effort, from 0 to N FTEs. The proposal draws on a work program that we're doing anyway for ATLAS (with IRIS-HEP help on iDDS). I say 0 as a possibility because NPP today will choose a list of proposals filling its LDRD A quota, and any proposal directing money to NPP goes into the NPP quota. So if we fall below the quota cut, I'll ask Meifeng to remove our funding with the understanding that we contribute on a best effort basis.

As I've mentioned before, Meifeng suffers from the same thing we do: cut off from ASCR funding because she works too closely with domain sciences, ie she works with people like us. So from my point of view the principal purpose of the project would be to help her out with attracting support, in an area where she and her group can materially contribute to an important effort of ours. (She understands the failed previous attempt to inject CSI effort into our program, and that won't be repeated.)

My zeroth order plan for finding effort would be looking to our Rubin contingent (some small fraction), who by the time of this project starting would be back to ATLAS, but that's open to discussion and how things develop. We were told Monday that SDCC will take care of sPHENIX's distributed computing needs so we don't need to balance effort in that direction.

Comments and thoughts appreciated.

  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