Skip to end of metadata
Go to start of metadata

Attendees

Goals

  • Improve strategies for finding and assisting users from non-traditional communities
  • Report and track work with NIP research groups

Discussion items

TimeItemWhoNotes
5 minWelcome and Agenda AdditionsPhilip Blood

Zoom Meeting ID: 647 898 865

 Join from PC, Mac, Linux, iOS or Android: https://zoom.us/j/647898865

Or iPhone one-tap: US: +14086380968,,647898865#  or +16468769923,,647898865#   

Or Telephone?US: +1 408 638 0968  or +1 646 876 9923  or +1 669 900 6833

 

 

15 minNIP focus: Harnessing the Data (and AI) RevolutionPhilip Blood, Sergiu Sanielevici, All

Should we position ourselves as those helping spearhead XSEDE's efforts to help users doing data-centric and AI work?

One of NSF's 10 big ideas for the next 10 years: https://www.nsf.gov/about/congress/reports/nsf_big_ideas.pdf

 

  • Sergiu Sanielevici: NSF questions at XSEDE mid-year review:
    • How does XSEDE get ahead of the curve?
    • How do we anticipate the needs of the community?
    • Where do you anticipate that community will need the most help?
      • Philip Blood: (Aside: Perhaps get data from Dave Hart's NIP reports and look at trends)
  • Sudhakar Pamidighantam: There are many Big Data projects and these will grow. It is more than NIP can handle alone. Perhaps it will eventually merit it's own section of ECSS
  • Philip Blood: Should we become the group that gets ahead of the curve – really novel and innovative? We are doing this but can tighten our focus. Additional training and information sharing. Attendance at conferences.
  • Sergiu Sanielevici: The questions from NSF above were driven by the question of onboarding researchers to new resources like Jetstream and Bridges
  • Alan B Craig: Maybe we should think of this as a matrix between techniques and domains/disciplines – This adds value as we can help connect people from different domains (leveraging domain champions)
    • Example: Streaming data techniques cross many fields
  • Emre Brookes: Like the idea of the matrix and NIP being the translator between them
  • Derek Simmel: How do we capture history and experience working with different areas and give people tangible examples of the kind of projects we have done to help people see how they would use NIP.
  • Alan B Craig: It may be good to have a third matrix dimension of the resources and what types of projects are intended (and maybe have historically run) on those systems.
  • Davide Del Vento: We need to be careful about just advertising expertise in techniques that many researchers may not be familiar with. May be useful to keep information about domains as it's easier for people to associate with.
  • Paul Rodriguez: Advertise simple consulting opportunity as a first step, e.g., if they heard about deep learning and wonder if it would be beneficial for their research
  • Derek Simmel: Can people direct potential NIP projects to us as requests come in?
  • Paul Rodriguez: Maybe we need something even before someone applies for an allocation
  • Sudhakar Pamidighantam: How do we present what NIP does to current users or potential users? Is there a service to help people with preparing proposals? Sergiu Sanielevici: at the moment this does not exist. And also there is a request for someone to talk to PIs who got rejected, but a systematic way for doing this has not been devised. Sudhakar Pamidighantam: Perhaps an ECSS person should be assigned to call/talk to users
  • Derek Simmel: Can we be involved earlier in the process, like during the presentation of "How to submit a successful XSEDE proposal"
  • @Tom Doak of NCGAS: How to submit successful proposal document is a bit terrifying for small/new users and doesn't apply as much to them. E.g.: scaling analysis is problematic for a history professor. Sudhakar Pamidighantam: a gateway is maybe a good solution or the Startup. Sergiu Sanielevici: Remember as XSEDE we are helping people get where they need to be, not just to the SPs, so if we help them find a better resource that is a success.

15 minFollow up on Discussion: Increasing number of new users from NIP communitiesPhilip Blood, All

Can we be more proactive in seeking out new users from non-traditional communities? Perhaps by more actively attending and presenting at strategic conferences? Other ideas?

Identifying NIP people who come to XSEDE:

  • Should people be able to request NIP help specifically (or at least self-identify as a project that might need NIP assistance?)
    • Highlight NIP as a new/special resources?
      • Put a one-liner in the high-level title so people know what it is
      • Add some description on the page and allow people to ask for help
    • Can we put something in the critical path that forces people to indicate whether they want NIP help?
      • Alternative if making NIP a requestable resource doesn't work
    • How do we process these requests?
      • NIP does initial orientation and mentoring and then hands off to other ECSS if serious work is needed.
  • Help researchers from non-traditional communities write proposals or rewrite rejected proposals
  • Specifically requesting that XRAC reviewers identify people in need of NIP support

Finding NIP people who don't know about XSEDE:

  • Targeting conferences
    • Who is attending conferences with non-traditional communities
      • Sudhakar: ACS meeting has experimental areas that are not traditional HPC users (cheminformatics)
      • Emre: crytallography, small-angle scattering, ultracentrifugation. A lot of these don't have unmet needs
    • Can we identify meetings where we should go?
      • Work with Domain Champions to get the right conferences and attend and present at those conferences
      • Conferences in a non-traditional domain with something like technology track may be of interest
  • Work with XSEDE Community Engagement and Enrichment
    • Have a standard NIP blurb that those doing outreach can share in all their presentations
    • Alan B Craig: has a one-pager about Humanities that Linda Akli shares at outreach events --> We could have domain-specific outreach material to share with CEE
    • Sergiu Sanielevici: Have a one or two page handout on NIP in general
    • Previously had a NIP tutorial at one of the XSEDE conferences (Sergiu Sanielevici, Alan B Craig)
      • Could look up evaluations of the tutorial
  • What about domains that we don't have experience in?
    • We'll triage based on skills and interest of current staff, request other ECSS personnel as needed, and watch for trends that suggest we need to bring new expertise into XSEDE ECSS
  • How will we handle new influx of NIP projects?
    • We will begin tracking all NIP projects in JIRA
    • Phil will periodically send email to ecss-nip@xsede.org requesting that people bid on projects they are interested in reaching out to or working with on a specific need.
    • Projects that are not bid on will be discussed with NIP staff and assigned. Phil will look at Skills and Interests in NIP staff Confluence profiles to identify likely staff members to work on projects. These will also be discussed in NIP calls

 

 

20 minCurrent NIP efforts + HighlightsPhilip Blood, All

Where are your current NIP efforts focused?

Paul Rodriguez: Ying Xie and Mina Ameri, University of Texas at Dallas. Paul is working with this group on benchmarking.

Feng Kevin Chen: Lars Hinrichs, UT Austin: https://jira.xsede.org/browse/ECSS3-4531?jql=text%20~%20%22hinrichs%22

Joel Welling: Anela Choy (MBARI): Transferred to NIP from ESRT: https://jira.xsede.org/browse/ECSS3-156
Sergiu Sanielevici was going to make initial contact with this project. Status?: https://jira.xsede.org/browse/ECSS3-4509

Projects we are aware of and helping. Sergiu has a list at the bottom of the main NIP page: WBS 2.2.3 NIP: Novel & Innovative Projects.

Marques Bland now has these projects in JIRA:

T Key Summary Assignee Reporter P Status Resolution Created Updated Due
Loading...
Refresh

We will have JIRA training at the SP sites. Philip Blood will invite non-SP NIP members to join the PSC training.

5 minAOB, next meetingPhilip Blood

Reminder: SGCI can now be allocated through XRAS ECSS (help with User Interface side).

Next meeting: Feb. 12, 2018

Action items