NGVLA CSV Working group: 2019 May 28 Telecon

Details

• Date: Tues 28 May 2019, 19 UT = 15 EDT = 13 MDT
• Location: CV-ER331, SO-280
  • Audio Hub: 434-817-6286
• Participants:
  • CV:
  • SO:
  • Audio: none

Agenda

1. Document status
   • CSV Concept: Last minor update was April 9 based on comments from this group. I've heard nothing from SAC review.
   • CSV Plan: I have created an overleaf document based on Concept, but outline and contents will need to be changed.
2. How to structure the AIV and CSV Plans: AIV group has started a milestones table (snapshot shown below)
   • AIV/CSV linked until 6-antenna array
   • Aiming for C-config-like capability first, followed by C+D, B+C+D, etc.
   • Filling in IPT delivery pre-requisites as the columns
   • CSV will be a separate sheet -- shall we define a list of tested capabilities and facilities as the columns?
3. Begin considering and collecting FTE estimates for the Plan
   • Kern: Testing new features in control system, accuracy in M&C data, reporting errors. I think the list of tasks is:
     • Antenna Motion Scripts (3 months @ 0.5 FTE)
     • Correlator Control Scripts (3 months @ 0.5 FTE)
     • Alert System (18 months @ 0.2 FTE)
     • Pointing Model (4 months @ 0.5 FTE)
     • Correlator Model (4 months @ 0.5 FTE)
     • Focus Model (4 months @ 0.5 FTE)
     • Sequencing (6 months @ 0.5 FTE)
     • Transition to Scheduling Blocks (3 months @ 1 FTE)
     • Monitor Database commissioning (18 months @ 0.1 FTE)
     • I think we should budget to spend 2 weeks of 2 FTE with each new software release commissioning the new capabilities at the M&C level before they are ready to be used scientifically.
   • Rafael - Development of observing scripts and scheduling blocks for tests
     • Based on ALMA Obs. Scripts, the estimated size is 30,000 SLOCs.
     • Using the historical production metric of 30.0 SLOC per day and developer, this accounts for 3 FTE-years, including design, coding, testing, etc.
     • Assuming that 20% of this requires a scientist, and 80% a developer, I estimate a total of 0.6 FTE-years for a scientist, or 3 years @ 0.2 FTE.
   • Bryan - Testing of the calibration plan; filling calibration database (including polarization)
   • Vivek - Testing performance to specification (astrometry, stability, etc.)
   • Barry - Short baseline commissioning; testing correlator modes
   • Walter - Long baseline commissioning; phased array + pulsar modes
   • Todd - Autocorrelation and single dish commissioning (I drew the short straw)
   • John - Testing and commissioning of automated pipeline for each observing mode
     • On-going effort (not per-mode):
       • PL SSS: 1 FTE
       • Communication (exclusive of SSS): yearly F2F + monthly 2hr mtg w/developers + weekly 2hr mtg w/group = 0.12 FTE per person
     • per-mode effort:
       • Reqts capture: 3mo @ 1 person/mode @ 0.2 FTE (1 workday/week) = 13 workdays = 0.08 FTE/mode per yr
       • Testing/acceptance: 4mo @ 2 ppl/mode @ 0.4 FTE (2 workday/week) = 69 workdays = 0.43 FTE/mode per yr
       • (Above I use 1 FTE = 43 workweeks/yr * 5 workdays/workweek *0.75 functional duty fraction = 161 workdays/yr)
     • e.g.: 2modes commissioned in 1 yr (4ppl + SSS) = 1+4*0.12+2*0.51 = 2.5 FTE
     • For comparison - the NAASC SW testing group charged 3.6 FTE for FY2018, and 4.0 FTE for 8mo of FY2019 (direct hrs billed, not burdened by vacation etc). This is mostly heuristic development/testing or basic operations testing.
   • Joan - Validation of First Look science and other SV data against existing data
   • Eric - Documentation of observing modes, and release of data
   • Chris: Providing assistance to AIV (including after the AIV/CSV split)
   • Rob: Providing reports to management, serving on hiring committees, etc.
4. Work schedule for CSV
   • Kern: I think that we will want 1 scientist working with the team while we are observing. If we’re working bankers hours that is probably 1 FTE then, but if we’re doing day and night observing it is probably 2 FTEs. I suspect we will want at least 3 FTEs of scientist effort doing the analysis. That kind of breaks down to 1 FTE for FEND type issues (Tsys, WVR, gain stability, bandpass, etc); 1 FTE for delay and correlator; and 1 FTE for antenna integration (pointing and focus models, temperature stability, gain curves, etc).
   • So if we had 5-6 scientists (the 4-5 I listed above +1 for things I forgot and an occasional day off) I think that would be reasonable. My guess is that it will take us 18-24 months to get to the point that we have routine stable fringes among 6 antennas (or however many we have by that point.