General Comments

** Single laptop for all presentations

** Are we going to provide the code for the reviewers

George:

  • Welcome to the review
    • introductions


Andy:

  • Add a slide - what is opsim (an emulator of the gross telescope and site properties and a scheduler)
  • slide 6: m2 vendor selected for the figuring will take 2 years
  • slide 8: slew time is minimum not median (dont change)
  • integration time needs checking as is 16+16+2+2
  • Total number of exposures in 10 years
  • slide 12: Replace org chart with slide 16
  • note it is systems engineering and telescope and site (Scheduler is
  • under telescope and site)
  • slide 15: note just validate changes also validate that we meet the SRD
  • slide 17: Total simulation resources
  • In FTEs, all sims, highlight baseline resources for opsim (break this down
  • into two for sys eng and T&S) have off project resources as bar with no color
  • have numbers for opsim at bottom of chart (FTE)
  • Make the data size the same in the slide

Abi:

  • Add a slide saying who will address which charge
  • Requirements
  • slide 3: add numbers (eg calibration or number or visits)
  • slide 4: determined -> determine
  • slide 5: state clearly that your assumption and the design is assuming that opsim represents the telescope and site and you build the scheduler within opsim
    • so are the api's correct (apis are not defined but the design will support it)
  • slide 6: state that opsim is a tool to optimize the science it doesnt optimize science (others do)
  • Talk is very verbal - quantify it (examples are easier including the successes)
  • Identify who will talk about the different aspects in future talks (e.g. Srini will talk about SSTAR)
  • Needs to be shorter (30% fewer slides) and tighter and talk more on the outputs with examples
  • You mention requirements and SRD but don't discuss them
  • Slide 21: will the reviewers understand the plan as it is fairly complicated
    • note these will be discussed in a wrap up talk
  • Slide 25 - what is the new scheduler - you said 3.0 was the redesign. Need to define what is delivered by 2016.5
  • Reduce slides and move images (talk over the figures)
  • Ran over time

Francisco:

** how does this relate to the SRD requirements
** Move slide 14 forward to the beginning and move the second half of the talk to the beginning for context

** Move Srini's talk ahead of this?

** Rework figures to make them easier to read and cleaner than a sysml output

  • slide 5: need labels on the boxes that are readable
  • slide 10: completely unreadable Labels need to be bigger (or provide a handout)
  • slide 11: as above
  • Requirement to minimize the slew time - should be an external constraint as depends on science requirement.
  • slide 12: as above
  • slide 13: what does this mean - put labels
  • Zeljko - lots of boxes in the diagrams (put statements into bullets and then follow the sysml tool)
  • George - need an overall flow diagram, how do we make a decision
  • Slide 14 is part of this but need to be nicer
  • Slide 15 fonts too small, change the names for the internal version (as the names
  • are duplicated)
  • Slide 21 - where do the parameters come from
  • Describe the algorithm in the presentation (give the logic of the process)
  • Flow chart of what opsim is doing - in Abi's talk (?). This exists in previous talks
    • Start with this explanation
  • slide 24 - Area distribution proposals (how is this algorithm set up)
  • Explain what has freedom and what doesnt (eg selection of field does but model for telescope isnt arbitrary)
  • Slide 29: make this an image then could easily explain program boost
  • How does the SRD feed into opsim requirements. Opsim needs a fidelity that can distinguish between baseline and minimum specs (eg number of visits).
  • For example, how do errors in the sky model impact the fidelity of opsim and your sensitivity to the results
    • need SRD slides and what can opsim do = why should we pay $2.5M

Srini

  • Are we making the code available?
  • Put more words on the slides that cover the details of what you are saying
  • Dont refer to opsim 2.0
  • make mysqlworkbench available for the june dry run (or at least a schema browser)
  • slide 11: modification to scheduler algorithm
  • slide 13: How long does it take to run and how much memory

Kem

  • Slide 3: single visit depth numbers are input to opsim (or remove line)
  • Note what 3.61 means
  • Slide 7: what does design-design mean in the table (very cryptic naming conventions)
  • Write down what JC means on a slide
  • Slide 9: identify which are old and new code in the figure
  • Hard to see why we are comparing to 3.61 (need to explain that it is good and why)
  • Lead with back of envelope comparison then go to 3.61
  • Perturbation test - change a parameter and make sure that it makes sense

 

Steve

  • SRD slide - give Steve's SRD to Abi for earlier in the presentation
  • Slide 10 - mark desired numbers
  • Slide 17: do aitoff plot for these points
  • Slide 25: make axes text bigger
  • drop last metric example as it is complicated to explain in a short amount of time


Lynne

  • explain why we need enhancements and why we are implmenting maf, e.g. to support cadence studies by the science collaborations
  • extensible - used by science and opsim
  • note this is redesign of sstar
  • note the use of python, git, public repo etc
  • compare metrics across simulations
  • slide 9 and 14 of steve and lynne's presentation (same slide and show what things MAF can do)
  • check the power spectrum plots to ensure they are the same
  • comparing multiple surveys examples

Abi

  • slide 4: need to add api for OCS
  • need to add OCS emulator
  • what additional functionality is required to create a virtual OCS
  • if virtual OCS is implemented does T&S need anything else


     
  • No labels