Quick Intro to the Metrics Analysis Framework Peter

Quick Intro to the Metrics Analysis Framework Peter Yoachim University of Washington Vera C. Rubin Observatory | SCOC-Science Collaborations Workshop | December 2020

So, we simulated 2 million observations. How well can you do science with them? The simulation outputs a database with all the relevant information you would expect from a log book or fits headers • RA, Dec, MJD • Exposure time • Filter • Seeing • Sky brightness • Airmass • Sun altitude • 41 total columns, documented in the repo: https: //github. com/lsst/sims_feature. Scheduler Vera C. Rubin Observatory | SCOC-Science Collaborations Workshop | December 2020

3 components of MAF • SQL query • filter=“r” • night < 365 • “” (just grab everything) • Slicer • Grid on whole sky or a subset of points • Metric • What to compute at each point from the slicer Vera C. Rubin Observatory | SCOC-Science Collaborations Workshop | December 2020

Example: Coadded depth over the whole sky SQL: filter=“r” Slicer: Healpix. Slicer Metric: Coaddm 5 Metric Vera C. Rubin Observatory | SCOC-Science Collaborations Workshop | December 2020

The metric code At every point the slicer has, take the 5 -sigma depths of overlapping observations, convert to flux, sum, convert back to mags. Vera C. Rubin Observatory | SCOC-Science Collaborations Workshop | December 2020

The slicer finds all the observations that overlap with a given HEALpixel center. For display purposes, we’re coloring the whole pixel by the value at the center (a fine approximation at high resolution, maybe not great at low resolution) Both MAF and the scheduler use a grid of reference points, not contiguous area. Vera C. Rubin Observatory | SCOC-Science Collaborations Workshop | December 2020

By default, MAF assumes all points within the science field of view get observed. Technically, sometimes a point will fall on a chip gap or wavefront sensor. Also, sometimes a point will be outside the science Fo. V and get observed. For the most part, this should even out, but expect metric values to be off by ~1%. We can turn on the full focal plane geometry, it’s just really slow (but not too bad for things like checking DDF dithering strategy) Vera C. Rubin Observatory | SCOC-Science Collaborations Workshop | December 2020

The other common way to run MAF: Define a population of objects, only look at those points on the sky Fast microlensing events • Distribute based on stellar density • Give each event a start time, light curve properties Vera C. Rubin Observatory | SCOC-Science Collaborations Workshop | December 2020

The metric code generates each light curve and tests a detection criteria. Vera C. Rubin Observatory | SCOC-Science Collaborations Workshop | December 2020

MAF tips If you’re not using the 5 -sigma depth of each visit for your science metric, you’re probably doing it wrong • We take observations in twilight, bright time, and sometimes 1 -5 s exposures • No simple way to only select “good” observations with SQL, need to do that filtering in your metric code • You probably want to compute the signal-to-noise ratio of some fiducial object (e. g. , an r=20 star or r=22 QSO) or • Compute the faintest object you can reach to some desired SNR. • You don’t need to select “only WFD observations”. There is no such thing as WFD-only. Just run on the full sky. Vera C. Rubin Observatory | SCOC-Science Collaborations Workshop | December 2020

Getting started using MAF • Can install locally: https: //confluence. lsstcorp. org/display/SIM/Catalogs+and+MAF • Can run via: https: //datalab. noao. edu/ Vera C. Rubin Observatory | SCOC-Science Collaborations Workshop | December 2020 11

It works! We might need to setup some shared data, but you can launch a notebook and go. Vera C. Rubin Observatory | SCOC-Science Collaborations Workshop | December 2020 12

Conclusions • MAF is up on: https: //datalab. noao. edu/ • We want your metrics! • Talk to us early. We’ve written lots of metrics at this point, and the most useful ones tend to be the ones with the most collaboration Vera C. Rubin Observatory | SCOC-Science Collaborations Workshop | December 2020 13