CadenceWorkshopOutcome

– Cadence Workshop –
Conclusions, Report,
Recommendations and Plans
Knut Olsen and Steve Ridgway
August 15, 2014
Sponsoring organizations: NOAO and LSST
The Organizing committee:
Richard Dubois (SLAC)
Eric Gawiser (Rutgers)
Zeljko Ivezic (U. Washington)
Ashish Mahabal (CalTech)
Knut Olsen - Chair (NOAO)
Steve Ridgway (NOAO)
Michael Strauss (Princeton)
Beth Willman (Haverford)
What did we set out to do?
• Get quantitative input (metrics) on how a
given LSST schedule performs for specific
science cases
• First step towards optimizing the ultimate
LSST cadence
• Get ideas for further cadence strategy
exploration, constrained by boundary
conditions of hardware and key science cases
How did we organize it?
• Metrics Analysis Framework tutorial
• Plenary session to give background and set
rough boundary conditions
• Breakouts: Transients and variables, static
science, mini-surveys, and main survey
optimization
• Plenary sessions on breakout reports, Deep
Drilling, and workshop wrap-up and future
plans
Breakout group deliverables
•
•
•
•
•
•
•
•
•
•
A list of science cases for which the groups would like to provide metrics
For those science cases, a list of variables that would enter into their metrics
A translation of those variables into the output columns delivered by OpSim
Performance metrics in rough analytical (or pseudocode) form
A list of assumptions made in constructing the performance metrics
Identification of e.g. modeling work needed in order to construct a metric that
can be calibrated to provide absolute performance for a given science case
A brief oral report of the breakout group discussion
A brief written report (few paragraphs) for the workshop report
Coded performance metrics in Python and MAF
Input for main survey optimization, mini-surveys, deep drilling, commissioning
We made progress on all of these deliverables
Breakout Group Leaders
Mansi Kasliwal Alex
Kim Michael Liu Ashish
Mahabal Warren Skidmore
Michael Strauss John
Thorstensen Tony
Tyson Kathy
Vivas Lucianne
Walkowicz Michael WoodVasey Hu Zhan
Select Highpoints from Static Science
Breakouts
• Static science does care about cadence
• Agreement between diverse science groups:
– Dithering!
– Uniform depth
– Calibration
• And some tension:
– Restricted airmass vs. extended airmass range
– Extra-long vs. shorter exposures
Static Science Breakouts cont.
• Both agreement and tensions argue for
finishing work started on metrics for all
science topics
– Some work completed already (dithering for LSS)
– Many metrics identified and discussed
Power spectrum with and without dithering
From LSS group
Circular Variance of RotSkyPos
from Weak Lensing Group
Select Highpoints from Transient and
Variable Science Breakouts
• Variety of variables much larger than represented
by attendees
• Need simulated catalogs containing contributed
light curves targets in simulated images
• For rapid transients, importance of sampling
more rapidly than strictly uniform (3-day)cadence
– need for rolling cadence, mini-surveys and/or
deep drilling to cover shorter time scales.
Transient and Variable Science
Breakouts - continued
• LSST model cadence uses fewer visits than ongoing programs for detection of solar system
object – caution in planning
• Consider separating “15-second” exposures to
give better time sampling (evaluate efficiency
cost)
• Value of forced photometry on recent visits to
support discovery of events
Transient and Variable Science
Breakouts - continued
• Recommend call for proposals for deep drilling
and mini-surveys
• Metrics will be needed which account for
timing of availability of follow-up resources
Sampling Summary Using the FWHM
of the Samping Window Function
from the Slow Transients and Variables Group
Visit Triplets with Delta-t Less Than 3
Days
from the Fast Transients and Variables Group
Uniformity of temporal sampling within a
season for supernova cosmology
From SN Group
Select Highpoints from Main Survey
Breakout
• There is a strong desire to front-load some
programs (e.g. deep coverage of the WFIRST and
EUCLID fields). This competes with “continuous”
uniformity. Simulations are needed to explore
compromises that support both.
• The “Ten-percent for mini-survey” estimate
should be re-evaluated based on science-metrics
• A scheduling algorithm based on an economic
model with virtual money allotted to different
proposals might give a different schedule.
Main Survey Breakout - continued
• There should be a small amount of time
reserved for urgent and unpredicted follow-up
(mini-TAC)
• Solar system cadence may not be needed
away from the ecliptic
• Sharp airmass and latitude boundaries should
be reconsidered.
Select Highpoints from Mini-Survey
Breakouts
• There is exciting science to be done with LSST in the
Bulge, Plane, and areas containing the Magellanic
Clouds
• Existing cadence simulations and metrics good for
some things (e.g. Magellanic structure at large scale)
• But higher cadence and number of observations
needed for variability (MCs, Bulge & Plane)
• Crowding an issue for MCs and B&P, argues for limited
set of excellent seeing observations legacy value
• Light curve library proposal for commissioning time
argues for using commissioning to anticipate broadly
useful reference datasets
• Confusion limits vs. radius
from LMC center
• Saha et al. (2010) surface
brightness profile
• Limits at which
photometric errors due to
crowding<0.1 mag for two
seeing values
Plans for maintaining contact
• Participant list and mailing list (grow with new
volunteers)
• Confluence page (current report is initial
content, add metrics discussion, metric
results)
• Contact persons for each topic: breakout
group leaders
Metrics Followup
• Continuing support for MAF – Peter Yoachim and
Lynne Jones
• Where to send metrics –check into git repository
– https://github.com/yoachim/ContributedMetrics
– instructions will be in report and on workshop
Resource page, or contact anybody in OpSim group
• Provide documentation of algorithm so
motivation and logic is clear
• Provide example config file that includes
appropriate captions for figures
Plans for Report
• Outline to be circulated in ~1 week
• Expect contributions of 1-2 pages, figures
welcome
• Want summary of work performed, lists of
science cases with associated metrics,
assumptions, work remaining to be done
• Indications of alternate cadence exploration
When Should We Have Next
Workshop?
• Goal: have users experiment with alternate
cadence calculations
• When: ~1 year from now
• Where: ?
• Pacing item: making OpSim a tool for
experimentation by community
• Intermediate work: continue work on getting
metrics coded into MAF, recommending
directions for cadence exploration
What worked well?
•
•
•
•
Time between sessions; relaxed pace
Availability of broad array of Project people
Enthusiastic and engaged group
Good preparation: workshop content, meeting
support
What didn’t?
• Conflicting parallel breakouts
Thanks
• The very enthusiastic and engaged participants
• OpSim group who devoted time and energy to
making this workshop happen (especially Lynne
Jones and Peter Yoachim for their work on MAF)
• The 23 Breakout leaders who volunteered their
time and effort
• LSST 2014 Administrative and Technical Support