Observing Strategy Considerations for Strong Lensing Science Intro - - PowerPoint PPT Presentation

Observing Strategy Considerations for Strong Lensing Science

Intro

• Strong gravitational lensed are rare events (1 in 104 galaxies

capable of being a lens)

• Requirements for general strong lens discovery:

○ Wide area with reasonable sensitivity in all bands (increases sample size) ○ Good image quality (to discern lensed images from lenses, better Rein sampling, accurate image positions) ○ Blue sensitivity (detect typically blue SFGs)

• Strongest observing constraints:

○ Strong gravitational lens time delays (lensed QSOs & SNe) ○ This talk: LSST considerations only, but in practice, high resolution imaging and spectroscopy follow-up are typically required

A quick word on static lenses

Most lenses have small Einstein radii. Most lensed sources are blue. Some of the good seeing time should be allocated to g-band. This will maximise LSST’s strong lens discovery potential.

Collett 2015

Theoretical Einstein Radius distribution ‘Log number of lenses’ Einstein Radius

Strongly Lensed Variable Sources

Light from a multiply imaged background source takes different paths through the lens potential Variable background sources have differences or time delays in their periodic variability

1/H0 Lens model Predict the delays

Quasar Time Delays

• Discovery

○ Single visit depth yields few thousand lensed QSOs (Oguri & Marshall 2010) ○ ~few hundred suitable for time delay science (Liao et al. 2015)

• Cadence

○ Capture delays of several days - several weeks

• Considerations

○ Night-to-night cadence ○ Season length ○ Campaign length Single galaxy lenses have best determined models LSST cadenced

• bservations (+

detailed follow-up) can provide the needed few % precision on time delay distance Need ~100 such time delay systems to constrain H0 to sub-percent level

Time Delay Challenge 1

Liao, Treu, Marshall, et al. 2015

• 5 different i-band light

curve datasets (“rungs”)

• 1000 lenses in each
• Challenged the community
• Assessed results through

○ Time delay accuracy ○ Time delay precision ○ Usable sample fraction

TDC1 precision, accuracy, success fraction - as an approximate function of observing strategy diagnostic

MAF Analysis

Sky maps of lensed quasar time delay accuracy, for minion_1016 Accuracy improves with campaign length (known). Bottom: 10 years, cf Center: 5 years Use all filters to increase night-to-night

• cadence. Top: ri only, cf Bottom: ugrizy

ugrizy improves distance precision from 0.42% to 0.24% (10 years) kraken_1043: early attempt to split visit pairs, shows marginal improvement in distance precision over baseline.

Better time sampling improves time delay precision

Alt_sched gives improvement

• ver baseline

by almost a factor of 2

Rolling cadence may not be best for lensed QSO time delays

Fewer seasons with good sampling - so accuracy falls. Need to compromise on accuracy threshold. Then, precision is comparable between rolling cadence and baseline. Need to revisit definitions in metrics!

Note on TDC2

DESC project, to be rebooted this year. Variety of cadence strategies. Multi-filter (TDC1 was i-band only). Lensed SNe time delays - better probe than QSOs?

Lensed Quasars: Cadence Needs

Maximize sky area, plus all three of season length, night-to-night cadence, and campaign length (no. of seasons). Alt-sched sims (Huber & Suyu) show significant improvements in success fraction and dt

• precision. However, mothra_2045 rolling cadence seems to give

approximately similar performance (Anguita et al, yesterday), but metrics need modifying to compute performance year-by-year.

SN time delays from LSST

Look for high luminosity SNe in elliptical “hosts” ~900 lensed SL SN 1a in WFD (Goldstein & Nugent 2017, Goldstein et al. 2018) ~100 (1a and cc) SNe spatially resolved (Oguri & Marshall 2010, Goldstein & Nugent 2017) suitable for time delay analysis

Goobar et al. 2017 Kelly et al. 2015, 2016 Goldstein & Nugent 2017

Lensed SNe time delays vs cadence

simulate lensed SNe light curves given LSST cadence strategy

Slide from S. Suyu & S. Huber

measure time delay PyCS (Tewes++2013, Bonvin++2016) alt_sched_rolling performs best: recover dtinput within ~4%

Lensed SNe: Cadence Needs

Lens sample size increases linearly with sky area and campaign length. Cadence needs are similar to unlensed SNe: higher night-to-night cadence is better. Alt-sched sims (Huber & Suyu) show best dt precision: 4% per system LSST only: time delays possible for only 10% lensed SNe and need higher cadence to achieve 4% accuracy Higher cadence = lower area: discover proportionally fewer lensed SNe for follow-up (i.e. with LGSAO, HST, JWST, Euclid, WFIRST, etc.).