Strong Lens Time Delay Challenge. Ii. Results of Tdc1

Abstract

We present the results of the first strong lens time delay challenge. The motivation, experimental design, and entry level challenge are described in a companion paper. This paper presents the main challenge, TDC1, which consisted of analyzing thousands of simulated light curves blindly. The observational properties of the light curves cover the range in quality obtained for current targeted efforts (e.g., COSMOGRAIL) and expected from future synoptic surveys (e.g., LSST), and include simulated systematic errors. Seven teams participated in TDC1, submitting results from 78 different method variants. After a describing each method, we compute and analyze basic statistics measuring accuracy (or bias) A, goodness of fit χ 2 , precision P , and success rate f . For some methods we identify outliers as an important issue. Other methods show that outliers can be controlled via visual inspection or conservative quality control. Several methods are competitive, i.e., give |A| < 0.03, P < 0.03, and χ 2 < 1.5, with some of the methods already reaching sub-percent accuracy. The fraction of light curves yielding a time delay measurement is typically in the range f =20-40%. It depends strongly on the quality of the data: COSMOGRAIL-quality cadence and light curve lengths yield significantly higher f than does sparser sampling. Taking the results of TDC1 at face value, we estimate that LSST should provide around 400 robust time-delay measurements, each with P < 0.03 and |A| < 0.01, comparable to current lens modeling uncertainties. In terms of observing strategies, we find that A and f depend mostly on season length, while P depends mostly on cadence and campaign duration.

Sn Refsdal: Photometry and Time Delay Measurements of the First Einstein Cross Supernova

“…The cost of this more flexible approach is that we lose the physical/empirical priors on the light curve shape and color that a well-matched template would afford. This second approach is therefore much closer to the methodology typically used for measuring lensed quasar time delays (e.g., Tewes et al 2013a;Liao et al 2015), where there is no way to apply an informative prior for the intrinsic light curve shape.…”

“…The time variation of quasars is stochastic, being driven by essentially random events on the accretion disk of the central supermassive black hole, so the intrinsic shape of a quasar light curve can not be known a priori. As such, quasar time delay methods must adopt a very flexible function to describe the light curve, and rely on purely empirical constraints (e.g., Liao et al 2015;Tewes et al 2013a). In contrast, for lensed SNe it should typically be possible to classify the SN based on both photometry and spectroscopy, and then identify a well-matched SN light curve template.…”

Crash testing difference-smoothing algorithm on a large sample of simulated light curves from TDC1

“…The publicly available simulated light curves of TDC1 with known true time delays are arranged in five 'rungs' having different sampling properties (see Liao et al 2015, Table 1), in increasing order of difficulty. Whereas COSMOGRAIL 3 -like rung0 light curves have observing seasons of 8 month duration, light curves of all other LSST-like 'rungs' have observing seasons of 4 month duration.…”

“…The TDC performance metrics as defined in Dobler et al (2015) and Liao et al (2015) are summarized below for the convenience of the reader. The success fraction or efficiency f of the time delay estimator is the fraction of light curves for which time delays have been submitted N sub with respect to the total number of light curves N available for analysis,…”

H0from ten well-measured time delay lenses

“…In order to test the robustness of our procedure for estimating the time delay and its uncertainty, we made use of synthetic light curves from the TDC1 stage of the Strong Lens Time Delay Challenge 1 (Liao et al 2015), which are arranged in five rungs having different sampling properties (see Liao et al 2015, Table 1). We applied our procedure on a sample of 250 light curves, 50 from each rung, selected such that we were able to reliably measure time delays from them.…”