Free Access

Table 2.

Estimator parameters used to compute the time-delay estimates presented in Fig. 3.

Free-knot splines Regression difference (*)
C2 η [days] [25, 35, 45] ν 1.7 2.2 1.9 1.3 1.8
A 0.5 0.4 0.6 0.3 0.7
ηml [days] [150, 300, 600] Scale 200 200 200 300 250
Errscale 20 25 20 25 25

ECAM η [25, 35, 45] ν 1.7 2.2 1.5 1.3 1.8
A 0.5 0.4 0.4 0.2 0.3
ηml [150, 300, 600] Scale 200 200 200 200 200
Errscale 20 25 20 5 5

SMARTS η [35, 45, 55] ν 2.2 1.8 1.9 1.3 1.7
A 0.5 0.7 0.6 0.3 0.7
ηml [150, 300, 600] Scale 200 200 200 150 300
Errscale 20 25 20 10 25

WFI η [25, 35, 45] ν 1.7 1.8 1.3 1.5 1.9
ml model Splines 3rd poly. A 0.5 0.6 0.3 0.4 0.7
Scale 200 200 200 150 300
ηml pos. [Free, Fixed] Errscale 20 25 20 10 25

Notes. η corresponds to the initial knot spacing of the intrinsic spline and ηml to the initial knot spacing of the extrinsic microlensing splines, in days. For the WFI data set, “ml model” indicates whether the microlensing has been modeled using free-knot splines or polynomials. For the former case, ηml pos. indicates the constraint on the microlensing spline knots. Brackets indicate that the values within have been tested in all possible combinations – each data set has thus nine different possible combinations. For the regression difference technique the parameters ν (smoothness degree), A (amplitude in magnitudes), scale (length scale in days) and errscale (observation variance in days) refer to the Matérn covariance function used in the python 2.7 Gaussian process regression implementation of the pymc.gp module. The rightmost column (marked with an *) has the Matérn covariance function replaced by a power-law covariance function, where ν indicates the power-law index.

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.