4. Geometry of PG 1115+080
Five images were used to measure the position of the lens galaxy. For each frame, the deconvolution procedure returned the coordinates of the four point sources, the deconvolved image as well as the image of the diffuse objects (Fig. 1). The latter was used to measure the position of the lensing galaxy.
To compare our results with K93's results, we transformed our coordinates to the coordinate system they used. The scaling factor and rotation angle were chosen to match the positions of the four point sources as well as possible, and the transformation was then applied to the galaxy coordinates. Fig. 2 compares the new positions with the positions from K93. The precision of the galaxy position has been increased by more than a factor of 3, from 50 mas with the pre-refurbishment HST images (K93), to 15 mas with the new data and deconvolution technique. Note that K93 used a pixel scale of 0:0004389, which Gould & Yanny (1994) revised to 0:0004374. Table 1 summarizes the geometry of PG 1115+080, in the same orientation as K93, but using the pixel size given by Gould & Yanny (1994).
Table 1. Relative positions of the four lensed images of PG 1115+080 relative to component C (in arcseconds). The intensity ratios in the I band are given for the NOT observations obtained on the night of 1996 June 7. The magnitude of A1 is
The galaxy position was measured on each deconvolved background by the first two authors independently, using either Gaussian fitting or the first order moments of the light distribution, resulting in 10 measurements (one for each author and each frame). We first averaged the two measurements for each frame and then computed a weighted mean of the 5 independent measurements, with a weight twice as large for the NOT images as for the CFHT images because the latter are corrupted by PSF variation across the field.
The standard deviation of the mean is mas for the point sources and mas for the lensing galaxy. The effect of a variable PSF was tested by performing numerical simulations which indicate that possible systematic effects could affect the point sources by another 2 mas and the lensing galaxy by an additional 10 mas. This leads to final 1 error bars of 3 mas for the point sources and 15 mas for the lensing galaxy.
The intensity ratio of A2 relative to A1 is given in Table 1. This ratio does not show significant variations among our frames or between our frames and the K93 frames, even though they were taken at four different epochs. This suggests that microlensing does not significantly affect the light curves of A1 and A2 in the I band.
The galaxy magnitude was derived on the NOT deconvolved frames by aperture photometry (0:009 diaphragm in diameter). The zero point was computed using several standard stars. Images with a much higher signal-to-noise would be necessary to derive the shape of the galaxy. However, the galaxy observed in the present data is compatible with a fuzzy circular object, broader than a point source.
© European Southern Observatory (ESO) 1997
Online publication: May 26, 1998