Results

We have been applying PSA to a number of scientific problems with some substantial success. We have been using it to analyze PC images of known gravitational lenses and QSOs which are candidates for lensing (Falco 1993). One inspects the power spectra for unusual features such as unexpected asymmetries or unusually rapid falloff at high frequencies. It is very easy to quickly assess whether any of the candidate QSOs have close companions by looking for fringes in the power spectra. Our greatest success to date has been in analyzing PC images taken on a known gravitationally lensed system, Q0142-100, which has a bright image to the west and a fainter image to the east. The lensing galaxy has been detected from ground based images (Surdej et al. 1988) and models for the system were generated by the same group. Examination of the power spectra from the PC images reveals that, in fact, the brighter image is actually double, more or less in a radial direction from the weaker image.

Fig. 1 plots the higher contours of the PC image of 0142 in the upper left, and its power spectrum in the upper right.

For comparison, the image of a star is shown in the lower left and its power spectrum is in the lower right. The power spectrum of 0142 shows fine fringes which are about from horizontal, due to the widely separated pair of lensed images. It also has a highly elongated envelope to the fringes which is characteristic of a close binary. In comparison, the stellar power spectrum is much more symmetric, particularly at high frequencies. If we fit the envelope of the power spectrum with a binary model, we find that best fit gives a separation of the two components of 1.3 PC pixels or about 58 mas. Since this separation is too close to actually resolve the two components in the image, one would not know whether this was an elongation of the object or a truly separated binary image. However, gravitational lens optics require it to produce two separated images rather than an elongated image. Surdej et al. (1988) have generated different models of the system which depend on the exact superposition of the lensing galaxy and the QSO. One model generates only two images, another breaks the bright image into separate tangential images, while a third gives radial components for the bright image. Our detection should precisely proscribe the system geometry. The fit also gives a relative intensity of the two components as 5 to 4. Due to the centro-symmetry of power spectra, it is not possible to tell which component is the brighter. However, this can be ascertained from the image, as described below.

Fig. 2 shows cuts through the 0142 and field star power spectra.

The upper two curves are - cuts through the star power spectra, showing no obvious asymmetry between two directions. The lower two curves are cuts through the 0142 power spectrum. The lowest curve is across the fringe and the higher curve is along the fringe. The strong asymmetry is obvious.

We have also gone back to the image and subtracted the PC PSF generated by Tiny TIM for that position, with alignment between the brightest pixel of 0142 and the PSF. There is a strong residual, somewhat fainter peak to the west of the brighter peak, presumably corresponding to the second component. A surface plot of the residual peak after subtraction is shown in Fig. 3. A similar subtraction from the field star results in much smaller residuals.

Better PSF modeling is needed to get really accurate fitting in the image. However, earlier attempts at detection of close companions using standard point source fitting routines (before it was found from the power spectrum) did not find this peak. Knowing exactly where to look from the power spectra makes the point source fitting much easier. This extremely close pair of images would have only a few days time delay between the optical paths, so if the QSO varies in brightness and regular observations were made, it could be an ideal candidate for making time delay measurements in order to calibrate the extra-galactic distance scale. One would expect that the relative brightness of the two components of the bright source would vary with time and cross-correlating the separate brightness variations would allow extraction of the time delay.

We are currently analyzing the data for a number of other lens systems and lens candidates. We hope to be able to use PSA for measuring the shape and position of lensing galaxies, for the detection of additional components, and for possible detection of structure associated with the QSO.