Example: PMN J0134-0931

This is a radio lens discovered by Josh Winn and collaborators. It was initially puzzling, because there are six ratio components (labeled A-F above) in an unusual configuration. This lens provided a fun case of using randomize to search model space and figure out what is going on.

We used the radio data to speculate that components A-E were images of one source, and F was an image of a second source. There is no way to reproduce the observed configuration with a single galaxy, so we postulated that there may be two galaxies. We used randomize to search through the parameter space of two elliptical galaxies, and kept track of the models that fit reasonably well. Here are the isodensity contours for 100 models:

Remarkably, the range of successful models is fairly narrow. The positions of the two galaxies are fairly well pinned down, and there seem to be two families of shapes. We see certain univeral properties in the critical curves and caustics as well:

Although we did not demand it, the models all predicted that the source S₂ producing image F lies in a 3-image region, just outside the 5-image region that produces images A-E:

This leads us to predict that there should be two more images of this source, which have not yet been observed: