Department & Program:
Department of Physics & Astronomy, M. Sc.

Funding:
Faculty of Science Graduate Studentship Award

Research Focus:
Gravitational lensing is an observed phenomenon that was predicted by Einstein using the theory of General Relativity in 1916. This effect predicts the behavior of light rays near massive objects - when a ray of light passes a sufficiently massive object (usually a cluster of galaxies), the gravitational field of the object causes the light ray to bend. When such an object is located between a distant "source" galaxy and an observer on the Earth, the source appears shifted from its actual position. In fact, at some positions behind the lensing mass a source may even produce multiple images. The properties of the images formed in this way are related to the density distribution of the lens, which can be expressed by a physical model characterized by a set of parameters that describe the physical properties of this distribution. With the correct set of parameters, the original source can be reconstructed from the observed distorted image. In practice this is very difficult since we do not know what the source should look like or the exact properties of the lens density distribution, since galaxies also posess an unseen halo of dark matter.

Through my research I have developed a unique method of finding the parameters of the lens density distribution using a powerful genetic algorithm called “Ferret”, designed by my supervisor Dr. Jason Fiege. Genetic algorithms are a class of search and global optimization algorithms that function in analogy to biological evolution. An initial population of models are selected and a “fitness” value is calculated for each member. These models are then subjected to “mutations” (random changes to parameters) and “crossover” events (in which pairs of solutions are combined). As in biological evolution, the fittest models survive, and the cycle repeats. The end result is the best possible reconstruction of the source galaxy, and a model of the dark matter distribution in the gravitational lens. This project is only possible because of our innovative techniques for dealing with gravitational lenses which were developed during the course of this research, and the power of the Ferret algorithm.