**CLUSTER COMPUTING**

The equations for gravitational lensing by a single point lens can be easily solved, but for more complicated situations involving a source star of finite size, and a lens with several components, the equations rapidly become intractable.

A simulation method known as ‘inverse ray shooting’ is available (e.g., J. Wambsganss,* MNRAS Vol 284, p. 172*)
in which photons are fired through the lens towards the source with deflection *4GM/c ^{2}b* for each component of the lens. This can treat situations of arbitrary complexity, but it is computer intensive.

The MOA group uses a cluster computer at the University of Auckland to implement the inverse ray shooting
technique. This includes 400 1.5 GHz PCs, and is described

at astro-ph/0204478.

*In the inverse ray shooting technique, rays are shot from the observer through
the lens to the source, with an appropriate deflection for every component of the lens. Rays that hit the source
star are accepted. *