Principal Investigator Martin Bazant
Co-investigators Alan Edelman , Rodolfo Rosales , Linn Hobbs
Project Website http://www-math.mit.edu/~bazant/AMCL/
The Applied Mathematics Computational Laboratory is a facility for high-performance scientific computation in the Department of Mathematics, featuring a 32-processor Beowulf cluster (16 dual-processor Athlon MP 2000+, each has 1.67 GHz CPU and 1 GB memory): heaviside.mit.edu. The AMCL is available to the entire Department and other special users at MIT for both research and educational purposes, subject to the usage policy. The Beowulf cluster is being used in simulations of granular flow in pebble-bed nuclear reactors and of radiation damaged ceramics, waves in lakes.
Currently, our largest parallel simulations (by Chris Rycroft) involve a granular dynamics code originally developed by the group of Gary Grest at Sandia National Laboratory. For example, it takes several days of real time on 20 processors to simulate the complete drainage of 50,000 monodisperse spherical pebbles from a model pebble-bed reactor. The MIT reactor will contain roughly 400,000 pebbles, so we are planning to increase the size of our Beowulf cluster and to improve its efficiency with faster switches.