Development of Tools to Enable the Port of Software to a Beowulf Cluster

This proposal, from a RUI institution in an EPSCoR state, developing tools and documentation, aims at reducing the learning curve and lowering the preparatory barriers that inhibit scientists from taking advantage of multi-processor computations. The project, proposing a 100 node Beowulf cluster, is expected to:

a. Assist in the conversion of single processor applications to a parallel computational environment, and

b. Reduce learning time and code porting time for scientists to effectively utilize the computational power of multi-processor systems.

The work spans two colleges, Engineering and the Arts and Sciences, and enjoys participation across six academic departments. Various problems, including topics in hydrology, seismology, civil engineering, atmospheric fluid mechanics, oceanic currents, wave propagation, mathematics (numerical methods and solving ODE's in parallel), electromagnetics, neural networks, will be ported to the cluster and include,

a. Inelastic Wave Propagation, Soil Dynamics, Soil Permeability, Atmospheric Modeling,

b. Parallel Computation, Character Recognition.

c. Hydraulic Tomography, Waveform Relaxation Methods,

d. High-Energy Astrophysics, and Nanotechnology.

The interaction between a team of computer scientists, client scientists, and students should result in a user paradigm facilitating the conversion of the diverse scientific problems. The platform should serve as an asset for future research and educational activities.