|
The use of high performance computing (HPC) is well established in research groups throughout the Case campus. HPC exists in the form of Beowulf clusters and as high performance, multi-processor SMP computers. Here are just a few of the Case faculty who are making use of such resources in their research:
In the Department of Astronomy, Professor Chris Mihos' Linux cluster "Muleteam" consists of a heterogeneous set of compute nodes, including:
- 16 dual processor 3.2 GHz Pentium 4 Xeon nodes with 2 GB of memory and Myrinet interconnect,
- 4 dual processor 2.4 GHz Pentium 4 Xeon nodes with 2 GB of memory and Gigabit Ethernet interconnect,
- 10 dual processor 1.0 GHz Pentium III nodes with 512 MB of memory and 100M Ethernet interconnect,
and 4 Terabytes of storage.
The cluster uses the SuSE Pro 9.1 operating system, the OpenPBS queuing system, and the Maui scheduler. It is used to perform simulations of galaxies and galaxy clusters, as well as to analyze astronomical data from Case's Burrell Schmidt telescope and other ground- and space-based observatories.
Professor Walter Lambrecht of the Department of Physics was the recipient of a ClusterOhio hardware award in 2004. His cluster consists of 24 compute nodes (dual Athlon processors), a login node, a storage node, and a Myrinet interconnect. It is used for electronic structure calculations of semiconductors and magnetic materials using MPI parallelized density functional codes, such as the linearized muffin-tin orbital (LMTO) method. Electronic structure calculations form the basis for predicting bonding, vibrational, optical, transport and magnetic properties. This facility is shared among several faculty of the Physics department and their research groups. Among them, Professor Rolfe Petschek conducts research in quantum chemistry, an area he describes as "very, very hungry for computer power". He runs quantum chemistry codes, notably the GAMESS generalized atomic and molecular electronic structure package, in order to compute the optical and electronic properties of organic and metallo-organic molecules. In addition, the research group of Professor Philip Taylor studies soft materials (especially liquid crystals and polymers) using molecular dynamics simulations, which require a large amount of computer resources.
Professor Thomas G. Gray of the Department of Chemistry operates a 24-processor / 12 GB cluster based on 3.0 GHz Pentium 4 Xeon technology and a Gigabit Ethernet interconnect. Computational research in his laboratory currently addresses excited-state properties of metal-metal bonded clusters and non-frontier orbital reactivity in partially reduced nickel(I) hydroporphyrin anions. These theoretical investigations provide atomic-level information not accessible experimentally, yet these results are absolutely essential for understanding laboratory observations. An ongoing effort calculates the excited-state structure of emitting W 6 X 14 2- and Mo 6 X 14 2- clusters; these species have been proposed as luminescence bioimaging agents, and outside funding for their study has been approved. The same clusters are also promising singlet-oxygen generators for photodynamic therapy. A detailed theoretical understanding of the clusters' triplet excited state is essential for developing these new applications.
A separate project analyzes the non-frontier orbital reactivity in nickel(I) octaethylisobacteriochlorin monoanion. This species is held to be the fastest-reacting transition-metal-based nucleophile known, yet textbook bonding paradigms cannot account for its extraordinary reactivity. Correlated post-Hartree-Fock and density-functional calculations are proposed to understand the nucleophilicity. It is hypothesized that reactivity originates from an orbital other than the highest-occupied (a predominantly d z² -orbital localized on Ni). An extended Hückel calculation indicates this orbital to be the HOMO-10; preliminary DFT computations indicate the same orbital to be the HOMO-1 (where HOMO designates the highest-occupies molecular orbital). Regardless, non-frontier orbital reactivity, although known to photoelectron spectroscopists for years, is rarely ever considered in inorganic chemistry, and appears never to have been addressed in catalytic contexts.
Professor Steve Hauck's research in the Department of Geological Sciences involves simulation of the deformation and evolution of solid planets and moons in our solar system. A key component of his future research will be the ability to leverage larger than typical computational resources to perform these simulations.
Professor Randall Beer's work in the Department of Electrical Engineering and Computer Science involves the simulation and analysis of models of the neural basis of behavior. He typically works with interacting models of neural circuits, bodies and environments, and also runs simulations of biological evolution and developmental processes. All of these simulations are quite computationally intensive.
Professor Ed White of the Department of Mechanical and Aerospace Engineering was the first Case recipient of a ClusterOhio hardware award in 2002. His cluster consists of 32 processors (550 MHz Pentium III Xeon), 16 GB of memory, and Myrinet interconnect.
Professor James T'ien's Computational Combustion Laboratory in the Department of Mechanical and Aerospace Engineering is an advanced numerical simulation laboratory for the modeling and analysis of combustion sciences. The facility includes high-end UNIX workstations and Pentium-based Windows NT computers that are capable of solving a variety of computational fluid dynamics and combustion problems.
| 
