Dr. Lorie Liebrock's research interests focus on issues related to parallel computing. One long term focus is on using problem topology during compilation. In particular, the use of topology to automate data distribution and allow application of regular application optimizations to partially regular problems. She has developed algorithms for automatic distribution of irregularly coupled regular mesh (a.k.a. composite grid or multiblock) problems, e.g., aircraft aerodynamics and water-cooled nuclear reactor simulations, via the use of problem topology. For use with these automatic distribution algorithms, she has developed a program template and a set of style guidelines for these applications that allow automatic transformation of an application code with no notion of data distribution into a standard High Performance Fortran program with complete distribution specification.

Adaptive parallelization is another aspect of her research. She is working with Pat Teller at University of Texas in El Paso to explore how to adapt parallel applications to new and/or changing parallel environments. By looking at a number of applications, tools that span a variety of uses will be developed to support building adaptive parallel applications.

Well posedness is another aspect of computation that she is exploring. She has done a preliminary well posedness analysis of the automatic data distribution problem. She is also exploring sensitivity analysis (aka stability in well posedness analysis) for complex additive systems with ICASA.

Additional Resources