- Allen Malony
With the advent of high-fidelity simulations of complex physical phenomena made possible with modern high performance computing platforms, the problem of verifying and validating simulations has become a task with complexity comparable to the simulation itself. Traditional methods based on the L2-norm and manual inspection (commonly known as the 'viewgraph-norm') have reached their limits of usefulness. In particular, naive quantitative methods fail to track with domain-expert opinion. Similarly, the reliance on expert judgment often fails to provide quantitative metrics for quality of simulation output, especially when extending beyond the realm of simply reproducing known results to that of predictive science. This lack of rigor leads to the requirement for more sophisticated and mathematically sound techniques for quantitative comparison of simulations to experimentally obtained data. The algorithms that have been proposed for building such comparison techniques are sophisticated, and touch on many interesting algorithmic areas from computer science to address issues in both performance and correctness. This talk will provide an overview of the V&V process and problem, discuss the flaws with the current methods used in practice, and introduce the methods that have been proposed to replace the current state-of-the-art.
Matthew Sottile is a research associate and adjunct assistant professor in the University of Oregon Computer Science Department. Prior to joining the department in the fall of 2007, he was a staff scientist at the Los Alamos National Laboratory in the Computational Physics group. He earned his Doctorate in Computer Engineering in 2006 from the University of New Mexico in the area of high performance computing and signal processing.