High Dimensional Model Representation (HDMR): Concepts and Applications
Genyuan Li
Department of Chemistry
Princeton University
Princeton, New Jersey 08544
genyuan@straus.Princeton.EDU
Joint work with Herschel Rabitz.
A general set of quantitative model assessment and analysis tools, termed High Dimensional Model Representations (HDMR), have been introduced recently for improving the efficiency of deducing high dimensional input-output system behavior. HDMR techniques are based on optimization and projection operator theory, which can dramatically reduce the sampling effort for learning the input-output behavior of high dimensional systems (i.e., a reduction of effort from exponential scaling to only polynomic complexity). HDMR can be applied for different purposes: creating an efficient fully equivalent operational model, identification of key model variables, global uncertainty assessments, efficient quantitative risk assessment, etc. In one domain of applications significant computational enhancements have been observed in certain atmospheric model calculations.