Title: Simulation Validation Using Causal Inference Theory with Morphological Constraints

Speaker: Dr. Bill Reynolds, LANL

Date/Time: Tuesday January 25, 2011, 9:30am       

Location: CSRI Building/Room 90 (Sandia NM)

Brief Abstract: We present an approach for the validation of complex simulation based on the structured elicitation of expert knowledge. Knowledge capture is based on the technique of Morphological Analysis, which is used to capture expert information on causal linkages and constraints in a system. This information is combined with Causal Inference Theory arguments to develop assertions about statistical dependency relations that should exist in a system. Causal Techniques for conducting these tests, which include the elicited constraint information are described.  Overviews of Morphological Analysis, Causal Inference Theory and Statistical Testing Approaches are provided in the context of a Bayesian simulation of an example problem.

Speaker Bio: Dr. Reynolds graduated in 1986, magna cum laude, with degrees in Physics and Comparative Literature at the University of Massachusetts. He followed this with a Ph.D. in Theoretical Physics from the University of California, San Diego in 1993.  He has been with LANL, SFI, and for the past 12 years has been head of Least Squares Software in Abq which specializes in analysis of complex systems for customers throughout the country.

CSRI POC: Mark D. (Danny) Rintoul, 505-844-9592