Principal Investigator George Verghese
The modeling of complex systems usually generates an overwhelming quantity of detail. Unwieldy systems can be made tractable by simplification through analytical studies, simulation and numerical computation. One challenge in simplifying a complex model is determining which variables are critical for systems-level understanding, and which variables can be discarded, suppressed, averaged or otherwise approximated. Modeling exercises often start by trying to incorporate all of the data about all of the components and their interactions, followed by sensitivity analysis to determine which variables have the greatest effect on system behavior. We are developing methodologies for identifying and retaining the important variables in complex biological models and reducing other variables so that we can simplify such models. We also are addressing problems such as noise, uncertainty and fluctuation so that we can build more robust models that do not generate and propagate artifacts.