Abstract:
|
Uncertainty quantification approaches are often necessary for a complete analysis of multi-scale systems. Multi-scale systems, where multiple small scale models are embedded in a larger physical system are common in the physical sciences and engineering. Without adequate representation and propagation of small-scale uncertainty, there is a serious risk of predictions that are inaccurate and/or overconfident. There are a variety of challenges in the analysis of multi-scale systems; large parameter calibration/emulation, complex model discrepancy, experimental design, and dearth of data for the full system among others. The methods to deal with these challenges will be discussed using multiple applications in the physical sciences and engineering, including laser induced breakdown spectrography (LIBS) measurements and forward models as well as chemical engineering/carbon capture systems and associated measurements.
|