A Similarity Metric to Assess Upper and Lower Probabilities.

Uncertainty abounds in physical system analysis and simulation. A variety of measures of uncertainty are built around measuring the amount of nonspecificity, strife and conflict in the information. The results of these measures are useful for decision making as uncertainty-based information. We are progressing towards a novel approach to uncertainty analysis in a related sense by exploiting an existing technique in linear algebra as a means towards a measure of ignorance. A test metric is proposed to measure the similarity extent between two probability distributions, upper and lower. We show analytically and numerically how this metric performs on two normal distributions (a standard and sample) for illustration.