The Uncertain Number: An Abstract Data Type for Measurements

摘要

A measurement is a process for estimating the value of a quantity. Because there is always some uncertainty in how close an estimate is to what you are really trying to measure, the accuracy is never perfect. Consequently, the level of uncertainty that is acceptable depends on why a measurement is needed-a seven-digit voltmeter is not required to discover if your car battery is flat. Uncertainty information is used to determine the range of values consistent with a measurement result. Whenever one measurement is compared (or combined) with another, uncertainty is needed to properly interpret the results. These days, with increasing demands for objective quality standards, both users of measurement instruments and instrument manufacturers are thinking harder about how to make use of uncertainty information. This trend will grow.