ABSTRACT
Numerical error is composed of two components, precision and accuracy. Precision and accuracy are two related terms that are often used interchangeably. Accuracy measures how close an estimate is to the true value. However, even with a small relative error R, it is possible for the effect to cause problems. Being in Newark when a job interview is in New York is catastrophic. Internally, R has a variety of ways to store numbers. Most numbers in R are stored as floating point numbers. The chapter provides a reference table for the binary-decimal-hexadecimal digits from 0 to 15. In addition to accuracy and precision in our source numbers, the idiosyncrasies of the floating point formats can create new errors. While round-off error happens in or near the last place of numbers, round-off error can appear in other places. Numerical stability is a desirable property to have in an algorithm, but it is generally found in an algorithm-problem pairing.
