Abstract

Presently, the IEEE 754 Float is the most commonly used standard to represent floating point values, which is capable of representing fairly extensive range of numbers. Recently, a new number format has been coined among the research community, termed Unum. The Unum format is in the evolutionary phase and is being standardized for efficient and accurate representations keeping the number of required bits quite limited at the same time. Many researchers consider the Posits feature of Unum as a viable alternate to the IEEE Floats as it provides number representations in a broader range and higher accuracy as compared to IEEE Floats for a given number of bits. This paper presents R-SoftPosit, an R-based software implementation of Posits to facilitate comparison between Posits and Floats. The article describes the steps followed for the development of the package, lists of classes and functions created, concerning documentation, and reproducible examples. This R-based package aims to provide a convenient tool to explore and test various features of Posits.