A suggested approach to computer arithmetic for designers of multi-valued logic processors

• Authors: D. E. Atkins
• Venue: MVL '78 Proceedings of the eighth international symposium on Multiple-valued logic
• Year: 1978

An approach to the topic of computer arithmetic is suggested which may have a particular conceptual, pedagodical, and practical appeal to the designer of multiple-valued logic processors. Computer arithmetic deals with the physical representation of finite sets of numbers and the design, analysis, and implementation of algorithms for mechanizing arithmetic operations on these sets. Finite number representation systems (FNRS) are specified by defining a set of symbols and a mapping from the elements of this symbol set to a subset of the real numbers. A formal definition of a FNRS provides a basis for a set of definitions which in turn provide the framework for the classification of a large set of number systems. Emphasis in this paper is on the following positive, fixed radix systems: unsigned, sign and magnitude, radix complement, and diminished radix complement. We offer an annotated listing of primitive digit vector algorithms for the four common number representation systems with an arbitrary, positive integer, fixed radix. These digit vector algorithms are ones which the designer of multi-valued logic arithmetic processors will need to implement to provide general arithmetic computation.