Floating-point number LISP

Kyoji Umemura
1991 Software, Practice & Experience  
A floating-point-number-oriented LISP has been developed. Though it performs type checking on every numeric operation, it runs as fast as Fortran for simple differential-equation problems. This paper describes the implementation, provides some measurements of its efficiency, and discusses the feasibility of this type of implementation. KEY WORDS Floating-point arithmetic LISP Symbolic computation DATA REPRESENTATION In Real Number Lisp, all numbers are represented in double-precision format.
more » ... h the integer 1 and the floating point number 1·0 are represented in the same way. Rounding errors have no effect on fixed-length integer operations because add, subtract and multiply operations in double-precision numbers are accurate within a 56-bit range on DEC VAX computers. All other data are represented as invalid floating-point numbers. Invalid numbers
doi:10.1002/spe.4380211003 fatcat:enf223amr5enrbfcqtevrfqsce