US Patent 7136891 Arithmetic and relational operations

In a method of carrying out arithmetic and relational operations on decimal numerals of arbitrary size, the operations produce exact results if the decimal numerals involved are terminating or repeating. The result of arithmetic operations is flagged to indicate whether the result is a terminating or a repeating numeral. To maintain computational accuracy, numbers are converted to rational fractions, whenever necessary and computations are performed using such fractions. The sign, decimal point, and digits comprising a number are treated as individual character symbols. All arithmetical and relational operations are performed on character strings, rather than on binary maps of the relevant numbers.