On Jan 12, 2016 18:40, "Damian Vicino"
Hi, I’m interested in developing basic support of “real numbers” for
computable calculus applications.
In computable calculus, reals have infinite precision ( certain
The usual approach is representing numbers as functions producing digits. An arithmetic operation produces a new function using the functions in
restrictions may apply). the operands and some simple expression manipulations.
When a number needs to be evaluated (e.g., for comparisons where expression rules are unknown), enough digits for reaching an answer are generated using the functions. For example, 3.14 < pi requires generating 3 digits to answer “true”.
The algorithms for arithmetic operations and evaluation are well known, and multiple implementations were developed (mostly in the 90s). However, I couldn’t find an open source implementation and had to come up with my own for working in a project during the last 2 years. My preferred reference for the required algorithms is the book: Aberth, Oliver. Computable Calculus. Academic Press, 2001.
In particular, my initial scope would be the 4 arithmetic operations (+, -, *, /), and the comparison operators (<, =) for reals.
Someone is interested in such a library in boost?
Yes! I really want this.