Computing thousands or millions of digits of PI with arithmetic-geometric means This is a proof of correctness for two algorithms to compute PI to high precision using arithmetic-geometric means. A first file contains the calculus-based proofs for an abstract view of the algorithm, where all numbers are real numbers. A second file describes how to approximate all computations using large integers. Other files describe the second algorithm which is close to the one used in mpfr, for instance. The whole development can be used to produce mathematically proved and formally verified approximations of PI.
opam install coq-pi-agm.1.2.1