SAM: Stochastic Arithmetic in Multiprecision

The SAM library implements discrete stochastic arithmetic in arbitrary precision. It enables rounding error estimation in arbitrary precision programs. The control of accuracy is performed on stochastic variables, the stochastic types being provided by SAM.
It requires the MPFR library, which itself requires the GMP library.

Two different versions of SAM are available.

Uniform precision version
All stochastic variables have the same precision that can be changed during the execution.
Souce code, documentation and example programs can be downloaded.
This version of SAM has been introduced in the following article:
S. Graillat, F. Jézéquel, S. Wang, Y. Zhu, Stochastic Arithmetic in Multiprecision, Mathematics in Computer Science, 5(4), pages 359-375, 2011.
DOI link, BibTeX reference

Mixed precision version
Each stochastic variable is declared with its own precision.
Rounding errors can be estimated in codes having variables with different mantissa lengths.
Souce code, documentation and example programs can be downloaded.