This is two small implementations of the language described in
D. Koller, D. McAllester, and A. Pfeffer. Effective Bayesian
Inference for Stochastic Programs. In Proceedings of the
Fourteenth National Conference on Artificial Intelligence (AAAI-97),
pages 740-747, Providence, Rhode Island, August 1997.
The implementations are in Java and Scheme. Both have bugs. The Scheme version is currently buggier, but is also much more concise.
There is a description of them available. [PostScript, 76k]
The source tarball is
distributed under the
GNU Lesser General Public License.
I've also implemented the probability
monad in Scheme.
There's now a very tiny OCaml implementation of this. It only has integers as a base type, but it's very small, so (hopefully) it's easier to understand what it's doing. Here is the output of running it on some tests.