Probabilistic concurrent programming using multiset transactions

The GAMMA parallel programming model, based on the multiset data structure and a succession of chemical reactions consuming the elements of the multiset and producing new elements according to specific rules, is a very useful new paradigm. This paper extends GAMMA model to its probabilistic version called P-GAMMA model. The P-GAMMA model can realise probabilistic, classifier, bucket-brigade, and genetic algorithms. We also describe how concurrency in GAMMA paradigm can be supported using transaction processing facilities provided with multilevel atomicity, consistency, isolation and durability features