GMLE is a distributed generic maximum likelihood evaluator implemented in the MPI/C and SALSA/Java programming languages. Users can define a function to be evaluated in a distributed fashion over distributed data. Different search methods can then be used to find the maximum likelihood. Currently, gradient descent, conjugate gradient descent, simplex and genetic search have been implemented for GMLE. |
GMLE is now version 0.2. Release Notes: The GeneticSearchProbabilistic search has been added for SALSA. This performs an asynchronous genetic search. The SALSA Physics package has been updated to use GeneticSearchProbabilistic and GradientDescent. The SALSA Astronomy package has been updated to use GeneticSearchProbabilistic and Simplex. |
Previous releases can be accessed here. |
Department of Computer Science |
Rensselaer Polytechnic Institute |