Features of the G-EMMA software
- Can operate with both linear least squares approximation of end-member fractions, and sampling of end-member fractions from a Dirichlet distribution. Standardization of concentrations to zero mean and standard deviation of one is advised when using lls approximation.
- Parallel processing: G-EMMA supports parallel processing on multi-core machines through use of the <a href="http://openmp.org">OpenMP</a> library. G-EMMA by default uses the maximum number of cores available on a machine, but this behaviour can be manipulated through the nthreads parameter in the ini-file.
- Adjustable re-use of iteration data between separate samples (speeding up calculation times), file storage permitting.
- Possibility to randomize both used end-members and solutes in end-member mixing calculations.
- Possibility of using either uniform or normal end-member concentration distributions.
- In lls approximation, flexible weighting of water balance closure.
- Flexibility in defining fuzzy likelihood measures.