Kind of strange, I know, but sometimes you just want to make sure that your best individual really is something special. An easy way to do this is to compare it against the worst individual.
… relative to the gene, not absolute like it currently is. Default mean = 1.0 Default std = 0.0333 This means that the values will be multiplied by 1.0 with a std of 0.0333. In other words, 68% of the time, the multiplier will be within +/- 0.999 (10%) of the gene value.
…anGradient The following four mutators are new: G1DListMutatorRealGaussianGradient G1DListMutatorIntegerGaussianGradient G2DListMutatorRealGaussianGradient G2DListMutatorIntegerGaussianGradient The main difference between Gaussian and GaussianGradient is that GaussianGradient uses a multiplicative modification rather than an additive. GaussianGradient's mu and sigma are absolute, not relative. So if the default values of mu=2 and std=10 (why not mu=0?) are used, the random gaussian number is a flat number around 2. If we're working on a huge range, like say 0-100000, this is a very small drift. GaussianGradient uses mu=1.0 and std=0.1 to generate a number around 1.0. This is then *multiplied* by the gene to provide subtle drift regardless of how large the range is. 2 new constants added, Mu and Sigma for the GaussianGradient routines.
…allow multi processing support
… related to the issue #5.
Removal of the MpiMigration.py Refactoring of the MigrationScheme Refactoring of the MPIMigration
…ng abs in append
…This is to avoid that individual who were lucky once will bias any subsequent evolutionary step
…osed. Closes #1. Signed-off-by: Christian S. Perone <email@example.com>
Added Eric Floehr as Contributor.
…s patch closes the Ticket #98.
…ernative, they were reimplemented in a more flexible way.
See ticket #85 for more information.