MATSim Code

CMAES restricting secondary reduction to be positive

negative does not make theoretical sense, and quickly no incentive to
improve clustering, as no effect of changes on it.

non-linear fitness

penalize more when far of one of the coordiates

fix wedge sampling and use it

tolerances for global convergence

log info on found optimum

adapt config file to new algo

also log/dump fitness value

classes for demand generation for the santiago (cl) scenario

Update, restructure and cleaning of heterogeneity package

Updated tests