Hi Alberto, You should be able to override the default partitioner class in pseudo distributed and the fully distributed mode. I used the same technique for the paper "Scaling Genetic Algorithms using MapReduce" [ http://verma7.com/wp/wp-content/uploads/2009/10/ISDA09_MR_GA.pdf] that you might find useful.
-Abhishek. On Fri, Jan 8, 2010 at 5:24 PM, Alberto Luengo Cabanillas <[email protected] > wrote: > Hi everyone! First of all, happy new year to all community!. I´m actually > developing an Map-Reduce approach to genetic algorithms, and I was > questioning the use and effectivity of Partitioners in the Map-Reduce flow. > I´ve been working in pseudo-distributed mode and didn´t find any example of > usage of Partitioner classes. Does this mean that using another Partitioner > -not the default one- have sense only on a Fully Distributed environment? > Which Partitioner should I implement with my GA approach? > Thanks a lot in advance! > > -- > Alberto >
