at moment have genetic algorithm (ga) runs quite while, , thought distribute using fabric because theoretically fits nicely microservice. first try @ fabric.
how should ? should have stateful service runs , aggregates other actors tasks ? it's kinda similar project: https://github.com/azure-samples/service-fabric-dotnet-data-streaming-websockets
i'm not sure how go , there not documented on subjects. ga extensive , our goal distribute it's calculations.
i implemented basic genetic algorithm app service fabric app building exercise. not sure if approach best way things scenario, can describe did.
my app consisted of actors, both stateful , stateless. had processor stateful actor provided management tasks , drove algorithm. because stateful, maintained history of genetic state across each of generations produced. had fitnessevaltask stateless actor. task responsible evaluating fitness of entity. input gene representation , output fitness value. idea you'd spinning instances of actor @ high rate , they'd distributed appropriately. processor app, being responsible driving algorithm, create necessary instances of fitnessevaltask actors , provide input , have them report fitness values , necessary processing afterwards. client process, simple console app, communicate processor actor initiate algorithm , perform necessary management tasks.
Comments
Post a Comment