Monday, October 25, 2010


Today a quick post on agent programming, using jade. The importance of a small wait between threads, as you might know when you program an agent you can have parallel behaviours running at any given time, if you multiply this by the number of behaviour times the number of agents this means a your processor will only have time for the agent programs, in fact the add of a small time delay (less than 200 milliseconds)can actually improve the performance. Plus it has the added value of you being able to use the computer while running simulations that include several agents :P

Taking a breather is always good policy, and this applies to your code ;)

Another interesting thing I came across is the limitation of the DF agent in returning agents that do a service, it is limited to 100 answers. Well for most applications I guess it is not a problem, but when using the basis of swarm and wanting to trigger a specific type of agent, 100 agents doesn't seem that much. Anyway I had to make an agent that creates and launches to maintain the info, and pass it on request. Honestly after looking around online this was the best option.

Well JADE is one of my posts with more feedback, and since I am currently facing some problems there, well some posts on this might happen :P.

Plus it is always good to relate a doWait(200) whatever you are doing ;)

P.S.-Support the sailing dream please click on the ads ;)

No comments: