Hi All, I have implemented memory funtion for the simulation. Now the user can mark any number of crime spots on the map. Each crime spot is associated with reward, risk and no. of times the agent has passed by the vicinity of that crime spot. It is represented as <reward, risk, number of times> in the label for the spot on the map.
You can have a lok at it here: http://gw76.iu.xsede.org/criminfo/index.html I have saved a sample simulation by the name - memtest. You can click on "load" and enter: memtest to load that sample simulation. Or you can run a new simulation yourself. The next step i guess would be determining whether the agent breaks into a crime spot. This would be a factor of reward, risk and the number of times the spot has been visited. We could assign different weights to the above factors. Please let me the model I should use asap. Thanks, Nadeem. -- Sent from my iPad
