The simulated environment contains the space information in the grid and the state information of all registered agents. To 'use' these informations appropriately we need another mapping from the actual state of an environment to the next one, saying . Such a mapping makes 'sense' if 'between' two such mappings the agents will be 'informed' about new inputs and they can compute new responses. Thus we need a repeated concatenation of the kind . A simulation is then such a finite (or infinite) repetition:
(4.65) | |||
(4.66) | |||
(4.67) |
To sum up: the interface function maps the state of the simulated environment onto the agents and vice versa, and the environment manager maps the actual state of a simulated environment into the next state. These two functions are constantly changing and thereby thexy are gnerating a sequence of environmental states which represents a possible history of the environment including the agents.
The exact details of all these functions will be defined as scilab functions.
Gerd Doeben-Henisch 2012-03-31