The following script is still unfinished. It describes an agent of type which follows the paradigm of an emotional LCS. See preceding chapters for more details.
The script is divided into the following sections:
Then you can see protocols of some runs.
Comment: I have written the algorithms for the LCS-agents as scilab scripts. This is at a first glance quite 'handsome' because the relationship between the theoretical text and the scilab functions is very close. For theoretical works is this a strong argument. Furthermore one can work with the same development environment under windows7 as well as under linux (ubuntu). Echanging files is no problem. But from a more 'practical' point of view it becomes more and more clear that the scilab development environment is a bit 'low graded' and can be very slow even on a 4-CPU high end PC if one has to change code -which is the most often event while coding- . Using screen output while running a script to improves understandability but slows the machine extremely down (whereby the 'logging' option with the diary()-command is nevertheless very convenient). There is an option to compile scilab; I didn't yet test this option. But even if this will work there is no simple way to improve the scripts in a way that allows a more object oriented development which is important for more serious work in the future.