Registered: Mar 2007
Posts: 1

Hello,
I´m from German and new in this forum.

After a long time of work I have developed an computer algorithm that is able to classify a given onedimensional CA with symmetric set of rules,i.e.
f([leftneighbours,cell,rightneighbours])=
f([rightneighbours,cell,leftneighbours]) ,
in one of the four Wolfram-automataclasses I, II, III or IV. Totalistic CA are a special case of symmetric CA!
I call this the ´Algorithmic Classification´ versus the ´Subjective Classification´ a human beeing gives when he studies the Space-Time-Development of the CA.

I needed several years to develop the Classification Algorithm und now it works with a reliability of about 99,7%.
The algorithm does not rely on the results of Andy Wünsche!!

With this Classification Algorithm you can do several things:
For example you can automate the search for the fascinating Class IV - automata.
I did it.
Or you can investigate the distribution of the four automata classes in the space of all CA with a given neighborhood structure and number of cell states.
I did it, too.
Or you can explore if the Class IV - automata form accumulations in the space of all CA with a given neighborhood structure and number of cell states.
I did it, too.

Is anyone interested in my work ?

I´ll append a demo program which shows the ability of the algorithm to classify a given symmetric 1D CA.
The demo program is written in C# for Windows OS and requires the .NET Framework from Microsoft.

Further: You are free to download a 48 page article from my homepage: www.ewetel.net/~martin.sollert/ , which contains all the results of my work.
Unfortunately this paper is written in German because of my bad English.

Regards
Martin Sollert

Attachment: algorithmic_classification_of_ca.zip
This has been downloaded 1309 time(s).

__________________
MS

Report this post to a moderator | IP: Logged