selection program

Glenn.Morton@ORYX.COM
Wed 31 May 1995 11:44 CT

I am releasing a new program which shows the power of random mutation coupled
with selection. It is based on the same mathematical algorithm of the two
previous programs. The program really addresses the often cited objection to
evolution that there hasn't been enough time or that random mutation simply
couldn't find the one useful permutation of a given set of amino acids.
In this program, as opposed to the previous two, the 4 points are real
numbers. This means that there are 10^312 different possible combinations of
the "genome" for my screen critters. If only one permutation will produce a
given form, then my program will never find it. What you will find is that
every time you run it you select a different genome which produces a similar
morphology. Thus the argument that one and only one permutation will perform
a given task is erroneous.
Each successively selected genome differes in one of the 8 numbers of
the genome from the previously selelcted form.Thus there is a step by step
mutational lineage from the single dot form and the finalselected
form. I do not allow more than one mutation per step.
I would suggest that you run the vertical or horizontal line several times
with the same parameters first. This will give you a feel for how common
certain solutions really are in the phase space of the 'genome'.

The programs can be access from ftp.oryx.com. Type cd /pub. Type "get
select.exe" "get 8514ai.bgi" and "get egavga.bgi" Put
them all in the same directory and type "select". There are more extensive
directions in the program.The other two programs are evolve.exe and
cambexpl.exe.

By the way, Brian Harper, are your there? I am still interested in any
comments you might have on the last two programs as well as this one.

glenn