🚀 go-pugleaf

I had once read about an article that studied and compared several
alternatives suggested to Silicon in the future. I followed it up with
my own research on the 'net, and collated a lot of information. The
alternatives considered most in depth were Optical computing, DNA
computing, Molecular computing and Quantum computing.

A recent post reminded me of DNA computing... it was a pretty
interesting concept, and the time I read it, it was not eminently
feasible, but it certainly seemed possible. A search on "DNA
computing" on comp.ai.genetic resulted in some pessimistic views on
the subject, but I think it might be feasible with the right
infra-structure and for the right problems- and with the right
approach.

The current approach involves providing enough DNA material to create
enough combinations so that the right one would eventually emerge, and
be filtered out (using some watchamacallit sieve). One of the
problems, as a posting mentioned, was that as the problem complexity
increases, the amount of DNA and time required increases exponentially
to impossible amounts.

Now what if it were possible to apply Genetic Algorithms to DNA
computing? You only provide enough DNA to get a good starting
population, extract the most fit solutions and discard the unfit DNA
sequences for re-combination and reuse. Maybe the fitter solutions are
replicated to offer a higher chance of getting more optimal solutions.
Or you could use any other techniques to increase the probability of
getting an optimal solution within the limited DNA material available.

And so you have a parallel implementation of GAs in hardware .. or
rather in "wetware", so to speak :-) Not only does it potentially
solve the material amount problem with DNA computing, it upgrades (?)
the purely combinatorial model of solving problems with a more
heuristic AI-ish approach.

However beyond the most basic concepts I am absolutely blank about the
science of DNA and genetic engineering... I have no idea how you would
combine, extract the fittest, or recombine DNA, or even if it is
possible. Yet from all that I hear, scientists now can get a
chromosome to do everything except sit up and beg. So if not now, in
the near future my uneducated guess is that it should be do-able.

Could this be possible? DNA Computing and Genetic Algorithms... the
connection is obvious, I am sure somebody would have thought of this
before. Hardware and software have never been more tailored for each
other! I would certainly appreciate any comments, enlightenments on
this.

kundi