Re: [asa] Of motes and beams

The computer simulations I've seen tend to use evolutionary principles to
solve optimization problems, rather than simulation of molecules. They use
natural selection based on fitness, and mutation (and often crossover), and
are based on having a simple "genome", usually a string of 0's and 1's
analogous to the four nucleotides in DNA. Each genome in the population is
decoded to become a candidate solution (for example by mapping bit strings
to numbers), and the fitness of the solution is then evaluated. Then
natural selection is used to select genomes for the next generation, and
then mutation and crossover are applied.

Despite the exciting nature of the idea (a computer algorithm drawing
inspiration from biology), when applied to real problems that you might want
to solve, they are somewhat disappointing, and a large amount of design work
is required to determine an appropriate decoding from genome to candidate
solution. They also perform poorly on high dimensional problems, due to the
so-called "curse of dimension" - the difficulty of finding the right
direction in a high dimensional search space. However, I now think this
limitation is due to the fact that mutations are assumed equiprobable
anywhere on the genome. According to Caporale's book this is definitely not
the case with real DNA - if mutations can be focussed on a few small "hot
spots", then maybe the curse of dimension is overcome.

I think the early chapters of Caporale's book are the most illuminating -
the later ones are a little heavy going for a non-biologist like myself.

When you say simple atomes "evolve" in time, you have to be careful what you
mean by "evolve". In the broad sense of the word, it just means "change
through time", and so any system obeying a differential equation in time
could be said to evolve. But I was meaning in the Darwinian sense -
replication, selection, mutation.

A thought occurs about your previous post, when you said that the system
adapted within the lifetime of the creature, so it didn't qualify for
evolution. However, in terms of the lifetime of the cells involved, of
course many lifetimes come and go during the fight against an infection -
presumably (I'm not an expert on this), the unsuccessful mutations that
don't bind on to the antigens effectively aren't selected for replication
and die off.

Regards,
Iain

On 7/18/06, Alexanian, Moorad <alexanian@uncw.edu> wrote:
>
> Thanks for the explanation. I am raising questions not
> questioning. Surely, one can have a system with simple atoms that "evolve"
> into complicated molecules as the system evolves in time from some initial
> conditions, which can be accomplished in short time scales. Are the computer
> simulations of evolution you mention of that type? I will have our library
> order "Darwin in the Genome" by Lynn Caporale.
>
>
>
> Moorad
>
> ________________________________
>
> From: Iain Strachan [mailto:igd.strachan@gmail.com]
> Sent: Tue 7/18/2006 9:52 AM
> To: Alexanian, Moorad
> Cc: Asa
> Subject: Re: [asa] Of motes and beams
>
>
>
>
> On 7/18/06, Alexanian, Moorad <alexanian@uncw.edu> wrote:
>
> The example is that humans have built-in abilities that can be
> used in new situations without meaning that they have evolved. It is a
> question of time scales. If a system can readjust itself to new
> circumstances in short periods of time, that cannot be called evolution.
>
>
>
> Can you explain why you say this? It's evolution if the principal
> mechanism of adjustment is mutation + natural selection. The timescale is
> irrelevant. You can make evolution happen in a computer simulation that runs
> in a fraction of a second, and it's still an evolutionary
> process. Admittedly the valuable applications of genetic algorithms are few
> and far between, but they do exist. I agree it doesn't _prove_ humans have
> evolved, but that wasn't what I was trying to prove - all I was saying is
> that you can't per se call evolution a bad thing, or say it has a "bad track
> record". If it didn't go on in your immune system you'd be dead very
> quickly, as tragically, victims of AIDS find out.
>
> For what it's worth, I was extremely sceptical about evolution of
> different species for a long time, based on my own experience with genetic
> algorithms, and how difficult it is to get them to solve any worthwhile
> problems. However, I'm reading a very helpful book at the moment that goes
> a long way towards answering the questions I had, namely "Darwin in the
> Genome" by Lynn Caporale. It is also refreshing that Caporale doesn't use
> her science as an excuse to bash religion (and explicitly says so). One of
> the key limitations of genetic algorithms appears to be that mutation is
> completely random (ie equally likely to strike at any point). However, a
> simple observation from Caporale's book (that the four bases A,C,G and T are
> slightly different), leads to the fact that mutation isn't equally likely
> everywhere - in fact there are localised hot spots where lots of mutations
> occur and areas where very few mutations occur. This is precisely what is
> going on in antibody evolution - the bit that binds on to the antigen
> mutates rapidly, and the fixed bit of the antibody (that sends some sort of
> signal for the rest of the sytem to act, as I recall) doesn't mutate. But
> the mutation process happens exactly as it does over generations - by
> inexact copying when a DNA molecule replicates itself. There is essentially
> no difference here between this type of replication, and when DNA replicates
> to produce sperm or egg cells, in that inexact copying gives rise to
> mutations.
>
> Iain
>
>
>
>
> Of course, the time scale of readjustment is of the order of the
> lifetime of the system.
>
>
>
> Moorad
>
>
>
>
> ________________________________
>
>
> From: Iain Strachan [mailto:igd.strachan@gmail.com]
> Sent: Tuesday, July 18, 2006 3:44 AM
> To: Alexanian, Moorad
> Cc: Vernon Jenkins; George Murphy; Don Nield; asa@calvin.edu
>
>
>
> Subject: Re: [asa] Of motes and beams
>
>
>
>
>
>
>
> On 7/18/06, Alexanian, Moorad <alexanian@uncw.edu> wrote:
>
> I am not sure I understand. When a human encounters a new
> mathematical problem and is able to develop a solution to it with the prior
> information he/she has about mathematics, can we say that the individual
> evolved or just that the human brain has the ability to "figure out" new
> situations with already existing elements?
>
>
>
> I wouldn't say that the solving of a mathematical problem, a
> sequence of logical steps based on some informed guesses, bears any
> resemblance to the evolutionary processes by which the immune system
> works. Unless you solve mathematical problems as in a cartoon I once saw
> depicting Einstein's blackboard, which had three lines on it: E =
> ma^2 (crossed out) E = mb^2 (crossed out) E = mc^2 (tick). No-one
> solves maths problems like that so to bring it up is an irrelevancy.
>
> Iain
>
>
>
>
> --
> -----------
> After the game, the King and the pawn go back in the same box.
>
> - Italian Proverb
> -----------
>

-- 
-----------
After the game, the King and the pawn go back in the same box.
- Italian Proverb
-----------
To unsubscribe, send a message to majordomo@calvin.edu with
"unsubscribe asa" (no quotes) as the body of the message.