On Mar 28, 2007, at 4:22 AM, Dave Wallace wrote:
> Jim Armstrong wrote:
>> But don't multiple-platform code modules often have functions that
>> are "switched on" or "switched off" depending upon the specific
>> hardware configuration of the host (thinking specifically about
>> military aircraft and weapons configurations)? JimA
>
> That is a very common way to deal with cross platform differences
> in programs that run on different operating systems or different
> hardware. Different features for different versions of a product
> may also be supported in the same way.
>
> Too much code "switched on" or off can become a maintainance
> nightmare and then the platform specific code may be isolated into
> specific platform dependent modules that get used depending upon
> the platform/feature set. Not sure if their is anything analogous
> to this process in the natural world.
This is too simplistic. The best computer science analogy I can think
of other than genetic algorithms which I will deal with later is
simulated annealing. When there is optimization on the computer you
worry about the local minima problem when you do greedy optimization.
Within your cooling schedule when the temperature is high you explore
more options that are neutral and very early on bad. The temperature
is slowly cooled and later on neutral options are explored less
frequently. What this algorithm does is allow for better optimization
without brute-force exploration of the entire design space. Note
something very significant here: note that design and random are not
mutually exclusive. As a designer, the random number generator is my
friend!
There was a recent paper on the observed evolution of viruses -- I'll
find the reference later -- that showed that observed changes went
in two phases. The first phase explore many neutral changes while the
second phase culled down the changes via natural selection. Genetic
algorithms do similar things. When you compare both genetic
algorithms and simulated annealing against a priori greedy design
algorithms you often get the better solutions. What you DONT get is
the global optimum. Both classes of algorithms are known as
heuristics because the problem spaces are often on the order of NP-
complete. To my engineers' eye, evolution has the hallmarks of design
and Dembski's tests are all wet because he is looking for greedy
optimization and that's not how intelligent designers do our job.
To unsubscribe, send a message to majordomo@calvin.edu with
"unsubscribe asa" (no quotes) as the body of the message.
Received on Wed Mar 28 09:05:13 2007
This archive was generated by hypermail 2.1.8 : Wed Mar 28 2007 - 09:05:13 EDT