Re: genetic programming

To: tesla-at-pupman-dot-com
Subject: Re: genetic programming
From: "Jim Lux" <jimlux-at-jpl.nasa.gov> (by way of Terry Fritz <twftesla-at-uswest-dot-net>)
Date: Mon, 12 Jul 1999 07:45:53 -0600
Approved: twftesla-at-uswest-dot-net
Delivered-To: fixup-tesla-at-pupman-dot-com-at-fixme


> 	It was all very intersting, and it occured that this could be very 
> usefull in designing primary circuitry, or power supply.  I was wondering
if 
> any one had ever tried this or concidered this.  I doubt that any one
has, 
> unless they hapen to go to MIT or U of I.  Particularly because most of
the 
> problems were run of a "beawolf cluster computer" that operates at 5
gigahz, 
> which I doubt any one on the list has (price tag $15,000).  But hey, it's

> worth checking, hey, maybe someone's into programming.

Actually, that's a "Beowulf" personal super-computer (aka a Pile of PC's),
pioneered at Goddard Space Flight Center. We have one here at JPL (called
Hyglac) and CalTech has a huge one they are working on (I think it fits in
what is called the "dreadnought" class of Beowulfs).  Oddly, I am trying to
get a small one built right now.. Could you give a specific reference to
the automated circuit design/optimization application because I'm looking
for something novel to have it do (FEM stuff is pretty standard), and
optimizing TC design, or modelling spark propagation might be an
interesting application.

BTW, for those of you interested, you can really build a supercomputer in
your garage for under $10K, if you scrounge for the processors, etc. For
when you get tired of TC'ing... I would recommend keeping the TC and the
Beowulf away from each other though... Or, invest in lots of shielding.