Something Similar 22.7.2004

Get Firefox

Scary Ballon Photoshops!

Seriously funny.

Bowser Balloon

posted at 21:57:20    #    neural activity []    trackback []


Going to have to look into this one. Might be handy for those "working on the go" times. Would be nicer to have a laptop on hand instead of a bootdisk (even if the bootdisk is fuckin' A) but I can't invest in one.. yet

A scientificly remixed Knoppix with clustering capabilities (through OpenMosix). What more could I want?

posted at 19:08:48    #    neural activity []    trackback []

Genetic Alogrithms and Me

I'm looking into doing research with Prof. Xi this next semester under the heading of genetic algorithms (related to evolutionary computation, evolutionary strategies, genetic programming, evolutionary programming and so on). It's being used in everything from art to artificial intelligence. Xi appears to be interested by neural networks but says that the field is "wide open" to me. He handed me a book called An Introduction to Genetic Algorithms by Melanie Mitchell and told me to copy the first and last chapter of it.

"If you read the first 30 pages, you will understand 70% of the field," he told me.

Do I really need anymore motivation than that? The application of GA (or EC or ES or EP or GP) in the world of physics is not a very well-studied issue or, at least, that is what my seach for information has led me to believe so far. I'm really hoping to find some interesting uses of GAs to solve physical problems and not just engineering issues, like most efficient wing shape (though that is an interesting application).

Anyone seen anything? Oh, I should note that there are lots of libraries for working with genetic algorithms. Scientific Python SciPy seems especially nice, if only because I really enjoy writing Python code. Unfortunately, it depends on ATLAS and since I'm a developer at Lunar Linux, I'm trying to build ATLAS from source so that I can write up a BUILD to add to our moonbase. As of yet, ATLAS is kicking my ass.

I think I might just cheat when I come home from my study time and install one of the binaries at the SciPy site. Ah well. The things I must sacrifice for the sake of science.

Update: Scientific Python and SciPy are not the same. Yeesh.

posted at 17:37:04    #    neural activity []    trackback []


I hate ATLAS' method of building from source code. It's the most cumbersome, pain-in-the-ass system I have yet to meet. Instead of being able to pass what kind of optimizations I want through a normal ./configure; make, I have to use this thing they call xconfig that requires you to interact with it. Not only that, but it doesn't automatically add in LAPACK support if you have the libraries. You have to fucking run ar and cp by hand to get the "complete" ATLAS install but depending on what architecture (Athlon/Pentium/etc) you compile the library for, it goes into a different directory. Argh! What is this? 1976?

I appreciate the amount of work that went into this, but couldn't they have made it just a bit easier for those who need/want to use source based distributions?

posted at 00:01:04    #    neural activity []    trackback []
July 2004
    1 2 3 4
5 6 7 8 91011

Tearing at the veil between art and science with every poorly written post.

XML-Image Letterimage

© 2004, Jeff Hodges