Page 1 of 1

PostPosted: Wed Apr 04, 2001 3:35 pm
by wossname
Hi, I have just downloaded version 4 of PocketC, oh how sexy it is!

I love the new @[] stuff, way faster than the methods i was forced to use earlier (concatenation).

Anyway, i am an Artificial Intelligence fanatic, and i'm looking for a new experiment to write a simulation for.

The last one I did was from the book "The Blind Watchmaker" (Richard Dawkins), about using a genetic algorithm to match a string (in this case a Shakespearian quote) starting with a random string and "breeding" from it!

Now though, I'm lost for ideas for new and interesting projects. Any ideas from you programming guru guys?

<img src=icon_smile_wink.gif border=0 align=middle>

Wossname :)

PostPosted: Thu Apr 05, 2001 8:37 am
by Thad
Okay, as your looking for a challenge, and are familiar with GA's how about writting a GA to solve the countdown numbers game.

Once you've got it working, compare it to the one I've got on my homepage, and see which is quicker!

Okay, so the countdown numbers game works as follows. You have six randomly selected numbers, and a randomly selected "target" number, and using each of these numbers no more than once find a way of getting as close to the "target" number as you can using integer maths .

For example
Source numbers: 1 5 10 250 500 7
Target number: 107

My Solution:
500 / 250 = 2
5 * 10 = 50
50 * 2 = 100
100 plus 7 = 107

My GA's solution (in 276 generations):
500 / 250 = 2
1 * 2 = 2
5 plus 10 = 15
15 * 7 = 105
105 plus 2 = 107


Edited by - thad on 04/05/2001 04:14:02

PostPosted: Thu Apr 05, 2001 1:42 pm
by wossname
Right I'll get right on it.

We'll see what Carol Vorderman has to say about it! :)

Wossname :)

PostPosted: Thu Apr 05, 2001 5:55 pm
by Glasswalker
The GA idea is a great one... I will fiddle with it and post back when I have a working chunk of code :) this should be fun...

By the way if you are interested in artificial intelligence I am working on a fairly complex Neural Network simulation for the palm myself, it should be pretty good. Send me an email if you are interested and I will send over the source and the prc file, I would love some help on it (as if you read some of my other posts I am having a problem with it automatically running the prefs program every time I use the Neural Net... wierd... anyhow...) Hope to hear from you...