Re: beastly_maths - a genetic algorithm

On Oct 30, 6:14 pm, Martin Andersen <d...@xxxxxxx> wrote:
Sapient Fridge wrote:
In message
<53838670-8e58-4654-bd07-bc17e49ab...@xxxxxxxxxxxxxxxxxxxxxxxxxxxx>,
spintronic <spintro...@xxxxxxxxxxx> writes
On 30 Oct, 21:00, Sapient Fridge <use_reply_addr...@xxxxxxxxxxxxxx>
wrote:
In message
<18f6db02-f0e9-4161-b5c4-9a2bb2d28...@xxxxxxxxxxxxxxxxxxxxxxxxxxx>,
spintronic <spintro...@xxxxxxxxxxx> writes

On 30 Oct, 17:35, Sapient Fridge <use_reply_addr...@xxxxxxxxxxxxxx>
wrote:
If anyone is interested I've uploaded a new version of my genetic
algorithm to which searches for expressions which evaluate to a
target
value e.g.

5+5+3-5^3+8^1*6*4^5+1^5^5-4*5-2-1^4-5*8+5+7 = 48990

It's written in perl and has simple instructions at the top of the
file
explaining how to run it.    Note that the file name has changed from
the previous version I posted (though the old link will still work):

http://www.spamsights.org/misc/beastly_maths.txt

The new version (Aardvark.Buffalo) has a slightly better fitness
scoring
algorithm along with improved comments and output.

For background as to why this was written read this:

http://groups.google.com/group/talk.origins/msg/29dc74b9af18ba50
Somehow, I feel like Sapient Fridge tapped me on the shoulder, and ran
away with the cookie jar.

Chuckle, I admit that you do have some cause for complaint there.

Your idea of a using a genetic algorithm to find expressions to reach a
target number was a good one, and I can't resist an interesting
programming challenge.

Do you want me to credit you with the idea in the code comments?

do realise that RSA can be broken with this algorithm?

I seriously doubt it, but feel welcome to prove me wrong.  You will
become famous very rapidly in the world of cryptography if you do.

Genetic algorithms only work when you can score your population in some
way.  Breaking RSA would involve working out the prime factors for
*very* large numbers.  It's impossible to assign fitness scores to
something like that as you either have the factors or you don't.  A two
value scoring system (true/false) will not work for genetic algorithms.

Your honesty and talent is wasted on this guy.- Hide quoted text -

- Show quoted text -

You call Pseudo-code talent?

[chuckle]



.

Relevant Pages

  • Re: Whats wrong with this proof that P = NP?
    ... Might it not be possible to somehow use pointers to make ... return a valid proof that W is false, leading to a contradiction. ... that no polytime algorithm like B exists, ...
    (comp.theory)
  • Re: How to compute a regex which is the difference between two regexes?
    ... The the one-letter strings b, ... Your algorithm which works directly on "regular expressions" is ...
    (comp.compilers)
  • Re: Float comparison
    ... For example, when using IEEE 754 arithmetic, you perform ... apply a simple rounding algorithm that produces a real number that has ... a representation, and that is the representation of the result. ... mathematical expressions expressible in C yield rational numbers. ...
    (comp.lang.c)
  • Re: NOT SURE ABOUT THE CONSTANT PROBLEM
    ... Let B be the class of expressions generated by ... The sine, exponential, and absolute value functions. ... the predicate "A=0" is recursively undecidable. ... there is no algorithm which will give the right ...
    (sci.math)
  • Re: When in doubt, drown em in lawyers
    ... You're using the word algorithm in a manner different from any I've ... expressions of theories in science and math come in algorithm form. ... methods for estimating celestial distance using parallax. ... test of invention. ...
    (comp.sys.mac.system)