Interesting Math and Visual Theory

[FrostyBoy]

I had this theory a few years ago, and posted it on Apolyton, but no one really cared. Anyway, times have changed and time to post it again.

The theory goes, if you have a computer that automatically generates random coloured pixels (including black, white and shades of grey) in a given size box and never repeat the same image produced, you will eventually get an image of everyone who ever lived and has yet to live, of every animal, every rock, every grain of sand - all at every possible angle and every possible lighting difference - and of course a LOT of images of just rubbishy noise patterns. (A nightmare if someone had to sort it all out, but I guess a computer could do the sorting out quite easily if it were to be programmed right).

So here's the math, someone do the workings for me, coz I suck at math, I am only good at coming up with ****.

100x100 square (10,000 pixels)
256 colours (don't really need 16bit colour to identify an image)
10,000^256 right? I think it's = 1024 + how many 0's?


Question, how many images would be produced, if each one were to be saved to disk?

Advanced questions:
How long would it take a supercomputer to accomplish?
How much disk space would be required?
Would it be hard to programme something like this, and would any of you be able to do it?

[Blake]

That's easy.

#pictures = 1 followed by 1024 zeros

This is 1 followed by 944 zeros times as many atoms in the universe.

So the answers to your other questions:
1) Longer than the lifetime of the universe. You'd run out of energy before even getting started.
2) If you converted all the mass in the universe into hard disks, you aren't even getting close.
3) Hell yeah, it'd be easy.

[Ari Rahikkala]

You're not thinking big enough.

[Dry]

Re: Interesting Math and Visual Theory

Originally posted by FrostyBoy
...never repeat the same image produced

That means your random generator is not purely random and has some intelligence.
You can then program it to be a little bit more intelligent by not generating rotated images, dividing the number of images by 4 (0°, 90°, 180°, 270°).
You can win another 4 factor by not generating symetric images (along vertical median, along horizontal median).
Ok, that's only 10^1024 / 16 = 6.25 * 10^1023, but it's a start.


EDIT: Oops, I s*** at math today, 10^1024 / 16 is 6.25 * 10^1022

[b etor]

Originally posted by Ari Rahikkala
You're not thinking big enough.

he can't even spell his own name right in his email address. DON'T TRUST HIM HE'S LYING!!!!

[Kuciwalker]

Re: Interesting Math and Visual Theory

Originally posted by FrostyBoy
a bunch of obvious ****

You failed algebra in grade school, didn't you.

[Kuciwalker]

Also, why the hell would you use an RNG to generate this. Just generate the images in lexicographical order.

[Lul Thyme]

Re: Re: Interesting Math and Visual Theory

Originally posted by Dry

That means your random generator is not purely random and has some intelligence.
You can then program it to be a little bit more intelligent by not generating rotated images, dividing the number of images by 4 (0°, 90°, 180°, 270°).
You can win another 4 factor by not generating symetric images (along vertical median, along horizontal median).
Ok, that's only 10^1024 / 16 = 6.25 * 10^1023, but it's a start.


EDIT: Oops, I s*** at math today, 10^1024 / 16 is 6.25 * 10^1022

The factor saved with what you described is only 8 , not 16 (a horizontal reflection followed by a vertical reflection is equivalent to a rotation).

If you want technical, it is called taking the quotient under the group of symmetry of the square, which is a dihedral group and has order 8.

[Koyaanisqatsi]

http://www.numeral.com/appletsoftware/eicon.html

[-Jrabbit]

Calling your own theory "interesting" in the thread title is a bit pretentious.

[Kuciwalker]

And a lie.

[VetLegion]

(A nightmare if someone had to sort it all out, but I guess a computer could do the sorting out quite easily if it were to be programmed right).

Not easily. The truth is computers suck monkey balls at image recognition. That's why Google has people identifying them http://images.google.com/imagelabeler/

[Kuciwalker]

Actually, one of my professors invented that. Google bought it from him (or from the university, not sure which).

[FrostyBoy]

Re: Re: Interesting Math and Visual Theory

Originally posted by Kuciwalker


You failed algebra in grade school, didn't you.

Nah, fell asleep.